Monday mirth – Old Reliable

DD More- In the description of the four trees in your yard. First, I hope they are only 9, 6, 3, and 3 inches in diameter.
I find this observation very interesting. My study of the morning sun shows that it is the brightest between sunup and 9:30 am local time. This brightness can only be seen during the summer months when the sun swings overhead rather than to the south. I have an upstairs apartment which faces east and west. The paint on the outside wall of the east side has been destroyed by the sun while the west side has held up fairly well. There are other observations around my place that confirms that the morning sun is harsh compared to the afternoon.
I set photo cells around the place to see if they could see the difference and wala, there is a definate peak of radiance at about 9am every morning. The pattern (which I have captured several times in the last two years) seems to begin when the solar cells first detect daylight (earlier than sunup) and has a rounded shape that peaks at 9am (10 for daylight savings) and ends around 10:30.
This pattern is correlated with the increase of temperature. It also is correlated to the increase of barometric pressure that ends when the noon diurnal takes effect.
I have asked the experts about this curious increase of brightness but it seems to be an unknown.
I suggest DD that you put up a solar panel on your roof and start recording the voltage that it puts out when it is tied directly into a voltmeter. My readings peak at 17.5v at 9am. I would suggest mounting them at the peak of the roof looking straight up and not facing the normal sun path tilt.
I also have a photo cell with the voltmeter set on the lowest scale to measure the reflected light from the moon. (used to determine when this pattern begins).
And if there are others reading this blog that are doing this test, feedback would be appreciated.

Mark W – He probably was but my mind was on the way we size trees..

richardscourtney,

Are you really trying to claim the confidence limits of HadCRUT4 are less than ±0.01 K?

My understanding is that Figure 3 is not a CI calculation, but rather the “raw” display of realization spreads from the data providers themselves. Figure 4 shows the results of the quartile distribution and 95% CI:
http://skepticalscience.com//pics/rss_ensemble_box.png
Figure 4: Uncertainty in the temperature satellite and surface temperature trends, estimated from the RSS and HadCRUT4 ensembles. The boxes show the mean and interquartile range of the trends on 1979-2012 in each ensemble. Whiskers indicate the 95% interval (2.5%-97.5%). Crosses indicate outliers.
The units are K/decade, not K, and that looks like ±0.03 K/decade for HADCRUT4 vs. ±0.13 K/decade for RSS.

Even by the laughable assertions of SkS that claim is daft!

I’m glad you think your own jokes are funny. You did however prompt me to more critically evaluate the article written by Kevin Cowtan (reviewed by Carl Mears of RSS), which includes a reference to Mears et al. (2011): http://images.remss.com/papers/rsspubs/Mears_JGR_2011_MSU_AMSU_Uncertainty.pdf
Table 2 of that paper reports 2-sigma trend uncertainty estimates for TLT as 0.044 K/decade, whereas Dr. Cowtan’s uncertainty estimate is nearly three times the peer-reviewed estimate. I have asked Dr. Cowtan to either correct my misunderstanding or justify his much larger uncertainty estimate: http://skepticalscience.com/surface_temperature_or_satellite_brightness.html#115622
While we’re waiting for his answer, I should like to ask you if you think 0.044 K/decade estimated uncertainty for RSS, to three decimal places no less, is any more or less defensible than 0.03 K/decade for HADCRUT4. When answering, consider what Dr. Spencer has to say about UAH TLT v6: http://www.drroyspencer.com/2015/04/version-6-0-of-the-uah-temperature-dataset-released-new-lt-trend-0-11-cdecade/
The new LT trend of +0.114 C/decade (1979-2014) is 0.026 C/decade lower than the previous trend of +0.140 C/decade, but about 0.010 C/decade of that difference is due to lesser sensitivity of the new LT weighting function to direct surface emission by the land surface, which surface thermometer data suggests is warming more rapidly than the deep troposphere. The remaining 0.016 C/decade difference between the old and new LT product trends is mostly due to the new diurnal drift adjustment procedure and is well within our previously stated range of uncertainty for this product’s trend calculation (+/- 0.040 C/decade).
One might reasonably expect uncertainty to improve to less than 0.040 K/decade in the new product. I’m not sure at what point very small numbers to three decimal precision becomes a laughable assertion, hopefully you will be able to clear that up for us.

