Foreword: The focus of this essay is strictly altitude placement/change of GHCN stations. While challenge and debate of the topic is encouraged, please don’t let the discussion drift into other side issues. As noted in the conclusion, there remain two significant issues that have not been fully addressed in GHCN. I believe a focus on those issues (particularly UHI) will best serve to advance the science and understanding of what GHCN in its current form is measuring and presenting, post processing. – Anthony
By Steven Mosher, Zeke Hausfather, and Nick Stokes
Recently on WUWT Dr. McKitrick raised several issues with regard to the quality of the GHCN temperature database. However, McKitrick does note that the methods of computing a global anomaly average are sound. That is essentially what Zeke Hausfather and I showed in our last WUWT post. Several independent researchers are able to calculate the Global Anomaly Average with very little differences between them.
GISS, NCDC, CRU, JeffId/RomanM, Tamino, ClearClimateCode, Zeke Hausfather, Chad Herman, Ron Broberg, Residual Analysis, and MoshTemp all generally agree. Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute. Whether one extrapolates as GISS does or not, whether one uses a least squares approach or a spatial averaging approach, whether one selects a 2 degree bin or a 5 degree bin, whether one uses an anomaly period of 1961-90 or 1953-1982, the answer is the same for virtually all practical purposes. Debates about methodology are either a distraction from the global warming issues at hand or they are specialist questions that entertain a few of us. Those specialist discussions may refine the answer or express our confidence in the result more explicitly, but the methods all work and agree to a high degree.
As we noted before, the discussion should therefore turn and remain focused on the data issues. How good is GHCN as a database and how serious are its shortcomings? As with any dataset, those of us who analyze data for a living look for several things. We look for errors, we look for bias, we look at the sampling characteristics, and we look at adjustments. Dr. McKitrick’s recent paper covers several topics relative to the make up and changes in GHCN temperature data. In particular he covers changes over time in the sampling of GHCN stations. He repeats a familiar note: over time the stations representing the temperature data set have changed. There is, as most people know, a fall off in stations reporting shortly after 1990 and then again in 2005. To be sure there are other issues that he raises as well. Those issues, such as UHI, will not be addressed here. Instead, the focus will be on one particular issue: altitude. We confine our discussion to that narrow point in order to remove misunderstandings and refocus the issue where it rightly belongs.
McKitrick writes:
Figure 1-8 shows the mean altitude above sea level in the GHCN record. The steady increase is consistent with a move inland of the network coverage, and also increased sampling in mountainous locations. The sample collapse in 1990 is clearly visible as a drop not only in numbers but also in altitude, implying the remote high-altitude sites tended to be lost in favour of sites in valley and coastal locations. This happened a second time in 2005. Since low-altitude sites tend to be more influenced by agriculture, urbanization and other land surface modification, the failure to maintain consistent altitude of the sample detracts from its statistical continuity.
There are several claims here.
- The increase in altitude is consistent with a move inland and out of valleys
- The increase in altitude is consistent with more sampling in mountainous locations.
- Low level sites tend to be influenced by agriculture, urbanization and other land use modifications
A simple study of the metadata available in the GHCN database shows that the stations that were dropped do not have the characteristics that McKitrick supposes. As Nick Stokes documents, the process of dropping stations is more related to dropping coverage in certain countries rather than a direct effort to drop high altitude stations . McKitrick also get the topography specifics wrong. He supposes that the drop in thermometers shifts the data out of mountainous inland areas into the valleys and low level coastal areas, areas dominated by urbanization and land use changes. That supposition is not entirely accurate as a cursory look at the metadata shows.
There are two significant periods when stations are dropped; Post 1990 and again in 2005. As Stokes show below.
FIGURE 1: Station drop and average altitude of stations.
The decrease in altitude is not caused by a move into valleys, lowland and coastal areas. As the following figures show, the percentage of coastal stations is stable, mountainous stations are still represented and the altitude loss more likely comes from the move out of mountainous valleys .
A simple summary of the total inventory shows this
| ALL STATIONS | Count | Total | Percent |
| Coastal | 2180 | 7280 | 29.95 |
| Lake | 443 | 7280 | 6.09 |
| Inland | 4657 | 7280 | 63.97 |
TABLE 1: Count of Coastal Stations
The greatest drop in stations occurs in the 1990-1995 period and the 2005 period, as shown above McKitrick supposes that the drop in altitude means a heavier weighting for coastal stations. The data do not support this
| Dropped Stations 90-95 | Count | Total | Percent |
| Coastal | 487 | 1609 | 30.27 |
| Lake | 86 | 1609 | 5.34 |
| Inland | 1036 | 1609 | 64.39 |
| Dropped in 2005-06 | |||
| Coastal | 104 | 1109 | 9.38 |
| Lake | 77 | 1109 | 6.94 |
| Inland | 928 | 1109 | 83.68 |
TABLE 2: Count of Coastal Stations dropped
The great march of the thermometers was not a trip to the beach. Neither was the drop in altitude the result of losing a higher percentage of “mountainous” stations.
FIGURE 2: Distribution of Altitude for the entire GHCN Inventory
| Minimum | 1st Qu | Median | Mean | 3rd Qu | Max | NA |
| -224.0 | 38.0 | 192.0 | 419.9 | 533.0 | 4670 | 142 |
TABLE 3: descriptive statistics for Altitude of the entire dataset
We can assess the claim about the march of thermometers down the mountains in two ways. First, by looking at the actual distribution of dropped stations.
FIGURE 3 Distribution of altitude for stations dropped in 1990-95
| Minimum | 1st Qu | Median | Mean | 3rd Qu | Max | NA |
| -21.0 | 40.0 | 183.0 | 441 | 589.2 | 4613.0 | 29 |
TABLE 4: Descriptive statistics for the Altitude of dropped stations
The character of stations dropped in the 2005 time frame are slightly different. That distribution is depicted below
FIGURE 4 Distribution of altitude for stations dropped in 2005-06
| Minimum | 1st Qu | Median | Mean | 3rd Qu | Max | NA |
| –59 | 143.0 | 291.0 | 509.7 | 681.0 | 2763.0 | 0 |
TABLE 5: Descriptive statistics for the Altitude of dropped stations 2005-06
The mean of those dropped is slightly higher than the average station. That hardly supports the contention of thermometers marching out of the mountains. We can put this issue to rest with the following observation from the metadata. GHCN metadata captures the topography surrounding the stations. There are four classifications FL, HI, MT and MV: flat, hilly, mountain and mountain valley. The table below hints at what was unique about the dropout.
| Type | Entire Dataset | Dropped after90-95 | Dropped 2005-06 | Total of two major movements |
| Flat | 2779 | 455 (16%) | 504 (23%) | 959 (43%) |
| Hilly | 3006 | 688 (23%) | 447 (15%) | 1135 (38%) |
| Mountain | 61 | 15 (25%) | 3 (5%) | 18 (30%) |
| Mountain Valley | 1434 | 451(31%) | 155 (11%) | 606 (42%) |
TABLE 6 Station drop out by topography type
There wasn’t shift into valleys as McKitrick supposes, but rather mountain valley sites were dropped. Thermometers left the flatlands and the mountainous valleys. That resulted in a slight decrease in the overall altitude.
That brings us to McKitrick’s third critical claim. McKitrick claims that the dropping of thermometers over weights places more likely to suffer from urbanization and differential land use. “Low level sites tend to be influenced by agriculture, urbanization and other land use modifications.” The primary concern that Dr. McKitrick voices is that the statistical integrity of the data may have been compromised. That claim needs to be turned into a testable hypothesis. What exactly has been compromised? We can think of two possible concerns. The first concern is that by dropping higher altitude mountain valley stations one is dropping stations that are colder. Since temperature decreases with altitude this would seem to be a reasonable concern. However, it is not. Some people make this claim, but McKitrick does not. He doesn’t because he is aware that the anomaly method prevents this kind of bias. When we create a global anomaly we prevent this kind of bias from entering the calculation by scaling the measurements of station by the mean of that station. Thus, a station located at 4000m may be at -5C, but if that station is always at -5C its anomaly will be zero. Likewise, a station at sea level in Death Valley that is constantly 110F will also have an anomaly of zero. Anomaly captures the departure from the mean of that station.
What this means is that as long as high altitude stations warm or cool at the same rate as low altitude stations, removing them or adding them will not bias the result.
To answer the question of whether dropping or adding higher altitude stations impacts the trend we have several analytical approaches. First, we could add back in stations. But we can’t add back in GHCN stations that were discontinued. The alternative is to add stations from other databases. Those studies indicate that adding addition stations does not change the trends:
http://moyhu.blogspot.com/2010/07/using-templs-on-alternative-land.html
http://moyhu.blogspot.com/2010/07/arctic-trends-using-gsod-temperature.html
http://moyhu.blogspot.com/2010/07/revisiting-bolivia.html
http://moyhu.blogspot.com/2010/07/global-landocean-gsod-and-ghcn-data.html
The other approach is to randomly remove more stations from GHCN and measure the effect. If we fear that GHCN has biased the sample by dropping higher altitude stations, we can drop more stations and measure the effect. There are two ways to do this. A Monte Carlo approach and an approach that divides the existing data into subsets:
Nick Stokes has conducted the Monte Carlo experiments. In his approach stations are randomly removed and global averages are recomputed. Stations were removed based on a randomization approach that preferentially removed high altitude stations. This test gives us an estimate of the Standard Error as well.
| Period | Trend of All | Re-Sampled | s.d |
| 1900-2009 | 0.0731 | 0.0723 | 0.00179 |
| 1979-2009 | 0.2512 | 0.2462 | 0.00324 |
| Mean Altitude | 392m | 331m |
Table 7 Monte Carlo test of altitude sensitivity
This particular test consists of selecting all the stations whose series end after 1990. There are 4814 such stations. The sensitivity to altitude reduction was performed by randomly removing higher altitude stations. The results indicate little to no interaction between altitude and temperature trend in the very stations end after the 1990 period.
The other approach, dividing the sample, was approached in two different ways by Zeke Hausfather and Steven Mosher. Hausfather, approached the problem using a paired approach. Grid cells are selected for processing if the have stations both above and below 300m. This eliminates cells that are represented by a single station. Series are then constructed for the stations that lie above 300m and below 300m.
| Period | Elevation > 300m | Elevation <300m |
| 1900-2009 | .04 | .05 |
| 1960-2009 | .23 | .19 |
| 1978-2009 | .34 | .28 |
Table 8. Comparison of trend versus altitude for paired station testing
FIGURE 5: Comparison of temperature Anomaly for above mean and below mean stations
This test indicates that higher elevation stations tend to see higher rates of warming rather than lower rates of warming. Thus, dropping them, does not bias the temperature record upward. The concern lies in the other direction. If anything the evidence points to this: dropping higher altitude stations post 1990 has lead to a small underestimation of the warming trend.
Finally, Mosher, extending the work of Broberg tested the sensitivity of altitude by dividing the existing sample in the following way, by raw altitude and by topography.
- A series containing all stations.
- A series of lower altitude stations Altitude < 200m
- A series of higher altitude stations Altitude >300m
- All Stations in Mountain Valleys
- A series of stations at very high altitude. Altitude >400m
The results of that test are shown below
FIGURE 6 Global anomaly. Smoothing performed for display purpose only with a 21 point binomial filter
The purple series is the highest altitude stations. The red series lower elevation series. Green is the mountain valley stations. A cursory look at the “trend” indicates that the higher elevation stations warm slightly faster than the lower elevation, confirming Hausfather. Dropping higher elevation stations, if it has any effect whatsoever works to lower the average. Stations at lower altitudes tend to warm less rapidly than stations at higher elevations. So quite the opposite of what people assume, the dropping of higher altitude stations is more likely to underestimate the warming rather than over estimate the warming.
Conclusion:
The distribution of altitude does change with time in GHCN v2.mean data. That change does not signal a march of thermometers to places with higher rates of warming. The decrease in altitude is not associated with a move toward or away from coasts. The decrease is not clearly associated with a move away mountainous regions and into valleys, but rather a movement out of mountain valley and flatland regions. Yet, mountain valleys do not warm or cool in any differential manner. Changing altitude does not bias the final trends in any appreciable way.
Regardless of the differential characteristics associated with higher elevation, changes in temperature trends is not clearly or demonstrably one of them. For now, we have no evidence whatsoever that marching thermometers up and down hills makes any contribution to a overestimation of the warming trend.
Dr. McKitrick presented a series of concerns with GHCN. We have eliminated the concern over changes in the distribution of altitude. That merits a correction to his paper. The concerns he raised about latitude, and airports and UHI will be addressed in forthcoming pieces. Given the preliminary work done on airports. (and here) and latitude to date, we can confidently say that the entire debate will come down to two basic issues: UHI and adjustments, the issues over latitude changes and sampling at airports will fold into those discussions. So, here is where the debate stands. The concerns that people have had about methodology have been addressed. As McKitrick notes, the various independent methods get the same answers. The concern about altitude bias has been addressed. As we’ve argued before, the real issue with temperature series is the metadata, its related microsite and UHI issues and adjustments made prior to entry in the GHCN database.
Special thanks to Ron Broberg for editorial support.
References:
A Critical Review of Global Surface Temperature Data Products. Ross McKitrick, Ph.D. July 26, 2010
Good work. I appreciate skeptics who are prepared to test if alleged biases are of practical importance. If only more would make the effort.
Very informative piece, and helpful in focusing the issues.
On behalf of those who respect Dr. McKitrick’s overall work, as I do, I trust that he will offer a correction or clarification, to the extent warranted, once he has had a chance to review.
So the remaining issues are: (i) metadata, (ii) microsite issues, (iii) UHI issues, and (iv) adjustments. Hmmm. Not to wave a crystal ball or anything, but I suspect (iii) and (iv) will be the most meaningful . . .
Nope, Steve you didn’t convince me.
The biggest drop in altitude correlates with the biggest increase in temps. Around 1998.
When you’re talking about reading temps in 1/100ths and 1/10ths of a degree, you left out the biggest part. The time of day that those thermometers are read. Lower altitudes are going to show a greater change/swing in temps every day and be more effected by UHI.
As you say, methods for computing global averages are sound. Of course.
It’s the data quality, and the many correction methods that are abjectly wrong and abused, that cause problems. Read the papers that justify the correction methods, note the many assumptions, note the error that should be associated with these assumptions, note the predominant effect of each method (to lower the values of past temperatures).
The data is junk, and the science is corrupt. It’s very sad.
It is not only altitude that effects temps, but when hundreds of stations closed in Canada and Siberia… Surely having lost 60% of the stations and most of those from cold countries, leaving stations in hotter regions, that must increase the average for the remaining stations.
[REPLY – But one can get around that problem by gridding the data. There are, however, other serious issues concerning Siberian stations other than the mere number that are there. ~ Evan]
Steven…..I know you and Zeke and Nick are all really bright guys. I know E.M. Smith is a really bright guy too but his work shows that the “march of thermometers” has had an impact on the temperature trends: http://chiefio.wordpress.com/2009/11/03/ghcn-the-global-analysis/.
The real question to me is why are all of you bright guys spending so much time on verifying (or not) that the use of bad data, regardless of methodology, produces similar (or not) results? Why isn’t there a concerted effort on everyone’s part to go back to the raw data and start over? What is the point in spending so much time, effort and research dollars on studying questionable data? Then again, perhaps you don’t agree that it is questionable.
As mentioned by latitude, there is an ‘apparent’ correlation between station count and rate of temperature change if you graph them together. Visually this has always made me raise an eyebrow. Has anybody been done a temp study for only the long lived stations that made it through the purge/s? Any links?
Because imo there should not be a significant difference in trend between this and what is graphed using the entire GHCN database.
Thanks, good post!
Dr. John Christy has done specific research on land use change and its effects on LT and surface temperature, one of which McKitrick cites in his paper. It disagrees with the conclusion of this essay.
http://climateaudit.org/2009/07/19/christy-et-al-2009-surface-temperature-variations-in-east-africa/
With due respect, not one single of your entries at WUWT or at Lucia’s have addressed the specific issues raised by Christy et al (his is but one, there are several). Until those issues are addressed and refuted, all that’s being done in these exercises is replication of the data, NOT analysis of the individual stations. It isn’t just microsite and UHI at play here.
Agreement amongst models is nothing to get excited about.
“The concerns that people have had about methodology have been addressed. As McKitrick notes, the various independent methods get the same answers”
I don’t trust it.
What I would like to see is each dropped station’s temp records analyzed.
I don’t think for one minute that any station was dropped “random”.
I think there’s was a specific reason for each station that was dropped.
Hey, Mr. Mosher-
“GISS, NCDC, CRU, JeffId/RomanM, Tamino, ClearClimateCode, Zeke Hausfather, Chad Herman, Ron Broberg, Residual Analysis, and MoshTemp all generally agree. Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute.”
I expanded your method and used your concensus theory on another real world problem:
I took a bitmap of Botticelli’s ‘Printemps’ and used the finest possible AlGorithm to average the pixel colors. Guess what! It’s gray!. Botticelli painted everything in gray, the same color as your elephant – the one that’s very like a poutine.
Has anyone else clicked onto the source website for that lovely picture? If so have you noticed that the pictures look so very different, yet are clearly identical.
Relevance to the topic at hand is: facts are tricky things, never the same when you cast a different light over them.
It seems easier to me to just look at the percentages of the stations for the sitings,
pre-1990 vs. post 2006 which did not change: 38.2% vs 39.9% Flat, 41.3% vs. 41.0% Hilly, .008% vs. .009% Mountain and 19.7% vs. 18.1% Mountain Valleys. One could argue the Valleys are under-represented by 5% post-2006. I am curious about the differentiation of mountains vs. valleys, as the top of Mt. Washington would not qualify as a valley in elevation in Western Colorado, let alone the flatlands in the eastern part of the state at 1700 meters. 99% of the population in the Roaring Fork Valley lives in a Mountain Valley, yet this is designated as high altitude as it is 1900 meters at its lowest point, and Aspen is still a valley town at the high end of this valley although its altitude is 2600 meters.
latitude says:August 19, 2010 at 8:47 am Lower altitudes are going to show a greater change/swing in temps every day and be more effected by UHI.
The day/night temperature swing is probably greater at altitude. I’ve been in Colorado for almost 40 years, and we get 30-50°F day to night change virtually every day. The inhibiting factor is cloud cover and or humidity*, both of which should be more predominant at lower and coastal climes.
*Absolute water content, not RH%
I still find this disconcerting: http://i27.tinypic.com/14b6tqo.jpg
It’s good to see these sorts of issues being dealt with at last. Only question, why has it taken a team of amateurs to do what the professionals should have done in the first place?
BTW, I would be interested in what E.M Smith has to say, but for now, good work.
Ken Hall says:
August 19, 2010 at 8:56 am
Surely having lost 60% of the stations and most of those from cold countries, leaving stations in hotter regions, that must increase the average for the remaining stations.
———-
No. This would happen is the analysis was a mean of the raw temperature data. But it isn’t. It’s the mean of the anomalies. All stations, everywhere, have an average anomaly of zero over the normal period (e.g. 1961-1990).
The contrary problem may exist, if there is Arctic amplification, dropping Arctic sites will drop the sites with the strongest trends, reducing the global trend
Is there a temperature graph that compares the trends up to 1990, 2005 of all the stations that were rejected, compared to all the stations that were kept?
“pace of global warming is not in serious dispute” ….. NO…..the trend shown in measured surface temperatures is not in dispute. The use of this metric as a measure of “global warming” is in serious dispute. Ocean heat content is clearly the better metric.
How about doing this from the other direction? Class the stations by trend, rising temps, falliing temps, maybe general shape of the curve, and then find, or try to find, what attributes they have in common. And of course how they are treated by GHCN.
If Nick Stokes says its so … I know it ain’t.
Table two has an error in calculation for the flat Terrain type. According to the way the other three lines are calculated the percentages should be 16%, 18%, and 35%, at least if my rpn calculator and my finger works correctly.
Type Entire Dataset Dropped after90-95 Dropped 2005-06 Total of two major movements
Flat 2779 455 (16%) 504 (23%) 959 (43%)
Sorry that was Table 6.
” Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute. ”
Considering that all of the GHCN anomaly caluclators use the mean:
1) The min could be going up
2) The max could be going up
3) Or it could be a combination of both
Having looked at the raw GHCN data, I can say the max is not going up. It is the min.
Therefore it is UHI.
As a proxy for whether the max is going up or the min, may i point out that 25 of the 50 state temperature max records were set in the 1930’s.
The 1930’s were the hottest decade (if you care about max temperatures).
It may be ture that the mean has gone up because the min has gone up.
But that is nothing to worry about.
Here in Wyoming the colder air settles into the valleys on the plains (hundreds of feet of elevation change in some of these draws and stream valleys here) summer and winter. Temperature inversions also effect even the high mountains where normally temperature decreases with elevation with the exception of inversions where it can be much warmer at higher elevations. These occur regularly here. As Latitude said, time of day will also effect these readings also amount of sunshine which is high here, averaging 270 days a year. Not sure what the change in elevation of temperature reading sites would do overall other than make the data inconsistent with past readings.
The valley in the photo looks like a classic river cut formation without meander though there is not enough of the countryside to say for sure regarding the hills on each side. Shape of these valleys could also effect temperature greatly.
I think this collaboration is extremely encouraging. Congratulations to the three of you.
Can we identify one data set from one single station that covers many years continuous to the present, that is widely regarded as accurate to a degree beyond reproach and is free of outside bias and adjustment? Such a data set, if it exists, covering just one single location, would give us a better indication of global changes than anything that has yet been published. With good numbers one can do accurate sums. With bad numbers one can do nothing but waste one’s time. There is no possible way to calculate the value of x where the values of its individual components are uncertain.
Two major issues with the assertions in the above posting.
1) Using data that ‘cuts off’ in 1990 does not capture the major issue which is that the way the data are handled post 1990 introduces a ‘hockey blade’ to the anomaly series. This is accompanied by a change of ‘duplicate number’ in the GHCN.
2) It ignores the impact of long duration weather changes, such as 30 and 60 year ocean cycles, as it will work through changes of the average VOLATILITY of the stations in the set. Ignoring volatility changes while looking at other attributes in isolation will fail.
http://chiefio.wordpress.com/2010/08/04/smiths-volatility-surmise/
The fact that a bunch of ways of calculating an average of an intensive variable from the same dataset gives similar results does not improve the quality of the dataset nor does it address that an average of an intensive variable is rather meaningless. The basic mathematics is wrong, so any arithmetic done on it is pointless.
http://chiefio.wordpress.com/2010/07/17/derivative-of-integral-chaos-is-agw/
So we have a bunch of folks doing number games for amusement and claiming to find truth. They aren’t.
So for starters, pick some individual long lived stations and look at their actual temperatures. When you do that you find either little to no “warming” or amounts consistent with UHI and / or for very long lived stations a rise out of the LIA. Sticking 92% of so of present GHCN thermometers at Airports in the USA (and similar percentages in France and ROW) doesn’t help. Tarmac stays hotter than a grass field, regardless of number of flights.
So, IMHO, we have crappy data and get crappy results from it. Admiring the uniformity of the crappiness does not yield much comfort.
I think it is clear that the adjustments done to the raw data is where the devil in the details rests. I have yet to see a reasonable UHI adjustment in the dozens of station in GISS that I have looked into. In all too many cases I have seen older records adjusted down (?) and newer records adjusted down by lesser amounts. Very few locations have experienced de-urbanization … the UHI adjustments should be increasing from old to new not the other way around.
Minor typo, para 4 sentence 2:
‘How good is GHCN as a database and how serious are it’s shortcomings?’
‘it’s’ should be ‘its’. Wouldn’t normally bother but the quality of the writing is good enough to make the error stand out.
I’m still reading, but with the Nepal business fresh in mind I am most interested in the subject of this essay (which I have been awaiting since Mr Mosher dropped hints in the comments over at Lucia’s).
My thanks to all three authors for their ongoing efforts, both for their own sake and for the example they set to others, here and elsewhere.
Dominic
[Thanx, fixed. ~dbs, mod.]
C James,
For better or worse, pre-1970s GHCN v2.mean is by and large the only “raw” data available to use. Post-1970s we have been playing around with using GSOD/ISH and other alternative datasets, though efforts so far indicate that over broad geographic regions they give results similar to GHCN.
You can easy recognize an honest approach when you read it!
Im catching up on details and I really appriciate the “tone” and the willingness invitation to get to the bottom of whats “allmost settled” and not.Its obviuos who will take responsability or not to gain end deserve trust! There are some aspekts and issues in the comments above that I think is intrseting to adress.The “correlation” or Coincidence” that the temprecord year was the same year a big drop in high altitude temps where dropped ?
Eric:
“So the remaining issues are: (i) metadata, (ii) microsite issues, (iii) UHI issues, and (iv) adjustments. Hmmm. Not to wave a crystal ball or anything, but I suspect (iii) and (iv) will be the most meaningful . . .”
I will give you my take.
Adjustments:
1. TOBS. the adjustment is valid, but its needs error bars
2. Instruments: same as above.
3. Station moves: bears investigation
4. Homogeniety: bears investigation
5.UHI: may not be possible.
UHI:
1. Depends ENTIRELY on the definition of Rural, that is metadata.
Microsite:
1. A small effect ( say .1C) is supported by prior studies.
2. The signal will be hard to find, isolated primarily in Tmin
Skeptics should not hope for more than a .15C adjustment
Warmists should Expect a wider band of confidence intervals. They should not fight the proper calculation of uncertainty
“Skeptics should not hope for more than a .15C adjustment”
Steve, that’s premature.
The lapse rate is 6.5C per km.
The average temperature in a grid box, for example, would therefore increase by 0.65C for each decline in the average altitude of just 100 metres. Your numbers are showing changes of that magnitude.
Now to the extent that each individual station is measured/calculated based on its individual anomaly only, this shouldn’t matter. But if it isn’t done strictly by station, anomaly only, it will make a very large difference.
[On the other hand, there is also some evidence that temperatures are increasing faster at the surface than higher up in the troposphere and, therefore, the lapse rate profile is also changing by altitude which will also influence the trend, even in a strict station anomaly only calculation].
Can someone clarify two things for me: (1) How many times a day is
the temperature recorded for the stations (in general), and what has
been used to determine the frequency and choice of time, and (2) How
is it known, or established, or agreed, as to what area of land is represented
by each station. Surely this latter is critical, because the area the station
represents needs to be known so that the proper weight can be applied
when working out the mean. It is the mean temperature per sq kilometre
that is surely the important thing, not just the mean per se.
Or am I way off the track here?
where did Table 6 come from?
Why do these simple to compute figures appear wrong? – and apparently wrong in a way we’ve come to expect!
BillyBob says:
August 19, 2010 at 10:21 am
” Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute. ”
Considering that all of the GHCN anomaly caluclators use the mean:
1) The min could be going up
2) The max could be going up
3) Or it could be a combination of both
Having looked at the raw GHCN data, I can say the max is not going up. It is the min.
Therefore it is UHI.
BillyBob:
You are correct that the min is going up much faster than the max. One region where this is especially strong is in the Swiss Alps. One problem with your conclusion is that green house gases are expected to reduce night-time cooling and to have a greater effect on the min rather than the max temperature. In the arctic and the in the mountains, the green house effect seems a more likely explanation than UHI for stronger increases in the min compared to the max temperature.
“The greatest drop in stations occurs in the 1990-1995 period”
It’s about 1990-1993, which is exactly where the hockey stick takes off.
I don’t think it’s coincidence at all and even though you guys did a “random” test,
I don’t think dropping the stations was random.
I’d still like to see someone look at the ‘trends’ from the stations that were dropped.
I would be willing to bet there was a reason for dropping those stations.
There is just too much coincidence that many stations were dropped,
and immediately after that, we had catastrophic unprecedented global warming.
EM.
“1) Using data that ‘cuts off’ in 1990 does not capture the major issue which is that the way the data are handled post 1990 introduces a ‘hockey blade’ to the anomaly series. This is accompanied by a change of ‘duplicate number’ in the GHCN.”
Well, That assertion makes no sense. The method for handling duplicates is well documented and many of us, SteveMc, RomanM, Myself and other people skeptical of the record keeping and math don’t share your undocumented concern. Further, we have tested handling duplicates in several ways, even performed monte carlo tests on the duplicates. That’s just standard practice. Finally you cannot assert the introduction of a “hockey blade” without knowledge of a shaft during the same period to compare it to. Ah yes, One can ALSO test the hypothesis by Dropping all stations that have ANY duplicates. The average station has .5 duplicates. The first quartile have NO duplicates. Answer doesnt change. Its getting warmer. its NOT getting cooler since the LIA, its getting warmer. Not cooler. I got zero evidence that it is getting cooler since the LIA. I certainly have seen no evidence supprting a contention that it is getting cooler. And it isnt staying exactly the same.
“2) It ignores the impact of long duration weather changes, such as 30 and 60 year ocean cycles, as it will work through changes of the average VOLATILITY of the stations in the set. Ignoring volatility changes while looking at other attributes in isolation will fail.”
Well the first thing one would have to prove is that there are 30 and 60 year ocean cycles of a DEFINITE volitility. That assertion would rest on data. Presumably SST data. You probably need to start looking at SST data. Start with Bob Tisdale he will help you avoid problems.
“So for starters, pick some individual long lived stations and look at their actual temperatures. When you do that you find either little to no “warming” or amounts consistent with UHI and / or for very long lived stations a rise out of the LIA. Sticking 92% of so of present GHCN thermometers at Airports in the USA (and similar percentages in France and ROW) doesn’t help. Tarmac stays hotter than a grass field, regardless of number of flights.”
1. Looking at the distrubution of trends over time this is not true. The distribution of the trends is slightly skewed to more warming stations than cooling stations. Its somewhat leptokurtic.
2. Tarmac. does stay hotter for a while. but that pesky 2nd law has something to say about it. The TESTABLE question is this. Does the Tarmac stay warm enough long enough to impact the Tmin reading.
For example.
tarmac feild
6pm 10 8
7pm 10 7
8pm 9.5 6.5
9pm 9.0 6.25
10pm 8 6
11pm 7 6
Midnight 6 6
In that TOY case, you see that the tarmac stays warmer longer, BUT by the time reading is taken for the day, it doesnt matter. because we DONT integrate temp over the day, we take the max and the min. Anyway, its a TESTABLE hypothesis that tarmac does, in fact, ( and not just in speculation) stay warm enough, LONG ENOUGH to make a substantial difference. so, test it. Until then its an interesting hypothesis. test it.
Second. It will change depending on the day: Windy days with windspeeds over 7m/sec. The excess heat is rapidly moved out. Airports have long clear fetches and nice laminar flow at the surface. for a reason. Landing planes also help create turbulant mixing at the surface if you dont have wind flow over the surface. Third. Windy and cloudy days are different. See the CRN study. 4th. rainy days are different. Tarmac warmer? yup. certain conditions, for certain periods, of variable magnitude.
So, can airports be hotter. Sure. How much? when? where? how long? all testable. I prefer to focus on testable things. And, the final thing is this. The biggest issue is the potential step change when an airport starts. its the trend baby. The farther you get away from that step change, the LESS it matter.
I ask you to step on scale in 1950. You weigh 175 lbs.
I measure you every year until 2010. By 2010 you weigh 225 lbs.
compute the trend.
Then, I study the records and I find that in 1951, they started weighing you with your SHOES ON. crap. Well I cannot reweigh you in 1950 with shoes on. So I have these choices:
1. throw out all the data and say we dont know.
2. Throw out 1950 and compute the trend
3. Leave 1950 in and realize that if shoes only weigh 2 lbs I dont have that much error. In fact, I can simulate this and see how IMPORTANT that error is to my trend estimation.
4. Make up a shoe adjustment. But I have no record of wether you wore flip flops or hiking boots.. Still I can try a distribution of shoe weights.
5. make confusing statements.
Now, If they changed protocal in 1960 I still have all the same choices, just different answers. In any case, we always and forever face these challenges in real world data analysis. They dont prevent us from knowing things or estimating things. They condition our knowledge but ALL knowledge is conditional. At least, as a skeptic, that is what I practice. conditional acceptance.
Anyway, the airport testing is coming. It’s gunna take a while, to disprove a case that hasnt been made, but its doable.
Mosh wrote:
“I ask you to step on scale in 1950. You weigh 175 lbs.
I measure you every year until 2010. By 2010 you weigh 225 lbs.
No, by 2010 he weighs 225 lbs after his weight has been adjusted. We don’t now what he actually weighs in 2010 – that’s the problem.
“to disprove a case that hasn’t been made”….Now there’s the scientific method in action….nothing like a nice bias to help design your study……and besides, you are studying surface temperature measurements which is a crappy metric for “global warming”. And I used to think you were open-minded.
Latitude:
Provide a hypothesis to test WRT lower altitudes. You want a record that contains ONLY low altitude?
You got it. that record show that OVER TIME low altitude and high altitude warm at the same rate. Change over time=the same.
For grins I compared the lowest site (-224m) with the highest site (4000+meters)
Answer? same trend.
So, I cut the data every way I could LOOKING TO SUPPORT the contention that there was a different trend over time. Could not find one bit of emprical evidence to support that notion. with no evidence to support it, I will retain my rational belief that dropping slightly higher stations makes no difference. On balance, I have a claim that is supported by the evidence. On the data given that belief is warranted. The opposite belief, has no such similar warrant. It is an unwarranted belief.
Now, to make the claim that dropping stations BIASES the record, you have to accept the data. If you want to question the data, then you have no knowledge of bias either way. you have “dont know” . That is just basic epistemology 101.
So: if people want to challenge the data, then they get to claim ignorance. They get to consistently say ” we dont know”. We dont know warmer or colder. we dont know LIA or no LIA. we dont know bias or no bias. We dont know. THAT is a consistent skeptical position. Almost no one takes it.
If we want to claim Knowledge of a bias, then you have to warrant that belief with evidence. You have to point to evidence you take as good and you have to articulate how your conclusion is drawn from that data and why other conclusions are unwarranted.
““to disprove a case that hasn’t been made”….Now there’s the scientific method in action….nothing like a nice bias to help design your study……and besides, you are studying surface temperature measurements which is a crappy metric for “global warming”. And I used to think you were open-minded.”
1. The claim in the paper was that the drop in altitude “comprimised the integrity of the data” As I stated, THIS needs to stated as a TESTABLE hypothesis. It wasnt.
I can see two possible hypotheses. First the silly claim that higher altitude are colder and so removing them hits the TREND. Anyways I tested that as well. Second, the claim that dropping Higher stations Increases the trend.
Thats the claim right;
IF you drop higher stations, THEN you will change the trend.
How do you test that claim?
A. You can add other stations in and measure trend before and after. Did that.
B. You can test the stations that are dropped ( using nicks method). Did that.
C. You can test all stations and stratify your sample by altitude, controlling for geographic homgeniety. Zeke did that.
D. You can stratify by altitude and by topography. I did that.
“you are studying surface temperature measurements which is a crappy metric for “global warming”. And I used to think you were open-minded.”
1. I am not studying surface temperatures. I am testing a claim made in a paper. To test that claim I accept the conditions IMPOSED by the claim.
A. That my method of calculation is correct, which Ross stipulates.
B. That the dataset we are talking about is GHCN
C. That dropping higher ( on average less than 100m) stations is
1. A move INTO valleys and low lands: False
2. A move into areas that will have warming trends ( UHI): No evidence to
support this, so conditionaly false. all conclusions are conditionally
true or false.
2. If you ask ME to pick MY metric? OHC and sea level. Surface temps, sure, look at them, but if I got to pick what should have been measured.. my wayback machine would ask for OHC. Absent that, you got what you got. You draw your conclusion and throw wide ass CIs on the thing.
Steve, you’re going to have to talk English, without inflections, to this biologist.
I didn’t question the data. I accepted the data.
I asked you a question about the collection of the data.
You can trust the data, but not trust the people that collect it, especially when they drop this many stations for no apparent reason.
I would like to see someone look at each station that was dropped, each individual station, and see what the trends were for that individual station.
For my money, it is too much of a coincidence that right after those stations were dropped, we had catastrophic unprecedented global warming.
Does not pass the sniff test….
“No, by 2010 he weighs 225 lbs after his weight has been adjusted. We don’t now what he actually weighs in 2010 – that’s the problem.”
Well, again, you have to be specific about “the adjustments” which adjustments made by who to what record for what reason. And. I dont need to know what he actually weighs, I probably cant. Dont forget, a thermometer is JUST A PROXY for temperature. a scale is just a proxy for what weight REALLY IS.
Next. I have his pants from 1950. he had a 32 in waist. His pants now are a 40in waist.
I also have photos of him.