…With his pretty little Crayons….We are so proud of his improvement !!

It’s almost like he doesn’t see that his graph produces nearly the same temperatures for both spreads…if you take the midline of the wider green spread, and the midline of the red spread, they are within 0.01 C of each other. Or did he mean to highlight that for us?

Marcus,

As always, Brandon is drawing pretty pictures from his cozy little insane asylum !!

The lunatics must be running it then, because when I turned on my computer today, I saw that some guy named Josh has apparently rebutted a video about satellite temperature reliability vs. surface station data by drawing a cartoon making fun of Michael Mann’s Hokey Schtick. Bizarre.
Nurse Ratched? May I have my meds now please? Stat?

JohnWho,

The main point of any SkS article is to deceive the weak-minded, gullible reader.

Ok, please inform this weak-minded, gullible reader which part of the SkS article is the deception, and why. Try to include something other than unsupported assertion and personal insults in your answer.

JohnWho,

Not in order of importance, but please notice along the right side of WUWT that the SkS site is deemed “Unreliable” partly because it has been noted as deleting comments that disprove what they want said, partly because of allegations of post editing where commenter’s post have been changed, and, I believe, mostly because the information provided is simply unreliable.

I didn’t ask you for someone else’s opinion of the entire SkS website, I asked for what part of this particular SkS article you think is the deception, and why: http://skepticalscience.com/surface_temperature_or_satellite_brightness.html

The SkS site, instead, provides what they deem “supporting” information to a non-established concept. Just that alone should make one, uh, skeptical of anything on that site.

I didn’t ask you of your opinion of the entire SkS website, about how well established any particular concept is or is not discussed.

Others with much more knowledge on the subjects discussed on SkS often de-bunk the SkS bunk.

I’m not asking what others think about the entire SkS website, I’m asking you what you think about a particular article.

Since you admit to being weak-minded and gullible, I would suggest you stay away from any site with dubious reliability.

I also asked you to try to include something in your response other than unsupported assertion and personal insults; you have essentially failed to do so and have completely failed to even directly answer my question as asked. I’m trying to understand how me pointing these things out is consistent with being weak-minded and gullible, but I’m probably just too stupid to figure that out.

Robert B,

After homogenising or before?

After.

If after, its not an estimate of uncertainty.

Homogenization is explicitly identified as one source of uncertainty in Morice et al. (2012): http://www.metoffice.gov.uk/hadobs/hadcrut4/HadCRUT4_accepted.pdf
In addition to the differences arising from data set construction methodologies, differences in computed climate diagnostics, such as regional average temperatures, can result from differing approaches to compensating for non-uniform observational coverage across the globe. The differences in temperature analyses resulting from the various approaches is referred to as “structural uncertainty”: the uncertainty in temperature analysis arising from the choice of methodology [Thorne et al., 2005]. It is because of this structural uncertainty that there is a requirement for multiple analyses of surface temperatures to be maintained so that the sensitivity of results to data set construction methodologies can be assessed. The requirement for any given analysis is to strive to both reduce uncertainty and to more completely describe possible uncertainty sources, propagating these uncertainties through the analysis methodology to characterize the resulting analysis uncertainty as fully as possible.
So, how certain can we be of the temperature evolution observed in a given observational analysis? A detailed measurement error and bias model was constructed for HadCRUT3 [Brohan et al., 2006]. This included descriptions of: land station homogenization uncertainty; bias related uncertainties arising from urbanization, sensor exposure and SST measurement methods; sampling errors arising from incomplete measurement sampling within grid–boxes; and uncertainties arising from limited global coverage.

BG-
“the net version changes are net COOLING to the tune of 0.10 K in the “pause” era, whilst the net version changes to GISTemp over the same interval are WARMING on the order of 0.03 K.”
LOLOL!! No Brandon. No.
Wrong. 1st-Anyone can see that the GISS data in your charts does not run through the year 2015 as does the UAH data. Why is that? Did you purposefully choose a chart that lied or was that an accident?
2nd- These are temperature anomalies, not actual warming or cooling…just deviations from the mean baseline (the Zero point)
3rd-The “cooling” shown in your chart shows a difference between -0.04C and -0.08C-a whopping interval of 0.04C NOT 0.10!! Check your math.
http://www.moyhu.org.s3.amazonaws.com/2015/12/uahadj.png
When graphed correctly you can see that the GISS data shows a 0.6C increase in trend and the UAH data shows a 0.5 C increase in trend. The difference between the UAH V 5.6 and V 6 is approx 0.1 C
Once again, you went to a lot of work to prove that you can’t even use honest, accurate charts.

Aphan,

1st-Anyone can see that the GISS data in your charts does not run through the year 2015 as does the UAH data. Why is that?

Here again is the plot by Nick Stokes http://moyhu.blogspot.com/2015/12/big-uah-adjustments.html :
http://www.moyhu.org.s3.amazonaws.com/2015/12/uahadj1.png
Read the caption: UAH 2015 versions diff vs. GISS 2015-2011 and 2015-2005
GISS 2011 stopped in … 2011. Can’t compare GISS 2015 to GISS 2011 if there are no data to compare.

2nd- These are temperature anomalies, not actual warming or cooling…just deviations from the mean baseline (the Zero point)

Umm, no, the plot shown above is the difference of anomalies between present and past versions of temperature anomaly data products.
Anyway, yes I know that plot doesn’t show actual heating or cooling. It does however show net heating or cooling ADJUSTMENTS between product versions.

3rd-The “cooling” shown in your chart shows a difference between -0.04C and -0.08C-a whopping interval of 0.04C NOT 0.10!! Check your math.

I did no math there, I just read values off the y-axis for each difference series.

When graphed correctly you can see that the GISS data shows a 0.6C increase in trend and the UAH data shows a 0.5 C increase in trend.

I’m not using that plot to support an argument about trend differences between data products, but rather the difference in monthly anomaly values from version to version between two data products.

Robert B,

Again, what is the point of the figure you titled “Meanwhile, back in reality” without the 0.16 error bars?

Read the caption. Carefully.
http://skepticalscience.com//pics/rss_ensemble_series.png
Figure 3: Spread in the satellite and surface temperature ensembles over time. Each line shows one possible temperature reconstruction from the ensemble (12 month moving average). All of the series have been aligned to a zero baseline for the 10 year period 1979-1988, so that the increasing spread after that period gives an indication of the variability in the trend.
Upon following the link provided to the entire article: http://skepticalscience.com/surface_temperature_or_satellite_brightness.html
… one will note that the next plot is:
http://skepticalscience.com//pics/rss_ensemble_box.png
Figure 4: Uncertainty in the temperature satellite and surface temperature trends, estimated from the RSS and HadCRUT4 ensembles. The boxes show the mean and interquartile range of the trends on 1979-2012 in each ensemble. Whiskers indicate the 95% interval (2.5%-97.5%). Crosses indicate outliers.
What? Aren’t those error bars?

Sure looks like they were ignored to me.

You may be right. Another possibility is that Dr. Cowtan didn’t realize that the HADCRUT4 ensemble doesn’t include all sources of uncertainty …
http://skepticalscience.com/surface_temperature_or_satellite_brightness.html#115558
Once he factored them in …
The original spread in the trends was about 0.007C/decade (1σ). Combining these gives a total spread of (0.0072+0.0022+0.0022)1/2, or about 0.0075 C/decade. That’s about a 7% increase in the ensemble spread due to the inclusion of changing coverage and uncorrelated/partially correlated uncertainties. That’s insufficient to change the conclusions.

Aphan,

Does anyone here even know what point he’s trying to make?