So too with the temperature record. I have UHA and RSS. I compare over that whole period from 1979 to 2010, what did UHA say about his weight? hmm. they have him growing from 200lbs to 220 lbs. I look at sea level, I look at glaciers. Hmm, All evidence ( of VARYING exactness and bias) point to it being warmer.
I have tried to follow Anthony’s admonition: keep to altitude placement and change of GHCN stations. Sorry, can’t. Too much “hail fellow well met” attitude on the part of those who want climate pseudo-scientists to be scientists. They aren’t, and under the current conditions of their pseudo-field never will be.
S. Mosher writes, “GISS, NCDC, CRU, JeffId/RomanM, Tamino, ClearClimateCode, Zeke Hausfather, Chad Herman, Ron Broberg, Residual Analysis, and MoshTemp all generally agree. Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute.”
Give us back the RAW DATA and openly show all analysis and then we can talk. Until then, Garbage In, Garbage Out. The current “climate” data has no scientific use since it has not been gathered scientifically. And no claims that using anomalies wipes out significant problems are valid. In my opinion, politeness is turning into cover for lying.
(By the way, when has there been a serious dispute that the climate has warmed since the Little Ice Age?)
When I understand the issues on Climate Audit (only sometimes) and I read the comments by Stephen Mosher, I am impressed. Not so today by any of the current arguments. (Hope this posts properly; I hit a key and the formatting changed)
Latitude:
“”I don’t think it’s coincidence at all and even though you guys did a “random” test,
I don’t think dropping the stations was random.”
1. The documentary evidence is clear on this. You can check Nicks post and others about the reasons. Guess when GHCN was compiled. Basically you have no evidence for supposing a non randomness.
“I’d still like to see someone look at the ‘trends’ from the stations that were dropped.
I would be willing to bet there was a reason for dropping those stations.”
Nick did that. Look at his study. Further we looked at all the stations which were kept. That is, Drop all the stations that were dropped in 1990. take ONLY the stations that have continuous records from 1990 to the end. Answer. NO DIFFERENCE.
“There is just too much coincidence that many stations were dropped,
and immediately after that, we had catastrophic unprecedented global warming.”
The warming actually starts around 1975, change point analysis. Crap I did that back in 2007 on CA.
You misunderstood my comment as I was referring to your case about airport testing where you seem to think you already know the answer before even testing a hypothesis. Will you admit it if you are wrong? But my main complaint is that you are equating the temperature trend estimates with “the pace of global warming”. I am sorry but not matter how you slice and dice the surface trends it will not provide sufficient rationale to make conclusions about CO2 driven AGW. You equated the two in the opening paragraph using the term “global warming” instead of surface temperature anomaly. The temperature anomalies, even if not contaminated by UHI or biased in any way are influenced regionally by a variety of first-order forcings (i.e land-use, etc.) and thus are not equivalent to the effect of CO2. So my bitch is the claim that surface temperature anomalies equal global warming.
All of that aside, I have to enquire: Why drop stations at all?
You see? As I see things, with fewer stations there arrives that neat ability to extrapolate temperatures to a wider area, even when those extrapolations are grossly inaccurate.
What’s worse? Those dropped stations are seemingly completely ignored where —if one were to evaluate matters on a quality level— those dropped stations could well be used to verify the extrapolations.
You say you don’t want to discuss UHI, but isn’t that really the big elephant in the room which you and others are seemingly going out of your way to ignore?
In your attempt to narrowly focus on only one aspect, you instead create even more doubt in people’s minds regarding your motivations for doing such.
That the min temperatures are increasing (while the max temperatures are not increasing) really undermines the alarmists’ claims that global warming is bad for us. I don’t think even wild animals are going to miss the lower night-time low temperatures, and it’s almost certainly good news for humanity. We can expect longer growing seasons, more places to live comfortably, even less fuel burned to stay warm overnight in cold climates. Of course that last suggested benefit is doubtful if it’s all a UHI effect — stop burning fuel overnight and the lower overnight min temperatures will tend to return.
You do realize until you un-moderate my last post, and post it, it looks like you’re talking to yourself. 😉
Reply: Mosh doesn’t moderate, but he has editing privileges which allows him to see unapproved comments even if he doesn’t realize that. ~ ctm
Steven Mosher says: …5.UHI: may not be possible. UHI: Depends ENTIRELY on the definition of Rural, that is metadata…”
You appear to be dismissing the UHI issue with a wave of your hand, here. But there must be some approximate way to quantify “non-rural” or to at least identify the location parameters that bias urban-measured temperatures.
But I’m not buying any of this. The entire notion of a global temperature as discussed here is nonsense. Any system that ignores atmospheric enthalpy and fails to account for the 1000 times larger oceanic heat sink is an exercise in futility.
Readjustment of column 5 (for policy makers)
The table below hints at what
Type Entire Dataset Dropped after90-95 Dropped 2005-06 Total of two major movements
Flat 2779 455 (16%) 504 (23%) 959 (43%)
Hilly 3006 688 (23%) 447 (15%) 1135 (38%)
Mountain 61 15 (25%) 3 (5%) 18 (30%)
Mountain Valley 1434 451(31%) 155 (11%) 606 (42%)
TABLE 6 Station drop out by topography type
Type Entire Dataset Dropped after90-95 Dropped 2005-06 Total of two major movements
Flat 2779 455 (16%) 504 (22%) 959 (38%)
Hilly 3006 688 (23%) 447 (19%) 1135 (42%)
Mountain 61 15 (25%) 3 (5%) 18 (30%)
Mountain Valley 1434 451(31%) 155 (16%) 606 (47%)
TABLE 6 Station drop out by topography type (readjusted)
“”” Rex from NZ says:
August 19, 2010 at 11:26 am
Can someone clarify two things for me: (1) How many times a day is
the temperature recorded for the stations (in general), and what has
been used to determine the frequency and choice of time, and (2) How
is it known, or established, or agreed, as to what area of land is represented
by each station. Surely this latter is critical, because the area the station
represents needs to be known so that the proper weight can be applied
when working out the mean. It is the mean temperature per sq kilometre
that is surely the important thing, not just the mean per se.
Or am I way off the track here? “””
Well probably they are read just once per day but apparently with a max/min thermometer, which gives you two numbers per day; but evidently at no particular time. That could be ok if the diurnal temperature cycle is a pure sinusoid; then the max/min readings would occur exactly 12 hours apart; and the average of those two numbers would be the correct daily average Temperature.
BUT, if the diurnal cycle is not sinusoidal, then it at least must contain some second harmonic component witha period of 12 hours, and so you would now need four readings per day at a maximum of six hours interval in order to get the correct daily average.
But if there are clouds blowing through; so the temeprature goes up and down in a random fashion, then even four measurements per day will be incorrect, so you can’t recover even the correct daily average.
And who cares how much area you assign to each thermometer. Hansen seems to think that the temperature stays the same out to 1200 km away from the thermometer, so that’s pushing a million square km area for each thermometer. And there aren’t too many of those thermometers placed out in the ocean so they could be even further apart that 1200 km.
But so far nobody has let that stop them from averaging all those thermometers. Well of course they dopn’t actually average the temperatures; just the differneces between the thermometer reading, and some other fixed value that they can’t determine accurately either.
“”The warming actually starts around 1975, change point analysis””
Then you and I do not agree on that either.
The 1975-1990’s saw a slight increase.
After 1990 temps could have just as easily flat lined or gone down.
That would be right in line with the past.
You’re really only seeing less than 1/2 a degree.
“”Drop all the stations that were dropped in 1990. take ONLY the stations that have continuous records from 1990 to the end. Answer. NO DIFFERENCE.””
No difference in trend is exactly what I’m talking about.
Dropping that many stations and not showing a difference in trend, does not pass the sniff test.
“”Nick did that. Look at his study””
I’ll go look again, but I didn’t see where each individual dropped station was looked at.
Fewer stations means you don’t have to record as many innacurate numbers.
Makes a person feel that much better about it. They’ve lied fewer times.
Andrew
REPLY: Doubtful. The people that record the data, especially airports, would be caught out if they lied. – Anthony
Fascinating disection. In the post climategate era I’ve seen a lot of very good disection of what IS VALID with AGW theory by skeptics. This is driving the debate from “an inert gas can’t change temperature” and “it’s all an evil scheme made up by people haters with malice” to a far more specific and thus relevent and honest debate.
Each time something is rigorously verified as valid, the focus of brain power and other resources can be applied to an ever shrinking set of criteria to investigate.
Perhaps we need a “state of the debate” executive summary that shows what we have seen shown to be credible, what is clearly not credible, and what still needs to be flogged.
Off the top of my head, seems we have the following to ascertain if there IS a valid AGW problem and the extent of the problem:
Historic record:
Data quality
Data modification methodology
Averaging methodology
Trend choices (years, High to High vs. mid to mid, etc.)
for both surface T and ppm GHG
Predictive Modeling:
Estimations of GHG increase due to natural and man made causes
Feedback/Forcing weightings, relationships, and missing pieces
Accuracy when run in reverse as compared to historic record and why there are discrepancies
Effects and changes to the world we can expect as result of a strong warming signal if there is one:
all the doomsday hypothesises [sic lol] (though this seems the most work for the least return on investment at the core of the debate)
The real questions:
Is it really getting warmer?
Will this pose a threat to the world?
How soon? to what degree?
What is it caused by?
What could we realistically do to prevent this?
How could we prepare for this?
some require a previous question to be true to even be necessary but you see what I’m saying.
My skepticism started when I saw hugely authoritarian and decisive statments being made by scientists about things which seemed beyond our current level of scientific ability to know with such extreme levels of certainty. It got a big boost when I started seeing the proposed solutions. But my “skepticism” is no emotional-religious-political driven belief one way or another. Which is why WUWT is oen of the few climate sites I truly enjoy reading every day.
A couple of points.
1. It appears you have set up straw men from McKitrick’s quote.
He says: “The sample collapse in 1990 is clearly visible as a drop not only in numbers but also in altitude, implying the remote high-altitude sites tended to be lost in favour of sites in valley and coastal locations. This happened a second time in 2005.”
Your first two points address his suppositions about the increase in stations over the many years before the 1990 and 2005 station reductions, and that doesn’t seem relevant to me. Interestingly, you agree that the average altitude has gone down, as when you say, “The decrease in altitude is not caused by a move into valleys, lowland and coastal areas.”.
2. You appear to be misunderstanding your own data from Tables 1 and 2. The percent of Coastal stations dropped in 2005 is 9.4% vs 34% of the total, while the percent of Inland stations dropped is 83.7% vs 64% of the total. Since Coastal by definition is near sea level altitude, while Inland is at a higher altitude, my read of your numbers is that higher altitude stations in fact got the axe at a higher rate in the 2005 culling.
3. Your argument on item 3 appears to say that items 1 and 2 are irrelevant. The fact that you decided to address the first two items suggests to me that your motive in dealing with them was merely to set up your theoretical superiority over McKitrick’s expertise. And then, when you run your analysis using 300m and 400m altitudes, it raises more flags for me – 300m is not quite even a mountain by international standards, while your Figure 6 does not display the results for 400m. Why didn’t you use a substantial high altitude test, say 1000m?
As a layman, I have to look beyond the “proof” offered and look at how it was put forth. While you could be 100% correct, your style makes me not want to trust you.
Frankly this representation does not show enough information. What is “mountainous”? Hills and mountains? Is a mountain valley average at, below or above 300m? What about flats and hills?
Where did 300m come from anyway? There are states that have their lowest point above 300m. In fact in half the states in the US, the average elevation is above 300m. So in your methodology, half the states in the US are high altitude. That doesn’t seem right. How many urban areas are above and below 300m? By what standard is 300m and above considered high altidude.
Your conclusion is that mountain valley and not mountain moved to flatland . Yet you do not show this. Since you did not state it, I would assume that the 61 mountain readings are the ones represented by an altitude above around 3,000 (thousand) meters. What are the rest of the above 300 but below 3,000 catagorized as? Hill, mountain valley, some of both, a lot of both? Any flatlands above 300m? How many?
Not saying your conclusions are wrong, just that from what you presented, I cannot tell if you are actually doing what you said you are doing.
You split hairs by saying it is not mountainous but mountain valley doing the moving, yet I do not see a plot of around 3000m or greater altitude trend compared to the rest. Barring that an actual plot of the 4 groups.
Frankly if you say you are going to see what “mountainous” regions compare with the others then you need to define what is “mountainous” like saying hills, mountains and mountain valleys above 300m and then plot it. You cant claim to see what flats, valleys, hills and mountains trend just by plotting the altitude if the altitude you pick has a different ratio of groups in them.
You may be able to say “high altitude”, but you cannot say mountain or “mountainous” if you did not bother to seperate them from the calculation.
REPLY: Doubtful. The people that record the data, especially airports, would be caught out if they lied. – Anthony
Not to nitpick you Anthony, I love WUWT but…
Your response is just an assertion. Like just about everything else in climate science.
And I’m still mad that I got snipped a couple weeks back due to inconsistent moderation. 😉
Andrew
REPLY: No, it is more than assertion, it is firsthand knowledge as a former pilot on how airports operate and airport data is used . If they fudge the temperature at the airport, then the density altitude calculation gets screwed up, and planes trying to take off on a hot humid day end off the end of the runway.
Try flying out of Leadville, Colorado on a hot humid day with bunged up density altitude calculation, you won’t make that mistake twice.
That being said, airports are a terrible place to measure climate, because the data comes from the runway areas. Great for aviation safety, the primary mission, not so great for climate, not part of the airport mission at all. – Anthony
MikeD, as long as we’re setting up climate questions to be asked, one that occurred to me recently is: Why is Greenland covered with glaciers? Take a look at this snow and ice map for middle of August 2010
http://www.natice.noaa.gov/pub/ims/ims_gif/ARCHIVE/NHem/2010/ims2010230.gif
Notice how Greenland is now a giant peninsula of white sticking far south into latitudes where nowhere else around the pole is there any glaciers or snow. It can’t be mountains because surely the northern Rockies in Canada and Alaska are more or less as high. To my mind this is a massive mystery and should be explained before we start trying to control the world’s climate.
Ken Harvey says:
August 19, 2010 at 10:49 am
Can we identify one data set from one single station that covers many years continuous to the present, that is widely regarded as accurate to a degree beyond reproach and is free of outside bias and adjustment? …
____________________________________________________________________________
For a large selection of mostly non-urban weather stations in both hemispheres there is this: http://www.john-daly.com/stations/stations.htm
Another useful data set is this: http://pages.science-skeptical.de/MWP/MedievalWarmPeriod.html
Graphs of oldest temp records: http://i47.tinypic.com/2zgt4ly.jpg
We are asked to confine discussion to altitude changes in measuring stations, but then we get this thrust at us:
“Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute.”
Apologies for going off topic, but this statement has to be addressed.
Apart from the fact that global temperature, even according to the dubious statistics under study here, has been pretty flat for 12 years, ocean heat content has (according to its own set of dubious statistics) been falling since 2003. So no matter what the atmospheric temp or SST is up to while the excess heat burps out of the ocean, that means some there is some ‘cooling in the pipeline’.
So quit it with ‘the pace of global warming’ rhetoric please.
Thanks.
I don’t see that the biggest difference between low & high altitude stations is addressed yet:
The diurnal, as in the difference between the high & low temps as a function of altitude.
A greater percentage of UHI night warming is injected into the data set by removing higher altitude stations, for the simple reason that thinner atmosphere cools quicker and further when the sun is down.
Recomputing GISS by restoring the station selection should remove some of the warming attributed to anthropogenic causes.
So, can airports be hotter. Sure. How much? when? where? how long? all testable. I prefer to focus on testable things. And, the final thing is this. The biggest issue is the potential step change when an airport starts. its the trend baby. The farther you get away from that step change, the LESS it matter….
POssibly but then possibly not. It depends on how the airport and its use changes over the years. Will a small municipal airport still record the same temperature if it eventually expands or sees an increase in traffic.
BillD says;
“One problem with your conclusion is that green house gases are expected to reduce night-time cooling and to have a greater effect on the min rather than the max temperature.”
****
*night time minimum and winter temps have been moderating througout the oldest temperatures records i.e CET since 1660 (not just since 1975).
* Medieval castles were built in the way they were as night time temperatures were warmer than in previous or subsequent periods.
* Frosts were a rarity during much of the MWP and during warm periods of the LIA.
In other words night time and winter temperatures have varied throughout the record without any help from co2 and in the past have been as warm or warmer than the modern era.
tonyb
Whenever someone says “move along, nothing to see here” one is reminded of the old New Yorker cartoon in which a street crowd is watching a man fighting off what looks like a giant squid and a passerby says, “It doesn’t take much to draw a crowd in New York”. No doubt the geomorphologists among us will be weighing in on the reductionist approach to topographic forms and their environmental contexts that is applied in this article.
E.M.Smith says:
August 19, 2010 at 10:52 am
…
So we have a bunch of folks doing number games for amusement and claiming to find truth. They aren’t.
…
So, IMHO, we have crappy data and get crappy results from it. Admiring the uniformity of the crappiness does not yield much comfort.
I have to agree bigtime with The Musings From The Chiefio..
The Chiefio has done a lot of meaningful work looking at the Altitude issue… for example: NCDC GHCN Africa By Altitude
http://chiefio.wordpress.com/2009/12/01/ncdc-ghcn-africa-by-altitude/
which includes the following table:
[chiefio@tubularbells Alts]$ more Therm.by.Alt1.Dec.ALT Year -MSL 20 50 100 200 300 400 500 1000 2000 Space DAltPct: 1849 0.0100.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 DAltPct: 1859 29.0 19.4 25.8 25.8 0.0 0.0 0.0 0.0 0.0 0.0 0.0 DAltPct: 1869 51.3 20.5 20.5 7.7 0.0 0.0 0.0 0.0 0.0 0.0 0.0 DAltPct: 1879 36.9 36.9 4.8 1.2 2.4 0.0 0.0 6.0 11.9 0.0 0.0 DAltPct: 1889 24.7 19.9 17.8 4.1 2.1 0.0 1.4 12.7 17.5 0.0 0.0 DAltPct: 1899 24.2 22.0 15.2 4.4 0.9 1.4 3.0 10.9 17.8 0.2 0.0 DAltPct: 1909 21.4 18.3 14.0 6.2 1.3 2.6 2.9 10.8 21.5 0.9 0.0 DAltPct: 1919 21.0 15.7 12.7 7.2 2.3 6.2 4.8 9.5 20.0 0.6 0.0 DAltPct: 1929 16.8 15.6 11.6 6.8 2.9 6.8 5.0 10.6 23.0 0.8 0.0 DAltPct: 1939 15.5 13.5 10.4 7.1 2.7 6.9 4.5 11.0 27.8 0.6 0.0 DAltPct: 1949 14.8 11.0 8.5 6.5 4.9 9.2 5.2 11.9 27.3 0.7 0.0 DAltPct: 1959 16.5 10.3 8.7 5.0 7.8 13.3 9.0 14.4 14.6 0.7 0.0 DAltPct: 1969 17.3 9.2 8.4 5.9 7.8 12.5 8.8 12.8 16.6 0.8 0.0 DAltPct: 1979 17.9 9.9 8.7 6.4 8.5 12.0 8.6 11.3 15.7 1.0 0.0 DAltPct: 1989 15.8 9.5 8.8 6.8 9.2 12.5 9.0 11.8 15.7 1.0 0.0 DAltPct: 1999 14.9 8.9 9.1 7.8 11.3 14.1 9.6 10.2 13.2 1.0 0.0 DAltPct: 2009 13.7 8.0 9.1 8.2 12.3 13.9 9.6 11.1 13.1 1.1 0.0If your paper was built upon this level of detail I might begin to take it seriously…
If your paper addressed any of the Latitude issues identified in the Africa By Altitude posting I might begin to take it seriously.
When I look at your Figure 1 my BS detectors are activated big time.
Reading this paper also activates my BS detectors big time.
The annual juggling of Stations really stinks… and Altitude is part of that stinking mess… as is Latitude, Airports, Coasts, UHI etc etc etc.
Steven Mosher says:
August 19, 2010 at 12:42 pm
Latitude:
Provide a hypothesis to test WRT lower altitudes. You want a record that contains ONLY low altitude?
You got it. that record show that OVER TIME low altitude and high altitude warm at the same rate. Change over time=the same.
For grins I compared the lowest site (-224m) with the highest site (4000+meters)
Answer? same trend.”
Unless I’m missing some important context, this equates to only one station in the world needed to determine temperature trend over time. If single low alt locations trend the same as single high alt locations, they would all certainly trend the same, all low alts and all high alts. Just pick one for the “warm” at the same rate” data.
Your comparison with the lowest and highest site (no proximity or other variables apparently considered) do not lend me confidence in your methods or conclusions.
Finding a different trend between one low and high alt station would be disproof.
Interesting discussion. These 2 statements seem to be contradictory: “Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute.” And the paper’s conclusion: “The concern about altitude bias has been addressed. As we’ve argued before, the real issue with temperature series is the metadata, its related microsite and UHI issues and adjustments made prior to entry in the GHCN database.” The pace of global warming is in serious dispute if you haven’t resolved if the GHCN data is accurate. The adjustments issue and UHI issues have been debated here at WUWT many times. The microsite issue is, of course, something that Anthony has worked very hard on, with his Surface Stations project. His comment that you are premature to say that .15C adjustment is the most skeptics can hope for is probably accurate. You say previous studies support this. Please cite them. My guess is that at the end of the day the adjustment will be significantly higher, perhaps 3-5 times or more higher than you claim.
Why do I say this? I personally surveyed 3 sites in the USHCN in SW Colorado for the Surface Stations project-Telluride, Hermit and Del Norte. The Telluride and Hermit sites are both high altitude sites which had significant environmental structure issues which would bias the temperature record upwards. The Telluride site had a metal sided shed 5 feet to the west of the MMTS, in addition to other site structure issues. As soon as the personnel at the regional NWS office responsible for this site learned we had been out to survey the site, they ran down and pulled out the instruments at the site, including the MMTS. But, as far as I am aware, the data collected from this site was never pulled out of the data base. (There is quite a bit more to the back story here.) The Hermit site was on private property, and the MMTS was placed 11.5 feet south of a greenhouse directly attached to a large solar design house. I mention the direction, because the prevailing winds in the winter (and winter is a long season when you are at +9000 feet) tend to blow down the valley, from north to south. The MMTS was placed in the heat shadow of the solar house and greenhouse. This site also had a large sheet metal roofed dog house 4 feet away to the SE, and a 3 foot boulder at the base of the MMTS. At both these sites I could clearly feel warmer temperatures near the MMTS than in the surrounding environment.
Of course, this is anecdotal, several sites prove nothing. What of the other sites surveyed for the project? Less than 15% of the sites meet NOAA/NCDC’s own standards for placing these monitoring stations, based upon the surveys done by the Surface Station project volunteers. What does this mean? Stations with CRN1 or CRN2 ratings are considered acceptable, with errors of 1C. And more than 85% of the stations surveyed fall into the latter categories. Would the microsite biases tend to cancel each other out-some reading colder cancelling out the warmer biases? While there are certainly a few stations that had colder biases due to vegetation or water nearby, the vast majority appear to be biased in a warmer direction. Prior to the publication of Anthony’s paper on the project, you can review the site data collected by the project yourself by clicking on the link on the right sidebar to the Surface Stations project. You can even see the individual site surveys, including the three I surveyed in SW Colorado. It is my understanding that the USHCN is part of the GHCN. I’ve heard it claimed that while the US data is just a part of the data in the GHCN, it is the best, most accurate record. If that is the case, then I would respectfully submit that disputing the accuracy of the GHCN data is a quite reasonable position, especially when the alleged temperature increase over the last century is less than 1.5C.
Funny, I have come to believe that GHCN has always been biased towards CAGW since they all started getting a more correct monthly salary.
This sentence somehow dropped out of the 3rd paragraph when I posted it:
CNN3, CRN4 and CRN5 rated stations would have expected errors of greater than >1C.
Apologies!
TABLE 6 Station drop out by topography type
Table 6 states that the greatest cumulative percentage dropout was for stations in Flat areas.
However, from the information supplied by the author, the greatest percentage dropout was for stations in Mountain Valley areas (42%).
This is followed closely by stations in Hilly areas (42%).
Stations in Flat areas coming in a poor third (35%).
Stations in Mountain areas, of which there was only 61 to begin with, were culled by a massive 30%.
The problem with this table was first referred to by Randy says:
August 19, 2010 at 10:18 am
and I have pointed it out twice already.
The author states: “The table below hints at what was unique about the dropout.”
Does it require a qualified statistician to pronounce what’s unique about this table? or are we supposed to “move along”.
It’s GMT where I am, and time for bed.
If I am wrong I apologise in advance, but i would appreciate clarification.
BillD: ” In the arctic and the in the mountains, the green house effect seems a more likely explanation than UHI for stronger increases in the min compared to the max temperature.”
I’ll give you an example. The last time I looked the GHCN station list for Canada dropped from 500 stations in 1975 reporting to less than 40 in the 2000’s. And sometimes less than 20.
Do I think GHCN has a lot of mountain and arctic stations still active? No.
“Give us back the RAW DATA and openly show all analysis and then we can talk. Until then, Garbage In, Garbage Out. The current “climate” data has no scientific use since it has not been gathered scientifically. And no claims that using anomalies wipes out significant problems are valid. In my opinion, politeness is turning into cover for lying.”
Anomalies dont remove problems. They prevent problems. Lucia has a wonderful explanation of this over on her sight. In the past people have misunderstood the process of creating an anomaly and what it actually does. Once you understand it, you slap your forehead and say.. Doh! But still I run into people who say silly things about base periods ( not an issue) and anomalies ( not an issue) In my mind as a critic of climate sciene, I do well to keep my criticism and barbs focused on the REAL issue.
Now WRT raw data. First there is no such thing as raw data. All data is processed. Second, I too would like to get my hands on the records made of the actual instrument readings. Until such time, we have what we have. So every claim has to be conditioned by that understanding. On the best evidence we have it is warming. How much? wow. I only know one way to START to answer that question and that is with analysis tools. Start by building the tools and work your way through the whole problem. Of course you can choose to throw you hands up and say we will never know and you can say we will never know until we get the raw data. Was there an LIA?
“(By the way, when has there been a serious dispute that the climate has warmed since the Little Ice Age?)”
I would not that on many occassion I get people arguing 2 things:
1. We cant say that its warmer today.
and later they say things like..
2. “Well we are coming out of a LIA.” a period for which there is very little measurement, Yet, Yet, no skeptic questions it. I find that ironic.
“When I understand the issues on Climate Audit (only sometimes) and I read the comments by Stephen Mosher, I am impressed. Not so today by any of the current arguments. (Hope this posts properly; I hit a key and the formatting changed)”
That’s my evil twin. I just go where the argument takes me. People( me TOO) said that the methods of calculating an average were flawed. I looked at that. they (I) were wrong. people (ME TOO! see CA threads) said that losing high altitude stations was an issue. Err they (I) were wrong.
Do I think there are issues with adjustements? yup. UHI yup? lack of proper metadata? yup. lack of access to the “least processed” data? yup. can I CONCLUDE anything from this? nope. I cannot conclude that it is garbage or not garbage. That would REQUIRE access to the whole process. since I try not to claim without evidence I hesitate to call it pure or garbage or just kinda dirty. But if I HAD access to the “raw” data and I compared raw to processed, THEN I could say whether it was garbage. Until then, I have my doubts both ways.
In furtherance of that what KINDS of issues would you look for between “raw” and “processed”
Glenn:
“For grins I compared the lowest site (-224m) with the highest site (4000+meters)
Answer? same trend.”
Unless I’m missing some important context, this equates to only one station in the world needed to determine temperature trend over time.”
NOTICE.. I SAID for GRINS. In The final analysis we of course looked at all the data. did you not read the paper. For one cut at the data I looked at the highest 100 stations and the lowest 100. highest 200 and lowest 200. Highest 50, lowest 50. Highest 1 lowest 1. Highest 1000, lowest 1000. Those above the mean, below the mean. Above the median, below the median.
Simply. The average station has gone up about 1C in the past century. Whether it was HIGH or LOW. the RATE of increase across long spans of time is similar. physics should tell you that.
I applaud this work but have to agree with many of the less appreciative comments. Taken as a whole I am not surprised that the data set stands up to scruitiny of this kind. Looking at the length and breadth of the data, by station, by country, by region tells me that the homogenisation is such that very large changes at local level are averaged out at the global level, such is the influence of adjusting stations in certain places.
For example the change of station characterisation method by GISS in February made very little difference overall: http://diggingintheclay.wordpress.com/2010/07/10/gistemp-plus-ca-change-plus-cest-la-meme-chose/ but at a regional level there are some huge changes. Is it a coincidence or by design that these cancel each other out at the lattitude and global levels? As a scientist I have to say I think it is the former. However all my instincts tell me that there is something being overlooked when this is worked up to global level and that there could be unintentional but nonetheless inherent bias in the data. Commonly you don’t see such biases, when you keep looking at something from the same angle. It is great to be able to replicate processing like this, but we may never see the biases if we keep on treating the data in the same manner.
dan Murphy:
1. If you research the history you will see that back in 2007/08 I pointed Anthony to the study by dr. LeRoy where the the CRN rating was documented.
2. There was a debate about WHAT the meaning of the number was.
3. LeRoy has a paper on the topic in French, but I have exchanged messages with his associate who did the field work.
Does CRN5 mean a bias of 5 degrees. Average bias?
There were 9back in 2007/08)two schools of thought. The school that said the PEAK bias was +-5C and the school that said the AVERAGE was 5. The school that said that microsite bias was HIGH frequency ( happens on certain days, under certain conditions) and those who thought that air conditioners ran continuously at every location.
Dr. Leroy’s associate who conducted the xperiements came onto CA and confirmed the position that the CRN ‘number’ was a PEAK bias. and that it was both positive and negative, but Overall the bias was WARM. On Lucia’s site he returned to discuss his experience with us and reconfirmed what he had said. His data showed a positive warm bias for all CRN2-4 ( 5 was NOT TESTED) in the range of .1-.15C with PEAKS going from 1-4C. The rating number is an expression of the RANGE of the bias, not the mean.
There was one and only one ( that I know of) field test. As such, it gives us an indication that the bias MAYBE small. As such that tell me that menne probably didnt have enough stations. As such that tells me to get more station data ( Anthony has done that) So, No conclusions about that issue only this observation:
The effect is likely to be small. That is based on the controlled field experiment. That experiment says to me: go look at more stations, BUT DONT HOPE. Dont HOPE to find a 2C difference. you wont. If the bias was 2C you could find that EASILY with a few stations. ( power of statitical tests and variance, see your stats 201 )
If you want to read his comments then he made them at CA and at Lucias. there is also a paper in french.
Murphy.
here:
ChristianP (Comment#38730) March 20th, 2010 at 1:37 pm
Sorry for my bad english, I am french and I do not speak english.
I know very well the classification of Michel Leroy in our national weather service (Météo France). I have classified some stations.
You can read the original classification here : http://www.ccrom.org/documents…..nement.pdf
The number of °C of error in each class, is one possibility in some stations for the class, but it is not right for all stations in the class.
I have an internal study (Michel Leroy) that gives the 95% uncertainty in each class CRN.
Available with our modern Stevenson radiation shield ( in ABS, tall et small Stevenson radiation shield) and Socrima radiation shield (In the comparisons, the Socrima gives the same values that the radiation shield Davis 7714)
CRN 1 – 2 (Temperatures for the days, in °C)
T max [-0.55°, 0.85°]
T min [-0.34 , 0.74°] (-0.2° for the small Stevenson screen)
T avg [-0.35°, 0.55°]
T inst (one minute) [-0.55°, 0.85°]
CRN 3
T max [-1.2°, 1.5°]
T min [-1.1° , 1.5°] (-0.2° for the small Stevenson screen)
T avg [-1.15°, 1.35°]
T inst (one minute) [-1.2°, 1.5°]
CRN 4
T max [-2.3°, 2.6°]
T min [-2.2°, 2.6°] (-0.2° for the small Stevenson screen)
T avg [-2.3°, 2.5°]
T inst (one minute) [-2.3°, 2.6°]
There are not study for CRN5 because the CRN5 is a station that we must closed, moved, or not use for the climatology.
There is an another classification in the OMM for the equipement :
http://www.ccrom.org/documents…..ssPerf.pdf
Thank you Lucie for your good blog and mathematical explanations !
Murphy:
When I asked the reasearcher to clarify his error bounds, were as I expressed them, he wrote back
ChristianP (Comment#38774) March 21st, 2010 at 4:53 am
Comment 38744 SOD,
“i disagree with this part. a CRN5 station should not be used for WEATHER. it is fine for climate, as long as it doesn t change.”
Yes, but in a CRN5 there were often significant changes and at departure when the station has been installed, it was not CRN5.
A CRN5 or CRN4 are not always very bad in the time, a CRN3 are often good in average annual.
For example in my site, I compared CRN 2 with CRN 4 (station too near the trees of the forest, the tree cut the sea breeze bias the T max (too hot in summer and in a part of spring (begin april), too cold with the shadows, in winter and the end autumn. The T min stay always good in this rural site), with CRN 5 on the roof, and CRN 3.
The CRN 5 on the roof (small roof 100 m2 in tiles terracotta, a radiation shield at 1.5 m of the roof, and at 3.4 m of the soil where there are the CRN2 at 40 meters) in the absolute here, provides “better” temperature that the CRN2 with Stevenson radiation sheild (I compared also with ventilated radiation sheild Young at 6 m/s in CRN2, at this tday, this radiation shield is considered as the best relative reference temperature by Météo France.)
In a average annual, the bias are lows, in the CRN5 on the roof, it is no significatif for this cas with a small roof “rural”.
In the CRN4, the bias is +0.33°C / CRN2. (the summer, for one month, in the average T max, the bias is sometime +2° ! On one day, the T max biais has been +4° in summer, but also in the terrible summer 2003, there are not bias for the days when the wind blow at the opposite of the trees)
In summer 2003, I have 5 days with T max > 40° in the CRN 4, for 1 day with T > 40° in CRN2 and CRN5.
Example of my comparisons here :
http://meteo.besse83.free.fr/i…..072007.jpg
(classe 3 = CRN3, Classe 4 = CRN4, Socrima MF is a small Stevenson of an new/old official climatological station in CRN2, This station is officially closed in 2009, but I keep it open by passion. 1/3 of these stations in France, 1000 stations will closed, because the state have not enought money…)
For me, we must study each site for corrected an annual bias, because each cas is unique.
For example, when I install an another CRN 4 to the opposite in my site and to the opposite of actual CRN 4, the annual bias is no significant, because the wind blow very often to the opposite of the trees that cut the ventilation of this site CRN 4 and there are not shadows with T max in winter.”
So, The bias can be big. but not every day. which means that the average on a monthly, yearly basis will be LES THAN the MAX observed. And this is not me speaking, but the guys who works for the inventor of the ranking.
Steven Mosher:
“First there is no such thing as raw data. All data is processed. Second, I too would like to get my hands on the records made of the actual instrument readings. Until such time, we have what we have.”
Until I can understand what what the “process” of the data entails and why it’s “processed”, I can never have an understanding of what is going on here. Nobody can. You certainly can’t, Mr. Mosher.
Andrew
Murphy: here from the same guy in France. 2007, last time we had this debate and settled it. ( the debate was what does the ‘5’ mean in the CRN5.
Christian
Posted Sep 17, 2007 at 4:45 PM | Permalink
Hello,
Sorry for my very bad English, I am a french observer for Meteo France (National weather in France).
5°c, it is a possible instantaneous error in one day, it is not a certainty and it is not systematic.
Classification does not indicate all. A station perhaps class 5 officially because a too important height of the shelter (> 2 m in released site. Example 2.5 m is classe 5) while being as good as a class 2 or 1.
Here, i have one station in class 2 and to 35 m another station classifies 4 of them natural. Trees and shrubs of the forest which couples the circulation of natural wind.
The maximum error in maximum temperature was of +4°, once . On average maximum temperatures, the error is of 1.5° with 2° for the summer months. The winter the station classifies 4 of them is colder in Tx with the shadows by the trees. I raised a weak average error of 0.1° in minimal temperatures.
See the class 2 and 4 here
The class 2 is an official radiation shield of Meteo France (on the picture : “Socrima MF”)
The radiation shield classe 2
Here you can see the the details of French classification of Michel Leroy.
Or with Excel here
Examples of errors of the maximum temperatures in August 2007 (°C) :
class 2 class 4
31,1 32,8
30,6 32,8
33,7 35,6
32,1 34,4
32,1 35,0
32,8 34,4
30,7 32,8
25,8 27,2
28,1 30,0
28,0 30,0
31,5 33,3
32,1 34,4
29,5 31,1
31,1 33,3
30,5 32,8
33,0 33,9
28,6 30,6
29,9 31,1
32,1 33,3
24,6 26,7
22,2 23,3
23,2 23,9
26,7 28,3
28,6 28,3
35,2 37,2
39,9 41,7
36,6 38,9
35,0 36,7
33,5 32,8
30,0 31,7
28,5 30,0
Why, if global warming is held to be so important an issue, are we collecting much less data than before?
Cost-cutting, perhaps? On “the greatest moral challenge of our time”? Surely not.
Or have hubristic climate scientists decided that their computer models are so accurate at interpolation that it is no longer necessary to make observations with the same rigor as before?
Good work, Steven. I appreciate all your hard work. I also marvel at E.M. Smith’s analysis. IMHO, when people tune in to the weather forecast, they are primarily interested in the projected high. Most people don’t care if the low tonight in Fair Oaks is 52 0r 54. But calculating the average, there is a huge difference. I believe that using an average of the high/low, to calculate temperature trends, gives misleading impression to the public, i.e. 94 as the high for mon. and tues., but low of 52 for mon. and 54 for tues. gives a 1 degree of “warming”. Really? Not to most people. fm
Malaga.
There is nothing there of substance about trends.
When this issue was first raised I went to Chief IO’s site and suggested a test.
he agreed that was a good test and one he wanted to do himself but was unable to.
CCC performed that test. No problem
http://clearclimatecode.org/the-1990s-station-dropout-does-not-have-a-warming-effect/
My idea:
http://chiefio.wordpress.com/2010/01/27/temperatures-now-compared-to-maintained-ghcn/#comment-2969
steven mosher
Hi Guy,
We havent talked much since dinner in SF.
Anyways here is how I would show the problem.
As you note there has been a falloff of stations from 1990
on.
Your contention: this makes a difference.
Gavin et al: makes no difference, the stations we have
post 1990 are good, and good enough.
Ok. Easy.
You have a list of stations used today:
You have GISSTEMP running.
Modify GISSTEMP: throw out the “excess” stations that were used before 1990.
Run GISSTEMP with only those stations used after 1990.
Run GISSTEMP with all the stations.
Difference the two.
HIS RESPONSE:
http://chiefio.wordpress.com/2010/01/27/temperatures-now-compared-to-maintained-ghcn/#comment-2970
E.M.Smith
@Steven:
Great idea! I had it early on 😉
The problem is that GIStemp is ‘brittle’ to station removal. It just crashes in a couple of steps. So you can’t just chop and run.
I got around this with the benchmark I did run via the use of “missing data flags” in a small part of the record and clearly defined by a single statement (date > May 2007 ). That approach is a bit too cumbersome for doing selected records at selected times (and perhaps with different choices based on modification flag…)
What I’m presently contemplating is trying it again, but with the “dropout” happening after STEP1. I *think* it is STEP0 and STEP1 that are the most sensitive to station dropout (as they ‘make assumptions’ about files matching each other line for line) and that I could simply step in after the “merge data sets toss out before 1880″ and “homogenize” steps… but then we still have homogenizing happening on the whole set with ‘in fill’ coming from ‘dropped’ stations….
The alternative is either:
Find all places that an exact record for record file match is expected and teach the code new tricks.
Find all the places that an exact match is expected and assure symmetrical ‘drops’ are done.
Write some code to replace about 75% of all data items with missing data flags and hope that does not cause a crash.
Invent a new benchmark method…
So, yeah, I’ve thought about it. Theoretically, it’s easy… 😉
################
bottomline:
LOW altitude stations WARM AT THE SAME RATE as HIGH altitude. its simple physics. Think about it. If the worls warms 10c over time, do you think that WARMING can be confined? the warm low sites get warmer and the high cool sites get warmer. and they warm at the SAME RATE over TIME.
Latitude. Over short periods of time higher latitude warm MORE. physics also.
Higher latitude are cool. but cool is not the question RATE OF WARMING. that is the Q.
Steven Mosher,
“5°c, it is a possible instantaneous error in one day, it is not a certainty and it is not systematic.”
Even with a 5deg C three sigma, how in the world are reports produced in standard divisions in 0.1/0.25/0.5 beside the obvious, natural/unnatural(measuement error) variation “swamps” EVERYTHING including altitude, UHI, ocean affects, planetary affects, etc., I feel that without a gauge r&r of the temperature network as well as a correlation with/to individual measurement gauges on the network, your comprehensive and very clever group of analysts work is “moot”.
In my industry, all the product would have been recalled, and our customer would be looking for another supplier.
Thanx guys. Love seeing the change in Gestalt that occured last November continue on.
Steven,
Thank you for replies. I’ve just returned home from the office, please give me time to take care of animals and chores before I can take the proper amount of time to read and digest in your replies. Respectfully, Dan
What this means is that as long as high altitude stations warm or cool at the same rate as low altitude stations, removing them or adding them will not bias the result.
It will not bias the anomaly. But it will bias the actual temperature.
I think this post will confuse many people.
He (i.e.McKitrick) doesn’t because he is aware that the anomaly method prevents this kind of bias.
Would you please prove this is what McKitrick was thinking.
See my post on the First Difference Method of computing temperature indices on CA at http://climateaudit.org/2010/08/19/the-first-difference-method/ .
The trouble with anomalies is setting an artificial normal of zero.
When you do that variations of .1 or .5 or .6 degrees looks huge.
All you have to do is look at Central England Temperature over the last 300 years and realize variations of 3.5C – 30 to 40% – from the normal of 9C is not unusual over time frames shorter than what some people claim for the whole of the last century.
And thats with thermometers much less contaminated by UHI than the airport temperatures we now measure to the exclusion of rural areas.
http://en.wikipedia.org/wiki/File:CET_Full_Temperature_Yearly.png
Whatever warming their may be (and I doubt the max NOW is higher than the 1930’s) is purely part of the natural cycle of the end of the last ice age.
And it will all end when the next ice starts or when some nutbar “scientists” are allowed to shoot SO2 into the stratosphere.
GHCN is unreliable and contaminated by UHI.
The authors claim that since high altitude stations have higher trends, the loss of those stations only tend to artificially lower rather than raise the overall trend. McKitrick explains why the loss of stations-at-altitude might matter he writes: “Since low-altitude sites tend to be more influenced by agriculture, urbanization and other land surface modification.” The authors cite and reprint that remark.
Unfortunately, this essay fails to demonstrate a lack of bias due to altitude changes. It purports to respond to McKitrick’s claim but does not.
Here’s why:
McKitrick is quoted as claiming that lower altitude stations have trends in greater excess of their GHG warming–due to land use changes. We can state that idea as the following decomposition:
The global average trend is going to be something of the form:
If we substitute the variables, we see:
Clearly, as weight1 increases and weight2 decreases, the LandUseTrend is assigned a bigger weight and therefore may bias the GlobalTrend upward if the LandUseTrend is larger than the difference between high and low altitude trends.
Ergo, the conclusion of the authors: “Dr. McKitrick presented a series of concerns with GHCN. We have eliminated the concern over changes in the distribution of altitude.” is overstated.
They have only demonstrated that its not enough for the LandUseTrend to be positive in order to bias the GlobalTrend up, its necessary for the LandUseTrend to be larger in magnitude than the difference in trends between low and high altitude stations.
Partagraph below Figure 5:
This test indicates that higher elevation stations tend to see higher rates of warming rather than lower rates of warming. Thus, dropping them, does not bias the temperature record upward. The concern lies in the other direction. If anything the evidence points to this: dropping higher altitude stations post 1990 has lead to a small underestimation of the warming trend.
“Thus, dropping them, does not bias the temperature record upward.”
The wording of this may be wrong. It may be better worded:
“Thus, dropping them, does not bias the temperature anomaly record upward.”
Taking cooler location stations out of the record will potentially bias the actual temperature upward while not statistically significantly affecting anomaly. Global warming is about 1/10ths of a degree. Removing cooler stations can make 1/10ths of a degrees difference.
Steve Mosher said:
John Christy said:
Have you done a detailed analysis to test John Christy’s findings?
Please … what are: GRINS … ?
Those studies indicate that adding addition stations does not change the trends
I don’t think there is a lot of arguing over anomaly (except in GISS) from either side of this issue.
But anomaly is not an important issue in the media. We only hear about hottest year ever this, hottest decade ever that, in temperatures, not in anomalies. Dropped stations in cooler areas can create 1/100ths to 1/10ths degrees of warming. I don’t hear any reports in the media saying anomaly this and that. Slight differences in trend that are not statistically significant in the trend/anomaly and are only measured in 1/100ths or 1/10ths of a degree in actual temperature. And again, actual temperature is what global warming is all about, and what warmest this and that reported in the media is all about.
E.M.Smith says:
August 19, 2010 at 10:52 am
So, IMHO, we have crappy data and get crappy results from it. Admiring the uniformity of the crappiness does not yield much comfort.
NOM!
As much as I agree with the criticisms of this article, I have to agree with the conclusions with one exception.
Some time ago I took the GISS gridded data and looked at the trend with the grid points that lost coverage after 1990 due to station drop out removed from the entire data series. To my surprise (being a hardcore skeptic) the result was less warming, not more. I confirmed this by trending the dropped drid points prior to being dropped, and sure enough, they exhibited a slightly higher warming trend (at least until they dropped out) than the grid points that remained. I cannot conlude from that that there was anything nefarious about which stations were dropped and when.
I do want to comment however on the use of anomalies. As Mr Mosher contends, the use of anomalies solves a great number of problems in comparing trends and data from a wide variety of sites at different altitudes, latitudes, and so on. But the use of anomalies does create a new problem and that is one of perspective. We are debating global temperature changes in terms of tenths of a degree per decade, sometimes hundreths of a degree are all that define a “record high year” compared to the previous record.
On a scale of -1 to +1 degrees, 0.6 degrees per century of warming looms large. To put it in perspective, take the temperature of a city like Moscow, or Winnipeg, or Edmonton and plot the annual temperatures and then plot them again with an extra 0.6 degrees added to every data point. Those cities have annual temperature ranges of 70 degrees or more, and when you plot the data in that fashion, the significance of the warming trend becomes very muted. To make the point further, plot the annual temperature swings of those cities on a Kelvin scale with a range from zero to 320. While the annual variation is still visible, a few tenths of a degree of warming over a period of a century just disappears from view. In fact, from that perspective, the marvel of our planet is just how incredibly stable the over all global temperature actually is.
As I skeptic I firmly believe that the earth has been warming since the LIA, but well within natural variability, and almost all beneficially.
Steven Mosher,
Your models are writing checks reality can’t cash…
p.s. didn’t see the error bars in any of your plots.
George E. Smith says:
August 19, 2010 at 1:40 pm
“”” Rex from NZ says:
August 19, 2010 at 11:26 am
Can someone clarify two things for me: (1) How many times a day is
the temperature recorded for the stations (in general), [–snip–]
Well probably they are read just once per day but apparently with a max/min thermometer, which gives you two numbers per day; but evidently at no particular time. [–snip rest–]
George,
Well, I dunno.
It really would depend upon the time of day, wouldn’t it?
What if the temp. were to be taken at the most likely times of min and max?
That would be more honest, would it not?
But WHEN it the key here.
WHEN?
Does ANYONE really know when ALL those temp sensors are actually read?
Are there ANY time tables for each and every device?
For all we know they might be read whenever …
And you know what that will produce, don’t you?
intrepid
“Even with a 5deg C three sigma, how in the world are reports produced in standard divisions in 0.1/0.25/0.5 ”
Quite easily. You do understand the LLN. And if the distribution of spikes is leptokurtic its even better.
http://en.wikipedia.org/wiki/Kurtosis.
But lets do a simulation.
Suppose we do a little field test and determine from that test
that the bias is ZERO. but in our test we have a 1 sigma value
of 3 thats a 3 sigma of 9. what can we expect if we sample that
100 times, 1000, etc
with a standard deviation of 3 and 3 sigma of 9
what happens to our mean as we sample more?
like take temps at a bunch of stations, say 5000
a day for 100 years, heck lets just do it for
50 to 60 years
> g mean(g)
[1] 0.4393237
> g mean(g)
[1] 0.04378581
> g mean(g)
[1] -0.06661664
> g mean(g)
[1] -0.01684307
> sd(g)
[1] 2.996614
Notice something?
For a period of nearly 35 years high altitude stations appeared to indicate cooler temperatures. What causes the bias oscillation in Figure 5?
Wow. Lot of work involved in that analysis. Thanks for sharing your assessment.
I have a rather fundamental question about GHCN data series. When stations were dropped as in 1990 – 1995 were the historical temperature values/anomalies for those locations purged/excluded from data sets used to compute global average temperatures or temperature anomalies? That is, is the GHCN data series downsized backward in time as well as going forward? If so, then it seems to me your conclusion that, “Changing altitude does not bias the final trends in any appreciable way” is not strongly supported by your analysis.
Table 5 shows the average anomalies reported for stations with altitude of 300 from the early 1950s till the mid-1980s. The divergence in those means is frequently 0.1 degree or more. From 1998 forward, the relationship between the two anomaly series reverses and values for stations 300 meters. Tables 4 and 5 report that more than half of stations dropped from GHCN in the 1990s and in 2005-2006 were positioned at altitude of less than 300 meters. That majority-or-better weighting factor for <300 meter altitude stations together with the relative temperature anomaly patterns observed over time suggests dropping stations from GHCN contributed to steeping of the post-1950 temperature anomaly trend through 1998 and retarded the decline in temperature anomaly trend post-1998.
FWIW
What I would really love to see is a study on diurnal variations from overnight lows to mid-day highs and how they have changed over the years.
latitude says:
August 19, 2010 at 8:47 am
When you’re talking about reading temps in 1/100ths and 1/10ths of a degree,
I see you’re talking about the same thing I am, 1/100ths and 1/10ths of a degree. I see it in comments now. I hope no one thinks I copied you.
We are probably seeing similar things. The focus on anomaly is a distraction to the real issue which is temperature. Anomalies can look virtually the same and be called “virtually the same”. But that is misleading. The uninformed person will think nothing is wrong between the data sets when hearing that, for example, that GISS and NOAA are not different than any other data set.
But when you talk about temperatures having only a 1/100th to 1/10th degrees difference, as in the GISS set, between being the hottest ever or not being the hottest ever, and that 1/10th to 1/100th of a degree is the only thing needed to make the media, and people like James Hansen, talk about dangerous global warming, then the average person will know global warming alarm is ridiculous.
david
There is a clear reason why altitude should make little/no difference.
if you think about it.
Imagine a planet with no people. It the flatland its 10C. at the top of a mountain say its 5C. Why? well we know why, the lapse rate.
Now, turn up the sun. Turn it up enough to raise the temperature by 5C. As the world warms up what do you expect the top of the mountain to be at? 10C.
and the flatland? 15C. Is the mountain still cooler? yes. Did they both rise the same?
yup. Why? Lapse rate. IF ANYTHING the theory predicts that in a warming world
The higher stations would warm MORE
” Both theory and climate models indicate that global warming will reduce the rate of temperature decrease with height” That aspect of the theory has yet to be verified because the change is small.
So in a warming world you might see a differential in the rate, but HIGHER would warm MORE..
So here is the silliness of claiming that the higher stations are dropped on purpose.
THEORY predicts that higher will warmer slightly faster because of changes in lapse rates. So they would be dropping stations that, if kept, would show MORE warming.
Basically, the physics says that in a warming world you might see a change in lapse rate that would have higher altitudes warming faster ( see the Fingerprint) And so skeptics position on that is what? The opposite? That in a warming world lapse rate will change the other way? on what basis? or do they think that lapse rate will stay the same? Well, which is it? if they think that lapse rate will stay the same, then they are saying that high and low altitude warm/cool at the same rate. If they think that higher will warm LESS that lower, then they have to explain how and why? and the have to explain the differential we do see in surface trends and trop trends.
WRT
“I do want to comment however on the use of anomalies. As Mr Mosher contends, the use of anomalies solves a great number of problems in comparing trends and data from a wide variety of sites at different altitudes, latitudes, and so on. But the use of anomalies does create a new problem and that is one of perspective. ”
I claim no such thing. I think you misunderstand. It doesnt SOLVE problems.
It prevents them.
A simple example:
3 stations:
1: 10 10 10 10 10
2: 4 4 4 4 4
3: 1 1 1
All measures in C
Average the temps:
5.0 5.0 5.0 7 7
You see? Droppin the station 3 CHANGES the average. Problem.
Now anomalize! subtract the mean of a station from itself.
1: 0 0 0 0 0
2: 0 0 0 0 0
3: 0 0 0
to Anaomalize you subtract the MEAN from the measurement for each station
so 10-10, 4-4, 1-1
Now average the anomalies
0,0,0,0,0
Wow. The temperature is FLAT. that is one of the points of doing an anomaly.
To PREVENT the problems of doing a simple average of temperature. NOW, IF
the records had no missing data, well you could use temps. I had section about this in the paper but we pulled it because it was too obvious.
When people focus on bogus issues they diffuse attention away from the correct problem. that problem is UHI and metadata and adjustments
“how virtually no change over the past 100 years, while the nighttime temperatures indicate the developed Valley has warmed significantly while the undeveloped Sierra foothills have not.”
Have you done a detailed analysis to test John Christy’s findings?
*******************************************************************
see two KEY THINGS here.
1. DEVELOPED VALLEY. The issue is NOT altitude the issue is DEVELOPMENT.
UHI. Period.
2. The stations dropped tended to be in mountainous valleys
It comes down to this. The real issue is UHI. Focus on that. Ross, tried to argue that the drop in altitude MEANT a move into more urban areas. My argument is this.
IF the problem is a move into urban areas, then focus on that. ALTITUDE pure and simple has nothing to do with it. MOVE into urban, REGARDLESS, of altitude is the issue. the only issue. The issue that needs 100 percent focus. NOT a change of 50meters in the average altitude of stations. Thats a DIVERSION. since I agree with christy that moves into developed areas is important, I will do everything to reficus the issue on the real concerns. altitude is a distraction. altitude as a proxy for development is silly when we can just look at the devlopment.
clear?
E.M.Smith says:
August 19, 2010 at 10:52 am
So, IMHO, we have crappy data and get crappy results from it. Admiring the uniformity of the crappiness does not yield much comfort.
LOL!
I know you were trying to make a serious point. But it’s a good thing there was no milk in my mouth at the time. 😉
very Nice Hu!
a while back Chad was starting simulations of the various methods.
I actaully prefer Romans method, Then first differences, then CAM a comaprsion will be nrat
Steven,
Thanks for the several posts. If I correctly understand your various points and references, the rating system was adapted from one in use by the French, and the error range stated for each CRN category was the maximum range of error, and not the mean error. And that a station may change rating category through the seasons due to environmental factors, say proximity to trees, which would tend to bias summer temperatures, but less so the winter ones. Reasonable enough, but note that I did not assert a possible error in the +2C range, but more in the range of 3-5 times +.15C, or +.45C to +.75C.
Based upon what you’ve shared with me, it seems clear that without knowing the mean error of the stations in a given CRN category, simply knowing what percent of stations fall in a particular category doesn’t produce a meaningful answer as to the temperature bias in the records. Dr. Leroy’s associate showed a +.1-+.15 error range over CRN2-CRN4 stations, but that was without the inclusion of CRN5 stations. You mentioned a study with a too small sample size which suggested that error range MIGHT be too conservative, and the only way to tell was to gather the data. Which is what was done by Anthony’s Surface Stations project.
One thing seems clear to me from ChristianP’s comments and from the sites I surveyed for the Surface Stations project, is that the status of stations needs to be constantly monitored, and that adjustments need to be regularly updated to reflect current conditions at the site. For example, I was told that the shed at the Telluride site had been built there about 11 months earlier, and that the City had put it there without asking or telling NWS manager for that site. Yet when he found out about it, nothing was done about the station until we did the site survey, and then they immediately pulled the station from that location entirely.
ChristianP also mentioned they were using a modern Stevenson Screen (in two sizes, I take it) and that reminds me that each site I surveyed had been “upgraded” to the MMTS from the Stevenson screens originally in place. In each case the placement of the MMTS was MUCH closer to buildings and other structures which would bias the temperature record upward, if not properly adjusted for. The MMTS has a data cable which must be run inside a building to a data recorder. The cable must be put in a trench. Someone must dig the trench. With a hand shovel! Naturally, most trenches were not dug very far from the structure the data recorder was put in. Not supposed to be that way, but try digging a 50 foot long trench in mountainous soil with a hand shovel sometime. (Much less out to the 100 foot standard!) I’ve looked at other surveys at the Surface Stations project, and this seems to be pretty common for other areas of the country as well. But, this shouldn’t necessarily make a station unfit to use if adjustments for each site are properly and regularly done, but that’s one of the real questions, isn’t it?
Steven, like yourself, I’m willing to be convinced with the data. In the absence of good and convincing data, my default position is that the null hypothesis is preferable: That any warming in the record can be explained by natural climate variations and observation errors. My sincere thanks for your ongoing contributions here at WUWT.
Dan
I’ll repeat an earlier question, and a new one.
People use the term GRINS … what does it mean?
And what does UHI mean?
Thanks.
Rex from NZ says:
August 19, 2010 at 10:38 pm
And what does UHI mean?
Urban Heat Island. Look at the top of the page. There is a “Glossary” tab.
I am with GeoChemist and Tallbloke. The implicit asumption behind this post is to “prove” McKitrick is mostly wrong and that temperatures match the “pace” of “global warming” with all the connotations that its man made, going to be catastrophic and it is all because of CO2. I don’t buy this package either. I live in NZ and here we have “adjusted” temperatures which match “adjusted” CO2 to prove CAGW. As a consequence, we have an emissions trading scheme. There are very few real scientists who are not taking an advocacy position.
Nick Stokes is a confirmed warmer and spouts on about the long term – “since records began” – trends, especially the Arctic. I can agree that some temperatures have gone up a bit over the last 100 years – so what? There are places where the temperature has gone down. If CO2 causes warming how can this be? Why must we use averages to hide those places declining. If physics can explain CO2’s effect on global temperatures, how come this physical effect has not happened since the start of the century? Arm waving that it doesn’t have to happen straight away is all I hear. If I drop a weight, I expect it to fall to the ground. If it doesn’t drop to the ground until one hour later, surely there would need to be a new law to explain it – not arm waving.
The Arctic is another example. I’m told the long term trend is down, ice free by any time between 2010 and 2020. The trend is not long term – just since satelites began measuring. Nick believes he can ignore earlier records because that trend would show growth in ice at the arctic and Greenland.
Glaciers are melting, oh no. In New Zealand all of the large river valleys were formed by glaciers which no longer exist. The Haast River Valley is vast – the glacier gone and not because of CO2 or global warming. Please explain this in a way which shows that glaciers no longer melt unless caused by AGW.
The sea will engulf Tuvalu, oh no. But the Pacific ocean is not rising. The latest news in Australia is the Murray Darling river is drying up. Scientists are worried about the future effects of global warming on its recovery. Everyone here knows that excessive water from the Murray Darling was taken for irrigation projects, so why blame GW, let alone AGW or for that matter CAGW. No – scientists are saying that the recovery may not happen because of CO2. There again, the recovery could happen when excessive water take is stopped. Might, could, possibly, likely and so on.
I get sick of being manipulated. I am sick of hearing when its hot its Climate Change, when it snows its Climate Change, when it is merely cold its weather. I know a weather forecast more than a few days out is unreliable, but someone tells me that a climate scientist can predict long term weather. How many times have they done this accurately? Zero. When they have correctly predicted two consecutive 30 year climates, I could be more attentive – but I will never see it in my lifetime.
And the punchline is the problem can be solved if I pay between $10,000 and $100,000 a year to somebody called Al Gore and partners. Zeech.
Alan
Steven Mosher says:
“Wow. The temperature is FLAT. that is one of the points of doing an anomaly.
To PREVENT the problems of doing a simple average of temperature. NOW, IF
the records had no missing data, well you could use temps. I had section about this in the paper but we pulled it because it was too obvious”
I’m afraid you missed my point. Yes, everything you said about anomaly comparison is true. But it still removes perspective. Hello Winnipeg, just so you know, because of global warming expect highs this summer of 39.7 degrees in August instead of 39.3. Oh, and this winter, your lows will be -39.7 instead of -40.5. I’d survey denizens of Winnipeg to see their reaction but with temps in the 30’s they’ve deserted the city for lake country and in the winter door to door surveys are not appreciated.
Anomalies have plenty of value, but to be meaningful they eventually have to be translated into the actual temperatures that actual people will experience in actual locations or the execrcise is menaingless. When you plot the magnitude of the anomalies against the temperature range that humans have lived at for centuries, you get a perspective as to just how insignificant the change is. Particularly when the coldest parts of the planet warm the most and the warmest parts of the planet the least. Winters warm more than summers and nights more than days. The average across all of this may be 0.6 degrees per century or something in that range on a global scale, but it still comes back to the same thing. An artcic reqion that warms will have slightly warmed summers but milder winters with most of the warming coming in the parts of the year where plant and animal life are both stressed to the limit by cold. OK Mr polar bear, you’ll have to survive -42 this winter instead of -49, but the really bad news is this summer you will have to put up with a swealtering +12 instead of +7.
Mr Polar Bear: “How do I get more of this “problem”?
Some people may not be understanding the difference in importance between anomaly and temperature. Think of it this way:
If you put back into the record all of the mountain, ‘mountainous valleys’, and all other temperature stations that are no longer used in the record, and instead, dropped all of the ones that are now in the record, the urban, and airport, etc., you would the same anomaly (virtually, theoretically) but you would have cooler temperatures.
Hello everyone. Thank you Steve, Nick and Zeke. I am on holiday, and on a slow dial-up, so I will do my best to follow this thread but my online time is limited until September. Here are my main comments.
“McKitrick does note that the methods of computing a global anomaly average are sound.”
– I’m pretty sure I didn’t phrase it like that. I have noted that the various gridding methods currently in use tend to yield similar results once the choice of input data is made. This has been demonstrated by the various groups using GHCN data, including Muir Russell’s team. As to saying whether the methods are “sound”, that goes further than I prefer to.
“the process of dropping stations is more related to dropping coverage in certain countries rather than a direct effort to drop high altitude stations”
– Altitude is likely not the deciding factor in dropping a station. It is more likely that a station is dropped due to cost of collecting the data or something like that, and the change in mean altitude is a knock-on effect.
As I see it, you have looked at 3 topics: did the global mean altitude change, was it due to a relatively large loss of inland and mountain sites relative to coastal sites, and did it affect the trends.
On the first, your Figure 1 shows variability in mean altitude but not the 1990 and 2005 discontinuities in Fig 1-8 of my paper, and hardly any change in the mean, by the looks of it. It would be helpful if you explained how you computed mean altitude. Are you arguing that there was, in fact, no change in mean altitude?
Table 1: What is the date for this distribution of stations? Is it something like, all stations over the entire 20th century? It would be helpful to the reader to know what exactly the sample is. Even more helpful would be the counts and % at some key intervals, like 1970, 1980, 1990, 2000 and 2010. You could make your point much clearer and more succinctly if you did that kind of tabulation.
On the second issue, you say “McKitrick supposes that the drop in altitude means a heavier weighting for coastal stations. The data do not support this” But then you proceed to show the sample of dropped stations, post-2005, is skewed towards inland sites. Only 9% of stations dropped are coastal as opposed to 30% in the larger sample, and 83.7% of dropped stations are inland rather than 64% in the sample. This represents a bias towards dropping inland areas and retaining coasts. If you disagree with this reading, you could make the whole point clearer simply by showing the coastal/inland tabulations at some key intervals.
In Table 4 and surrounding text you appear to be arguing that the distribution is not consistent with falling altitude. Looking at your numbers, compared to the entire sample (Table 3), the sample that disappeared post-1990 has a slightly lower median altitude (183 vs 192) but has a higher mean altitude (441 vs 420), a higher 3rd quartile (589 vs 533) and almost the same max (4613 vs 4670). This indicates to me that the distribution of the 1990 fatalities was skewed to higher altitudes compared to the whole population. Then the stations dropped after 2005 are even more skewed: the mean is 510 (vs 420 for the whole sample) and the 3rd quartile is 681 (vs 533 for the whole sample). You then conclude: “That hardly supports the contention of thermometers marching out of the mountains.” Well, first, that’s a straw man. Second, your own numbers show that the distribution of dropped thermometers was, in fact, skewed to higher altitudes.
Another straw man: In one comment above Mosher says, presumably referring to my paper, “The claim in the paper was that the drop in altitude “comprimised the integrity of the data”.” Where is that quote from? What I said was “the failure to maintain consistent altitude of the sample detracts from its statistical continuity.” That’s a pretty carefully-phrased assertion, especially since I said it in the context of a series of charts showing the magnitude of discontinuities in the nature of the sample around 1990 and 2005. So far you are not convincing me that the altitude distribution remained continuous across those intervals.
Going on to Table 6, the 4-way division of landforms is intrinsically less interesting because it does not directly map onto altitude. Why look at these terms as indicators of altitude, when altitude itself is available? Presumably a “Mountain Valley” is itself a high elevation site, at least compared to a coastal valley. Likewise Flatland can be high-elevation flatland or low-elevation flatland. Your table is hard to read, again because it doesn’t have a date to define the sample with, it doesn’t show a before/after comparison, and the % calculations are not defined. Since columns don’t add to 100% we can eventually figure out what you’re doing, but it isn’t what you need to do to make your argument. You want to argue, presumably, that the different land forms each lost a similar percentage of their stations. But your numbers show that the losses were not similar across categories, they were Flat 34.5%, Hilly 37.8%, Mountain 29.5%, MV 42.2%.
On the 3rd issue, at one point you say “What this means is that as long as high altitude stations warm or cool at the same rate as low altitude stations, removing them or adding them will not bias the result.”
– No, even if their trends are the same, there can still be a bias in the overall average, if the discontinuity in the sample causes the mean anomaly to change in a way that cannot be corrected.
Regarding the comparisons of trends, I think you have made a reasonable case that, where they can be compared and all else being held equal, high elevation sites in the GHCN sample have similar or slightly higher trends than low-elevation sites. Whether this is a general principal, I do not know. I am aware of Christy’s work on California valley-vs-mountainous trends, and it seems plausible to me that his findings may apply more generally.
But as regards deciding whether “it matters”, this is very difficult and I would caution against making strong statements, since it’s easy to fail to find an effect when you lack the data to measure it directly. In effect you have 2 sets of data: X and Y. X terminates at 1990, Y continues in the form of y. And we know that X and Y have different sample characteristics. If you only use the information contained in y, Y and X to estimate x (the continuation of X after 1990), depending on how you do it you will tend to estimate it based on the portion of X explained by Y, in other words the correlated portion. Ideally what you should have is some other data, W, which is correlated with X but not with Y, and which continues after 1990. Then you can do some proper statistical modeling. It’s a mistake to be overly nihilistic about the surface data, but it’s also a mistake not to properly quantify the possible biases.
As you have alluded to, the big issue needing attention is (in my terminology) the application of ex post rather than ex ante testing to assess the quality of adjustments, gridding, and all the rest. The testable hypothesis is: If the surface data have been properly adjusted, and If the GCM’s are reliable in telling us the dominant influences on surface temperature trends, Then after the adjustment process is complete we should observe certain patterns in the data and not observe others. I expect you all know my thinking on this, based on the research I have done up to now, and why the IPCC’s treatment of the topic in the AR4 was so horrendously inadequate. I have 4 more projects in the works on this: one nearly in print, one in prep for a conference in the fall, and two just at the data collection stage.
BillD says:
BillyBob says:
August 19, 2010 at 10:21 am
” Given the GHCN data, the answer one gets about the pace of global warming is not in serious dispute. ”
Considering that all of the GHCN anomaly calculators use the mean:
[…]
Having looked at the raw GHCN data, I can say the max is not going up. It is the min.
Therefore it is UHI.
BillyBob:
You are correct that the min is going up much faster than the max. One region where this is especially strong is in the Swiss Alps. One problem with your conclusion is that green house gases are expected to reduce night-time cooling and to have a greater effect on the min rather than the max temperature.
While it’s pretty clear it’s the MIN that gets hit, the reasons are many. Station siting, for one. Over asphalt, the warmth persists longer into the night than over grass. Over the decades airports have added far more night flights. When I was a kid, they were modestly rare. Now the business is 24 x 7 some places. That the temperature series warms in sync with the airport explosion of the Jet Age and has 90%+ thermometers at Airports in many countries is a blatant problem. (And that the “QA” process for USHCN demands a station be acceptable to it’s ASOS neighbors via a comparison to an average of them will also clip and replace more low going temperatures than highs… an average will never go as low as a single station.)
But the bottom line is that it’s much more ‘instrumentation error’ than anything to do with CO2. It’s the IR off the airport tarmac at night onto the badly sited thermometer, not the IR hitting a CO2 molecule in the stratosphere.
@Steve Mosher: Do a strict “self to self” anomaly process on the data comparing a thermometer ONLY with itself and there is a clear ‘shaft’ and a clear ‘blade” and it happens at a clear point in time when the “duplicate numbers” change. Easy to see, easy to show. I’ve done it a few different ways. A modified form of First Differences and a simple long baseline method. You have tended to treat all the duplicate numbers of one ordinal value as a group. You can’t. For some it’s a change from 2 to 3, for others with more ‘dups’ it may be a 4 to a 5. They have no specific meaning and are assigned sequentially. So do it by date and you get the blade.
One version:
http://chiefio.wordpress.com/2010/04/22/dmtdt-an-improved-version/
The one I like better, but is less ‘standard’:
http://chiefio.wordpress.com/2010/04/11/the-world-in-dtdt-graphs-of-temperature-anomalies/
The pattern of changes shows clearly there is impact in some months of the year and not in others, often going in different directions. It is driven by station and process changes, not by CO2. At it’s most basic, it is a ‘splice’ artifact from splicing together different types of stations over time via anomalies. Codes like GHCN will also do this ‘splice’ and get the splice artifacts, but does it by much more convoluted means via ‘homogenizing’.
You completely missed the point on Volatility. Read the link. It’s the STATIONS that have differential volatility. The weather cycles just put them at a warmer or colder end of their ranges on a periodic basis (substantially the definition of a major cycle.)
Not interested in toys. Interested in the real world. Any motorcycle rider can tell you that at night it gets a lot colder out where the plants grow and it is much more nicely warm over the tarmac and in town. Yeah, it’s testable, but be careful that you are not biasing the test… So you take a peach orchard, it transpires a load of water and cools off quickly. Tarmac not so much. (Ridden by many of each, even late at night and early mornings).
And yes, there will be some days that have more AHI than others. This is well known. But since airports are typically near cities, you also have the issue of from which way the wind blows. I’ve found an interesting example near Miami where you can watch the temps rise when the wind comes over Miami. Gave a several sigma event / excursion. You don’t need EVERY day to be extra warmed to bias your averages up over time. Just “enough”.
Per “step change”. Good luck with that.
Airports are far more dynamic than that, and on a time scale that covers the ‘warming’ found. SJC was a near nothing in 1940. In the 1960-70 era it had one terminal and you could walk from your car to boarded on the plane in 10 minutes. I did it in 5 once, seat to seat. Now it’s 2 major terminals and far more runway. Huge amount more traffic, loads more hours of operation. The same thing happens around the world. Airports are not built once in 1950 and frozen in time. The grow and expand. A Lot. Fuel use rises. We swap from pistons to jets with massive hot exhaust. We move from 707 to 737 to 747 to… and all the time the passenger load / year climbs. More than enough stations like that to move an average.
So pick long lived stations that have 100 years of time in service and are not airports and look at them. The “Industrial Revolution” does not warm them. (I think it was Tony B did that work).
Per the ‘cooling since the little ice age’. Don’t know where that comes from. We warmed out of the little ice age into about 1825, then did some cyclical warm / cold cycles with the 1930’s being quite warm, especially in the USA where most of the thermometers are located. But we’ve been cooling lately (despite hysterics from some folks) and are in a downhill slide since the PDO flipped.
One other point. It does little good to say a given station trend is all that matters when what is compared in codes like GIStemp is NOT a station trend. It is a Grid Box trend that is found by using one set of stations in the baseline and a different set in the present. The “trend” if found by comparing my 1967 VW with my present Mercedes SLC and finding that cars are faster now. Sure, there is a bunch of magic sauce applied to try to say that things have all been adjusted to make it fair. But it isn’t. THE basic issue in the present climate codes is the use of “box” anomalies computed by comparing one box of thermometers THEN to a different box of thermometers NOW. So any analysis / proof / validation done comparing trends of particular individual thermometers does nothing to show the validity of the method actually used. (It can show the nature of the data, and it shows no warming to speak of from 1825 to 1990, a ‘dip’ in the 1950-1980 range, and a ‘hockey blade’ in 1990).
And that is where the volatility issue comes in. A ‘high volatility’ group of stations records a low value in 1950-1980, then a ‘low volatility’ group of stations can never reach that low value again. They lack the ‘range’ to reach it. Explained in detail in the link with example stations data.
It is the failure to use actual “station to itself” anomalies that is the broken part of the process used by CRU and GISS and NCDC.
So you do a lot of ‘station A vs B’ warming trends. But you don’t do “Station A during cold PDO and station B during warm PDO with with A and B having different volatility ranges during each” then splicing the two series together to get a ‘trend’ via various box filling and homogenizing. You miss the trick so declare the magician really did saw the girl in half.
Per the comparison of pre-1990 stations. The problem is that we don’t have the data for them POST 1990. We don’t have both sides of the volatility problem to see. As long as you are doing “box to box” you can not find out what “station to itself” would have done with the stations that are not there. So that “pre-1990” with / without test is incomplete.
One final note on bias. The Meteorologists in Turkey did an analysis with all the available stations and compare it with the “GHCN kept” stations. They found cooling. GHCN finds warming. They have a published paper (I’ve put links up to it several times) and it basically shows that station selection bias turned a cooing trend into a warming trend in Turkey.
That’s one mighty big cockroach I see… wonder how many more in in the shadows…
Now one of the things I’ve noticed in wandering through GHCN that initially caused me some puzzlement was that when taken in aggregate things would often show no change, but when looked at in particular, there was an impact. This could be an accidental result of an attempt to avoid bias by a broken means (forcing specific changes to a standard no-change average) or it could be a direct result of assuring some metric is keep ‘nominal’ while the deck is being stacked.
I come from a forensics background, so I’m more used to this kind of thing than most. But, for example, if you have a company that orders equal number of buns and burgers, you expect the two counts to match. If someone is stealing burgers, buns build up to excess. So you cross check those two. A clever burger burgle would take both buns and burgers so the counts matched. Now audit the details and find that on even days burgers disappear and on odd days buns disappear… Hmmm…
It’s that kind of forensic mind set that leads to all the highly detailed (and sometimes tabular) comparisons I did. You ‘cut the deck’ every possible way, not just the way the dealer suggests… And that is also why I’m not very impressed by “studies” that show one particular cut of the deck finds everything matches up just fine. I would expect it to.
And that brings me to the Atlas Mountains.
I found the thermometers running INTO the mountains. Yet Morocco was ‘warming’. Odd. Until you realize that there is a cool ocean current off shore and the Atlas Mountains were much more desert warmed.
OK, now compare the ‘anomalies’ of those two ‘station to itself’ and you may not find much. But what happens when the cold thermometer is in the “box” at the start, then a warmer one in the “box” at the end? (Remember, we don’t have an anomaly between those two time periods for the thermometer to itself, only box to box…) And what if the cold thermometer was in during a particularly cold AMO to warm cycle, then moved to the desert where the waters can’t influence it. Gee, we can lock in that AMO warming trend and hold onto it. Avoiding the next cooling turn to the water and the breezes off shore.
So it’s not enough just to look at “by altitude”, though it gives some strong clues. It’s much better to look at “by altitude, by region” or “by altitude, by country”. And over aggregating things into an “average bucket” (or box…) just assures you are not looking up the right sleeves… While using a test of ‘station trend’ to validate ‘box trend’ is just missing the whole basis of the trick. (or error… to leave the metaphor behind).
Rex from NZ says:
I’ll repeat an earlier question, and a new one.
People use the term GRINS … what does it mean?
And what does UHI mean?
UHI is Urban Heat Island. AHI is Airport Heat Island.
“just for Grins” means “just for fun” or “as an example for fun illustration and not a formal proof”. Or, as my sons chem teacher derisively answered when I asked about dong practical chemistry for the kids: “Oh, you mean DEMONSTRATIONS”. (He didn’t see the point, being a dull and slow fellow.) So one might hear “Lets go do some cow tipping just for grins” or “assume I have a million dollars, just for grins, what would we do with it?.
Scan this video to 2:35 and play to 3:27
Dave says:
I have a rather fundamental question about GHCN data series. When stations were dropped as in 1990 – 1995 were the historical temperature values/anomalies for those locations purged/excluded from data sets used to compute global average temperatures or temperature anomalies?
The data are kept in. New data is not added. So one might have a station in Sacramento and one in San Francisco during the cold period, then drop any NEW data from Sacramento. GIStemp then makes up numbers for Sacramento based on the temperatures in San Francisco and the historic relationship between them.
Bolivia, for example, ends about 20 years ago. All the temperatures you see for Bolivia in the global maps are based on the OLD real thermometers compared to present made up ones.
That is, is the GHCN data series downsized backward in time as well as going forward? If so, then it seems to me your conclusion that, “Changing altitude does not bias the final trends in any appreciable way” is not strongly supported by your analysis.
While true, it’s worse than that. The missing data are often ‘made up’ via various means. Depending on the details it can be called ‘homogenizing’ or ‘The Reference Station Method’.
So take SF vs Sacramento and establish their relationship during a cool time period (when Sacramento is much cooler than SFO). Now during a warming time, ‘make up’ Sacramento by looking at SFO. SFO can be VERY HOT during hot excursions, so you will now make up a real scorcher for Sacramento. A value that will not be real. This is the volatility issue in the link above. SFO can get to 100-105F during hot excursions, about 5 F behind Sacto. But during cold excursions SFO may be 5 F warmer than Sacto. So establish your ‘cool’ offset as a minus 5 F, then apply it when SFO is really +5F and you get +10 F (when Sacto is really not that hot). As a fictional example.
Basically the cold range of Sacto is far colder than SFO while the warm range is only a tiny bit warmer. You can use that ‘delta V’ to your advantage via splicing over time and ‘fill in’ of missing data to create any anomaly trend you like. The artful addition and deletion of stations based on volatility and where you are in a long duration cycle like the PDO is all it takes. It exploits the small comparison period of The Reference Station Method to establish a fixed relationship when it really is more dynamic and it exploits the station volatility with long duration cycles (that is typically ignored in the codes) to let you make any anomaly trend you would like as a side effect of “box to box” anomalies.
Or more simply:
It’s all about the splice. What, when, and how.
Steve Nick and Steve
Mosh
You have done some interesting work here, very well done. I know how difficult it is to write an article let alone put it up for public scrutiny.
I appreciate that in this article you are interested in the effects of Altitude, although as is always the way the discussion has broadened out considerably. As there are many laymen lukers here I wonder if it might be useful to put your article into a historic context for them?
I have long been interested in historic temperature records and in order to examine the frequent assertions that we have no instrumental records of the LIA established my site here.
http://climatereason.com/LittleIceAgeThermometers/
To set the scene I wrote about the reliability of global temperature records here.
“Article: History and reliability of global temperature records. Author: Tony Brown
This article (part 1 of a series of three) examines the period around 1850/80 when Global temperatures commence, and looks at the long history of reliable observations and records prior to the development of instrumental readings.”
http://wattsupwiththat.com/2009/11/14/little-ice-age-thermometers-%e2%80%93-history-and-reliability/
Within that article was a tabular version of Chiefios’ wonderful ‘March of the thermometers- which has been referenced in this thread several times.
In a fine example of Anglo American cooperation Verity Jones and Lucy Skywalker kindly put the information into an excellent graphical form which clearly shows the changing numbers of thermometers over the years at various locations. (It is towards the top of my article here).
http://climatereason.com/UHI/
Delving into the records -both instrumental and written/observational- it became increasingly clear that far from Global warming starting in 1880 we had seen rising temperatures since the 1690’s (the LIA in its severest form effectively ended by 1698)
It became clear that James Hansen in setting an arbitrary start date of 1880 for GISS had missed out on a whole history of rising temperatures which I wrote about here
Article: Three long temperature records in USA. Author: Tony Brown
This article links three long temperature records along the Hudson river in the USA. They illustrate that a start date of 1880 (Giss) misses out on the preceding warm climatic cycles and that UHI is a big factor in the increasingly urbanised temperature data sets from both Giss and Hadley/Cru
http://noconsensus.wordpress.com/2009/11/25/triplets-on-the-hudson-river/#comment-13064
and in a strictly UK context here;
http://noconsensus.wordpress.com/2010/01/06/bah-humbug/
The gentle slow rise over the centuries has been graphically reproduced with some of the longest data sets here
http://i47.tinypic.com/2zgt4ly.jpg
http://i45.tinypic.com/125rs3m.jpg
The UK figures from 1660 can be more clearly seen here together with its linear regression.
http://homepage.ntlworld.com/jdrake/Questioning_Climate/_sgg/m2_1.htm
So we have a world that has been generally warming for the past 300 plus years. However it does not appear to be a global thing-(perhaps why terminology has recently been shifted to ‘climate change’. There is plenty of evidence to show that there are a number of locations that have bucked the general warming trend over a statistically meaningful period and are static or cooling. I hope to do a post on this in due course with some colleagues.
The main point I wanted to get across is that it would be useful for researchers to see current temperatures as part of a very long established natural trend, rather than that it has come about recently and must therefore be ‘our’ fault..
I will leave the nuances of the discussion to others here especially as Ross and E M Smith have turned up who can discuss the key part of your study concerning altitude.
Tonyb
I’m currently working on a temp reconstruction with the criteria that it shows the average, max, and min.
The way I’m getting over the station dropout issue is that I’m doing the averaging at the very last step. I’m using the daily GHCNv2 unadjusted.
From that I’m taking max/min temps with the only criteria being that they don’t exceed world records and that they’re balanced. I add all of these temps (they’re in tenths of a degree so they are all integers) together for a given year (I just want to see yearly for the moment) and record the number of observations. These two figures along with the min and max for all the years that station records are then stored in a file.
When I need to calculate a region or country, I open up the temp station files for that area and then add all the temp sums and observations for each year and only then compute an average for a year.
If stations drop out or are added, all that happens is the overall sum of temps and the observation count rise or fall in step with each other, and the computed average is valid. It’s a much less compute intense way of getting a Global Average Temp, with the advantage that none of the information is lost.
I’m still about a month away from releasing it (I still have to earn a living 🙁 ), but what I can say is that it does show that min temps have risen more than max temps, and that that is what seems to be driving global temps up. This is not a signature one would expect from CO2 induced warming.
Ross,
I calculated the average altitude of 1990 fatalities – yes, there were 919 of them, ave altitude 550.6 m (high). This was the year of Turkey, Canada, China and Japan.
It seems to me that if there is a suggestion that selective reduction of high or high-latitude stations biases the trend, the most direct test is Monte Carlo simulation of such a reduction. “Try it and see”. I’ve tried various selection strategies without producing a major trend effect, but I’d be happy to try others if there’s something I’ve missed.
E.M.Smith says: August 20, 2010 at 12:24 am
“Bolivia, for example, ends about 20 years ago. All the temperatures you see for Bolivia in the global maps are based on the OLD real thermometers compared to present made up ones.”
We do have over thirty GSOD stations reporting in Bolivia during this period. The story they tell is not substantially different to the GHCN account deduced from neighboring stations.
I have a comment/observation about anomalies.
In the piece above, it states “When we create a global anomaly we prevent this kind of bias from entering the calculation by scaling the measurements of station by the mean of that station.”
This indeed would eliminate biases due to dropped stations. However, is this what is really done? Hansen 1999 section 4.2 states “As a final step, after all station records within 1200 km of a given grid point have been averaged, we subtract the 1951-1980 mean temperature for the grid point to obtain the estimated temperature anomaly time series of that grid point.”
There is a big difference in calculating an anomaly by subtracting the station average from the station data and subtracting the gridpoint average from gridpoint temperature. Following Hansen’s method, dropping colder stations in later years would bias the more recent temperature anomalies upwards.
Wouldn’t it?
“C James says: August 19, 2010 at 9:11 am
The real question to me is why are all of you bright guys spending so much time on verifying (or not) that the use of bad data, regardless of methodology, produces similar (or not) results? Why isn’t there a concerted effort on everyone’s part to go back to the raw data and start over?”
The analysis described here uses GHCN v2.mean, which is raw data. That is, it comes straight from the CLIMAT forms submitted by the various Met organisations. That’s what people like Zeke, Steven and I use. GISS and CRU may subsequently adjust it (with good reasons).
” Paul says: August 19, 2010 at 7:59 pm
Unfortunately, this essay fails to demonstrate a lack of bias due to altitude changes. “
Paul, you say that the analysis doesn’t discriminate between Land Use changes and GHG changes. That’s true – we only used the measured temperatures, which don’t show that distinction. All this kind of analysis can show is whether selecting low altitude stations biases trends as observed.
Steven Mosher [August 19, 2010 at 5:09 pm] says:
I would not that on many occassion I get people arguing 2 things:
1. We cant say that its warmer today.
and later they say things like..
2. “Well we are coming out of a LIA.” a period for which there is very little measurement, Yet, Yet, no skeptic questions it. I find that ironic.
Uggh! Sounds like you are calling ‘skeptics’ hypocrites. That quoted dialog looks exactly like something Joe Romm or Grant Foster would write.
The kindest way for a ‘skeptic’ to interpret this would be as clumsy strawman argument with your anecdotes 1) and 2), points which are not even mutually exclusive, a requirement for hypocracy (or irony 🙂
So, to be kind I ask you to stand at 5th and 33rd in NYC and look up a quarter mile to the top of a famous building. Now roll the clock back 20,000 years and see the glacier that is twice (or possibly 4) times as tall. Roll the clock back further to 4 million and 65 MYA and you’ll notice something quite different.
Now Steve, I can and will state with a probability of one that the climate is warmer today than 20 KYA in that location without possessing a single measurement. Am I still a hypocrite (or ironic)? So unless you are now with those trying to disappear the recent LIA (at least there are a few measurements) I would re-think this one.
A less kind way for a ‘skeptic’ to interpret the above quote would result in gifting you a psychoanalysis book, with the chapter on Stockholm Syndrome tabbed and highlighted 😉
tonyb says:
August 20, 2010 at 1:52 am
“Chiefios’ wonderful ‘March of the thermometers- which has been referenced in this thread several times.
In a fine example of Anglo American cooperation Verity Jones and Lucy Skywalker kindly put the information into an excellent graphical form…”
Tony, the ‘correct version’ is here. http://2.bp.blogspot.com/_vYBt7hixAMU/SzpY-r3HTbI/AAAAAAAAALI/bIMyN3qFjGE/s640/March+of+the+Thermometers.bmp
IIRC correctly E.M.Smith chose 50 Deg as one of his latitude cut off points, and both my and Lucy’s previous versions connected to the wrong latitude lines. Anyway we have sonce realised that the GISS lattitude bins e.g. 64N-90N give equal areas.
Steven Mosher says:
August 19, 2010 at 6:41 pm
bottomline:
LOW altitude stations WARM AT THE SAME RATE as HIGH altitude. its simple physics. Think about it. If the worls warms 10c over time, do you think that WARMING can be confined? the warm low sites get warmer and the high cool sites get warmer. and they warm at the SAME RATE over TIME.
You might be right if we forget about natural influences like Ocean Currents, Jet Streams, Weather Systems, Continental Effects, Inversion Layers, Diurnal Ranges, Sea Breezes, Prevailing Winds, Climate Cycles etc etc etc…. You might be right if you forget about human influences like UHI, Jet Engines, Land Use, Population Growth, Economic Cycles, Depopulation etc etc etc
Unfortunately, real world temperature measurements are impacted by such things… so your bottomline is really scrapping the bottom of the barrel.
If we could just use the Lapse Rate then we could start reducing the number of stations to, say, one per degree of Latitude…
OH! I see they have started doing that already! I wonder why!
Steven Mosher says:
August 19, 2010 at 9:55 pm
Wow. The temperature is FLAT. that is one of the points of doing an anomaly.
And the other reasons for doing an anomaly are Obscuration, Manipulation, Misdirection, Disinformation, Deflection etc etc etc.
They are pre-baking the Station data so they can produce a fully baked grid of extrapolated data that drives a meaningless Global Average. Get Real!
Nick Stokes says:
August 20, 2010 at 3:05 am
We do have over thirty GSOD stations reporting in Bolivia during this period. The story they tell is not substantially different to the GHCN account
Graphical evidence? What does ‘not substantially different’ mean in this case Nick?
deduced from neighboring stations.
Neighbouring as in “in another country”?
I am in agreement that knowing the quality of the data at its source is of paramount importance. It is the foundation of the pyramid and therefore very important.
I take exception to the first paragraph of this post:
Several independent researchers are able to calculate the Global Anomaly Average with very little differences between them.
So?
The crucial question is : does the Global Anomaly Average have a meaning as far as energies impacting on the whole globe and therefore local temperatures?
I am sure that several independent researchers would calculate the same average phone number for the phone book of New York. Does the output have a meaning?
I have said several times that anomalies are third level convolutions and have a distorted connection with the primary objective, energy received and radiated.
The first level of convolution is from energy to local temperature, the second level is from local temperature to local anomaly, the third level is from local anomaly to global anomaly.
There is absolutely no way one can deconvolve from a global anomaly to a local energy, and hence to a global energy budget.
Local anomalies at a more esoteric level also include various mechanisms of heat transport that have tenuous ( again through convolutions) connection with radiative energy. Example : there are at times in the winter 15C and more anomalies in the arctic. These are the result of air currents and not radiation. Nevertheless they are happily averaged in the global anomaly. The recent heat in Russia, is due to stalled air, and should be demonstrating how important the air transport mechanisms are on the temperatures displayed.
The whole global average business needs a rethink from the foundation, in my opinion. Of course if even the data are not to be relied upon, one should just give up.
Steven Mosher says:
August 19, 2010 at 6:41 pm
Latitude. Over short periods of time higher latitude warm MORE. physics also.
Higher latitude are cool. but cool is not the question RATE OF WARMING. that is the Q.
=======================================================
Steve, I would like to see all of the stations that were dropped (high, low, medium, valley, under water) analyzed. It’s too much of a coincidence that many stations were dropped and there’s an immediate 1/2 degree jump in temps.
Those stations had something in common. The only way to figure that out, is to take apart each one and figure it out.
And over short periods of time, higher stations cool more. If they are taking two readings a day, the lower night temps is where they lost that 1/2 degree.
High, low, medium, doesn’t matter. I would like to see every dropped station analyzed.
A STORY OF OUR TIME
Some poor guy has a very serious road accident… the ambulance arrives and carts him off to a good teaching hospital… where he dies.
This guy’s wife is obviously very upset and asks the hospital “What Happened?”.
The hospital replied:
We’re sorry. Our top specialists did everything they could do.
Our top Dietician said he was not overweight – especially as he had lost a lot of blood.
Our top Psychiatrist said your husband was not mad… just angry and in pain.
Our top Nail Technician said your husband’s remaining hand got a lovely manicure.
Our top Chiropodist said your husband didn’t have any ingrown toenails.
Our top Dentist said your husband did not need any fillings in his few remaining teeth.
Out top Cleaner said your husband was a model patient and never left his bed.
Out top Priest gave your husband the last rights although he was not a Catholic.
Our top Lecturer said he demonstrated typical accident injuries to her students.
Our top Administrator said these specialists worked to the highest possible standards.
==============================
This is the sense of unreality I get when I read postings validating AGW data.
PS
Our top Speech Therapist stated that your husband could articulate “I need a doctor now!” very clearly and spoke with a lot of expression. However, our top Speech Therapist was worried about your husbands very limited vocabulary because that is all he ever said.
Nice effort. UHI and ‘adjustments’ have always been the difficult issue.
Unfortunately, I do not think those issues will be so easily addressed as the altitude effects of dropped stations. Comparison of satellite lower troposphere temperature trends for low population, high population and ocean areas may help with the UHI question, or at least better define the possible size of the UHI effects.
However, I fear the issue of adjustments could be endlessly contentious, especially since at least some individual stations can be identified where known adjustments seem difficult to justify. If some reasonable upper bound for the combined effects of station adjustments and UHI can be defined, then that would make a great contribution to the debate, since such an upper bound would automatically set a reasonable lower bound for overall warming. It would be constructive if most everyone could agree the instrument record shows a warming between a lower limit of ‘A’ and an upper limit of ‘B’ degrees.
Addendum to my
anna v says:
August 20, 2010 at 5:30 am
Somehow, in this game called AGW morphed to Climate Change the ball of the very basics has been lost.
1) The black/gray body radiation is used for balancing energies to tell us whether we are warming or cooling, BUT, the temperatures used are average temperatures at 2m in the atmosphere. Most of the time, that temperature has a tenuous connection with the skin surface temperature that is the one that is radiating the heat away. The atmosphere has very little heat capacity to be really representative.
2) Have a look at http://earthobservatory.nasa.gov/Newsroom/NasaNews/ReleaseImages/20040421/01_lstday_modism.jpg
to see real temperatures, the ones that should go into the T^4 calculations an ponder the logic of averaging day/night etc
Look at the scale of the first image, the globe goes from -25C to 45, and again, think T^4.
The whole anomaly chase in the context of real energies in a real world reminds me of the ancient greek word Omphaloscopy,
i.e. navel gazing.
.e
You know what would be cool?
A map of the anomaly difference between adjusted and raw by .5DEG grid cell. For each year since 1930.
Where do the anomalies come from? And when.
That kind of anomaly would be useful.
Nice work in the post, and also heartening to see Dr. McKitrick weighing in. Everyone won’t agree (duh), but the issues get a full and fair airing.
Minor point: the introductory text reads
I am almost certain that “degrees” in that sentence refers to degrees of latitude and longitude, rather than to degrees of temperature. It is a reference to the size of the bins (squares of 2 or 5 deg longitude x 2 or 5 deg latitude) as used in geographical gridding procedures.
Terrific thread, from which I’ve learnt a lot. Thanks especially to E.M Smith.
Nick:
Glad to see we’ve establish a baseline about what the essay says.
Some editing of the essay is in order. The essays leaves the reader with the impression of having vitiated McKitrick’s claims, whereas the actual claims made are finely worded and specific in a way that the essay ignores.
Indeed, your essay supports the original claim that we should care about the impact of altitude changes because of a disparate impact of land use changes. That’s not a different issue. That is the issue that was raised.
tonyb says: August 20, 2010 at 1:52 am
Tony, there certainly are long term records. GHCN, IIRC, has records (Berlin) going back to 1701. But GISS is in the business of compiling hemisphere and global averages, and you can’t do that with such sparse coverage. CRU goes back to 1850, but is often criticised about the sparsity of the early data. Even GISS’ 1880 start is sometimes criticised for the sparse coverage at that time.
Paul says: August 20, 2010 at 8:42 am
“Indeed, your essay supports the original claim that we should care about the impact of altitude changes because of a disparate impact of land use changes. That’s not a different issue. That is the issue that was raised.”
Yes, Ross did talk about land use changes etc. But the evidence available, that he discussed, is the GHCN land temperature record, and that does not discriminate attribution. In his summary, he said:
“The collapse in sample size has not been spatially uniform. It has increased the relative fraction of data coming from airports to about 50 percent (up from about 30 percent in the 1970s). It has also reduced the average latitude of source data and removed relatively more high-altitude monitoring sites. GHCN applies adjustments to try and correct for sampling discontinuities. These have tended to increase the warming trend over the 20th century.”
A few things to try to clarify there:
1. An adjusted temperature file is offerred as part of the GHCN data, but the major indices (GISS and CRU) don’t use it. They use the raw data file.
2. I don’t think the adjustments are trying to correct in any way for the change in numbers of stations in the record. They try to correct for changes within each station’s record. In fact, they make no adjustment to a station without a period of data before and after, which means they can’t adjust for the “dropping” of a station.
3. It’s true that Ross attributes the increase in warming trend in that para to the adjustments, rather than an effect on the raw data. But in th epaper he goes on to talk about the effect on GISS and CRU, which use as input not the adjustments, but only the raw data. And the only way they could be affected is if the selection effects that he describes did show up in the raw trends. We are trying to show that they don’t.
Nick at 10.03
Fortunately we have literally hundreds of thousands of highly reliable records that back up the instruments and tell us what was happening. Far from being anecdotal they are much more credible than lumps of old wood.
As regards to ‘sparse’ data, thats why I had earlier referred to Giss’s arbritary start date of 1880. A more logical time would have been 1910-1920 when there was better coverage in both hemispheres and the Stephenson screen had been universally adopted.
Dr Hansen measured from a dip in temperatures in 1880 which accentuated the subsequnt upswing. Have you any thoughts why such a rational person would have chosen such an irrational start date?
tonyb
Nick Stokes: “But GISS is in the business of compiling hemisphere and global averages, and you can’t do that with such sparse coverage.”
Maybe GISS should stop measuring temperatures in the USA (and elswhere) at JUST a handful of UHI contaminated airports.
tallbloke says: August 20, 2010 at 4:59 am
“Graphical evidence? What does ‘not substantially different’ mean in this case Nick?”
The graph shows it best. But the trends are similar. 0.0373 ± 0.0982 C/Dec for the GHCN estimate based on stations over the border. 0.0559 ±0.118 C/Dec for GSOD for the same region, including 30+ stations within Bolivia. And 0.0703 ±0.116 C/Dec for GSOD data, Bolivia alone. Small trends, differing far less than standard error.
Paul
“Indeed, your essay supports the original claim that we should care about the impact of altitude changes because of a disparate impact of land use changes. That’s not a different issue. That is the issue that was raised.”
If the Issue is land use changes, then the focus should be land use changes.
Altitude changes have not been shown to be associated universally with land use changes. If the concern is land use changes, then that can be assess directly without reference to altitude changes.
change in altitude per se is not a problem. We know physically why it should not be a problem. We showed that it is not a problem. Altitude changes may be associated with land use changes, but that has not been established. The claim was that there was a move out of mountains into valleys and low lying areas. Well, there wasnt.
Further, if altitude were a perfect proxy for land use changes, then we just proved that land use changes dont matter!
My preference is to focus on the thing that matters. Land use changes. Not a proxy for it, the thing itself.
Latitude:
latitude says:
August 20, 2010 at 5:30 am (Edit)
Steven Mosher says:
August 19, 2010 at 6:41 pm
Latitude. Over short periods of time higher latitude warm MORE. physics also.
Higher latitude are cool. but cool is not the question RATE OF WARMING. that is the Q.
=======================================================
Steve, I would like to see all of the stations that were dropped (high, low, medium, valley, under water) analyzed. It’s too much of a coincidence that many stations were dropped and there’s an immediate 1/2 degree jump in temps.
Those stations had something in common. The only way to figure that out, is to take apart each one and figure it out.
###############################
STATIONS WERE NOT DROPPED. in the early 1990 there was a COMPILATION of the EXISTING DATA. those stations that were available were compiled. That is why you get the big number. THEN they decided that going forward they would collect data via CLIMAT reports. hence fewer stations. Dont worry, there will be another compilation in 2011. WRT to looking at all the dropped stations. DID THAT. that’s what a study of RETAINED stations does. no difference.
“And over short periods of time, higher stations cool more. If they are taking two readings a day, the lower night temps is where they lost that 1/2 degree.”
you still dont get it.
High, low, medium, doesn’t matter. I would like to see every dropped station analyzed.
Been there done that. you are welcomed to the data. The simple fact is this. Drop every dropped station. Drop the WHOLE THING. Take only the stations that start in 1900 and end in 2010. dont add a thing. no “adding” of high altitude stations in the 50-90 period. Same answer. The drop doesnt matter.
Tony rogers.
None of us use Hansens method. Not zeke, not me, not nick.
Nick doesnt even use an anomaly. That is WHY we argued in our previous paper that all these methods give substantially the same answer
Ross:
“Going on to Table 6, the 4-way division of landforms is intrinsically less interesting because it does not directly map onto altitude. Why look at these terms as indicators of altitude, when altitude itself is available? Presumably a “Mountain Valley” is itself a high elevation site, at least compared to a coastal valley. Likewise Flatland can be high-elevation flatland or low-elevation flatland. Your table is hard to read, again because it doesn’t have a date to define the sample with, it doesn’t show a before/after comparison, and the % calculations are not defined. Since columns don’t add to 100% we can eventually figure out what you’re doing, but it isn’t what you need to do to make your argument. You want to argue, presumably, that the different land forms each lost a similar percentage of their stations. But your numbers show that the losses were not similar across categories, they were Flat 34.5%, Hilly 37.8%, Mountain 29.5%, MV 42.2%”
Sorry, the much longer version of the paper covered this in detail>
Basically I’m looking at the prior distribution of stations before the drop
and the distribution after to see if the loss was uniform per category or not
or if there was a skew: in terms of distribution the fraction of flat goes up
moutain stays the same. Hilly stays the same. valley goes down. All very slight movements. Hope that explains the table. distribution of topography prior.
distribution after, roughly the same. Putting a fine point on it ( I like that) the move is out of MV into flatlands. The percent ( of the total) of MV goes down The percent of flat goes up. Hilly and Mountainous stays the same. Since mountainous valleys ( one could argue) are higher than flatlands you see a small change in altitude overall. But altitude, per se, doesnt matter. If its land use change we are after we should look directly at that. Which you did. Unforntuanely, people might be left with a perception that ALTITUDE in an of itself matters. It doesnt. I find it hard to make people follow the argument about land use, when issues get conflated. That is why I keep harping on the same point: the problem isnt altitude, per se, its not latitude, is not the dropping of stations. Its the land use. its UHI. the other points are confusing what should be a clear explication of the problem
So more clearly here
Flat: 2779 38%
Hilly 3006 41%
Mountain 61 1%
Mountain Valley 1434 20%
After the drop of 1
1820 40%
1871 41%
43 1%
828 18%
Note that about 14 percent of the stations are above 1,000 meters. Approximately 25 percent of the Earth’s land surface is above 1,000 meters. The Highland climate could be described as “terra incognita”. Indeed it’s a big “H” in the classification system, which is so often ignored.
Mosher: “STATIONS WERE NOT DROPPED”
In Canada (using GHCN v2) they went from 500 stations reporting Max/Min in 1975 to 40 in the 2000’s and then even lower.
The GHCN v2 mean station count for Canada reporting hit 700 in 1965 and started dropping in 1981 and is now in the 30’s.
There were more stations in the the late 1870’s to 1899 reporting than in the 2000’s.
Question for a statistician:
How significant is the calculated global average warming trend of the measured local anomalies against the null hypothesis that the actual global average warming trend = 0? I motivate that question with the following observations: We have a very large, very fluid object – the global atmosphere – and we want to know whether its temperature trend is positive, negative or 0. So we randomly (or not so randomly) distribute thermometers through it and periodically (or irregularly) observe each thermometer over periods of time randomly (or not so randomly) distributed both in duration and in starting/ending. This gives us a large number of temperature records, from each of which we can extract a trend. Each trend should come with an error estimate. Some of the trends will be positive, some negative and some, within the bounds of error, will be 0. Now we combine all these local trends into one giant global trend, which should be stated with some estimate of error. It, too, could be positive, negative, or, within the bounds of error, 0. So, to return to my initial question, given the actual errors of the actual thermometers used to create GHCN, and the actual distribution of these thermometers in space and time; and considering their readings as ESTIMATES of the actual local temperatures at the time of the readings; what is the expected error of the calculated global mean trend as an ESTIMATOR of the actual global mean trend? And is the calculated global mean trend significantly different from 0, given the structure of the sample in relation to the population, and the errors inherent in the measurements?
Station altitude per se is not the issue, it’s how representative that station record is of climatic temperature variations in the region (not city) surrounding it. That seems to be the crucial point that is missed in the sausage method of stitching together anomalies from an ever-variable set of stations and then pointing to “the observed anomaly trend,” as if it were an inherent climatic feature.
Anomalies are not legitimately interchangable from station to station, unless there is complete spatial homogeneity of temperature variabilty. That is almost never the case. The march of thermometers that led to a sharp decrease in “rural” stations in key regions makes the constructed anomaly series scientifically invalid. And as Anna V points out, quite beside the central thermal energy issue.
There is something wholly totemic in the premise that, because a similar result is obtained by different people using the same broken data set and only somewhat different processing techniques, the result is thereby validated. Climate data crunchers need to learn a lesson from dumpster divers in recognizing the difference between rotten fruit and something perhaps bruised, but edible. I agree completely with E.M. Smith when he calls the GHCN data crappy and concludes: “Admiring the uniformity of the crappiness does not yield much comfort.”
JT says:
August 21, 2010 at 7:48 am
You pose a very good question to which I have time to provide only a brief answer.
The crux of the matter lies in the fact that bona fide climatic time-series do NOT exhibit a consistent trend. What is obtained by fitting such to available data is highly variable values that depend not only on record-length but also upon start time. In other words, the stochastic structure of climatic variations is such that the trend plus noise model simply doen not fit. Because that structure involves oscillatory components, which may be modelled as autoregressions of fairly high order, there is no analytic formula quantifying the uncertainty. You can be sure, however, that the confidence intervals are very much wider than those for an AR(1) process, which AGW alarmists often invoke. Hope this helps.
Here is a great post from a guy who has worked with Gisstemp.
He is a fellow who helped me with my google earth code.
Station dropout. If the concern that dropped stations will impact the trend ( raise them) Then there is a simple way to test that, as EM Smith noted. Look at what Gisstemp gives you if only look at long lived stations.
Say you start with 1000 stations, then more stations become available through 1990 and you get to 7000. After 1990, you are back down to 1500 or so.
Concerned about the stations gone missing? well DONT ADD THEM IN. If taking them out raises the trend, then never put them in to begin with. If there is an effect to taking them out, then there should be an effect to putting them in, adding in all those super high altitude and high latitude sites should have an impact ( they dont)
http://oneillp.wordpress.com/2010/06/25/the-effect-of-station-dropout-on-gistemp/#more-910
BillyBob says:
August 20, 2010 at 5:05 pm (Edit)
Mosher: “STATIONS WERE NOT DROPPED”
In Canada (using GHCN v2) they went from 500 stations reporting Max/Min in 1975 to 40 in the 2000′s and then even lower.
###########################################
You still dont get it. When the GHCN was compiled and published, the published
WHAT THEY HAD. Its not like they were collecting data all along and dropped stations in the early 90s. After the compilation of the HISTORICAL data (GHCN.. H=historical) they wanted to turn it into a database with monthly updates, so they focused on incorporating stations using CLIMAT. stations are not dropped. Its not a conspiracy.
http://journals.ametsoc.org/doi/abs/10.1175/1520-0477(1997)078%3C2837:AOOTGH%3E2.0.CO;2
This chart gives a better idea of what’s been happening to the temperature record.
You can see where the raw & urban readings begin to diverge.
http://homeclimateanalysis.blogspot.com/2009/12/continuous-stations.html
Kevan Hashemi said…
By special request: the trend we get from stations that reported during at least 35 out of the 40 years from 1960 to 2000. Analysis code is here, and the graph is http://www.hashemifamily.com/Kevan/Climate/Cont_80_1960_2000.gif . And using the same code, but a different reference period, we have the trend from stations that reported for at least 20 years during the period 1840 to 1880 as well, http://www.hashemifamily.com/Kevan/Climate/Cont_80_1840_1880.gif
Mr. Mosher, with all due respect, it appears you have not understood the argument made by E.M. Smith. After carefully reading all his posts on the subject, and your analysis here, I’m afraid Mr. Smith has the better argument.
The basic problem is that you, and all the others who believe in the GISS and other datasets, I don’t mean to imply that you are alone in this, take the data as valid. It indeed may be valid for that moment in time when the observation was made.
But, when the long-term data for a given site consists of multiple stations from different locations, e.g. a station move, or an instrument upgrade, or any of the other changes that occur in far too many stations, there are insurmountable problems. For example, if one has 10 years of data in one location, then 10 subsequent years after relocating the instrument, then yet another 10 years at still a different location, one cannot establish the 30-year norm or mean for purposes of computing an anomaly. Mr. Watts has posted on WUWT several examples of really bad observations that illustrate this.
All the effort expended in defending the GHCN should be better targeted to curing the defects that are created from splicing together disparate data segments.
The summation by E.M. Smith, that of admiring the uniformity of the crappiness, is completely on target. That phrase gets my vote for Climate Comment of the Year.
Smokey says:
August 22, 2010 at 1:29 pm
The USHCN yearly chart you present is very interesting and highly similar to a QCed compilation made via a much-different sampling scheme. Is the numerical data series available? I’d like to do a comprehensive cross-comparison with said compilation and share the results with you.
sky,
I got that chart here. It’s apparently from a June ’09 article. Sorry I can’t be more helpful. ☹