Main point was that the trend uncertainty for RSS is greater than HADCRUT4. Secondary point made in follow-on posts is that if the same skeptical questions about reliability of HADCRUT4 uncertainty are applied to RSS and UAH (A)MSU TLT anomaly products, it’s entirely clear that asking skeptical questions of both is entirely appropriate. Conclusion: When Judith Curry testifies in front of Congress and says, “I think we should look at the satellite data because they’re the best data we have” I’m thinking that such a claim isn’t supportable by what the data providers say about their own product in peer-reviewed primary literature.
And then I get an ever so subtle taste of irony over the name of that particular hearing: Data or Dogma?
Is that short and sweet enough for you to parse?

Aphan,

Ok. RSS and HADCRUT and GISS have never been exact matches, but they are amazingly similar all things considered.(in terms of tenths/hundredths of degrees)

Don’t forget UAH. I would not put as much stock in sat temperature anomalies as I do if RSS and UAH were not also remarkably similar, and if both of those also did not track well against radiosonde temperature anomaly products. The one fly in that ointment is the RATPAC-A radiosonde time series which shows greater warming over “The Pause” era than both satellite products to.

Based on the fact that the surface temperature stations are so out of whack that 90% of them aren’t cited properly, and their readings are in question, then yes, I’d agree with Curry that the satellite data is the “best” we’ve got all things considered.

I may have more to say on that when Watts et al (2015) makes it through peer-review and is published. One elephant in the room there is that USHCN (post-adjustments) tracks very well against USCRN since its inception, and USCRN was purpose-designed as a reference network to avoid the siting, calibration, time of observation, enclosure, UHI and other issues that are known to be problems in USHCN requiring homogenization.
All things considered? There are a lot of things to consider. Neither of us are going to be able to consder all things, ever, and especially not in the format of a blog comment section.

Seems like a lot of back and forthing for such a simply point really.

I thought my main point was abundantly clear and succinct from my very first post.

Aphan,

No one here, that I know of, has ever claimed that the satellite data, from ANYWHERE, should be taken as some kind of gospel truth.

I cannot recall anyone phrasing it exactly that way. Then again, I am not arguing that phrasing … I quoted Dr. Curry’s congressional testimony wherein she opines that it is the best data we have.

But almost everyone here is familiar with Anthony’s work with surface stations and know that the SSData SUCKS even more than the satellite data

Apples and oranges, Aphan. Two completely different methods of estimating temperature trends going on here. Go read some of the papers I’ve cited here, I think you’ll find that while the sources of error are different, the satellite teams use remarkably similar methods to identifying, quantifying and attempting to correct them as the surface data teams do.

so all you’re really doing here, besides poking richardscourtney until he has a stroke, is making points that everyone here pretty much knows and most likely agrees with.

You don’t speak for everybody here, and I respectfully disagree with your opinion about what others here think.

But heaven forbid that one thread on the internet should lack a lengthy, irrational discourse from you.

Heaven forbid you make a post that doesn’t make sweeping generalizations about other people.

Aphan,

There are only 4 satellites used to retrieve data from. Thus the levels of uncertainty are limited, and as you have repeatedly proven in this thread, the data results are amazingly close to each other with very small amounts of bias.

RSS lists 16 currently active: http://www.remss.com/measurements/upper-air-temperature
… though it is true that only 4 have been available all the way back to 1978. I wouldn’t expect that fewer instruments entails greater “limits” on uncertainty, but rather quite the opposite. I have “proven” nothing on this thread, and certainly not that satellite uncertainty is less than that of the surface record. I believe that I have shown compelling evidence that, by using the various data providers’ own uncertainty estimates, that HADCRUT4 has about 5 times less trend uncertainty from 1978-2012 than RSS TLT over the same interval.

The surface based stations number in the thousands and are wildly biased and uncertain. See here-http://www.surfacestations.org/

Been there many times, read it with much interest. I don’t recall seeing any quantified estimates of uncertainty, and no comparisons of same to any satellite temperature time series. Your qualifying terms “wildly biased and uncertain” in this are meaningless for any kind of reasonably objective comparison.

The surface station trends move in all kinds of different ways, and do not mimic each other closely at ALL.
7.9 % of surface stations show difference of 1 C or less.
21.5% of surface stations show differences of more than 1C
64.4 % of them show differences of more than 2 C
and 6.2% of them show differences of more than 5 C

Those are not trend differences, but rather estimated single-measurement uncertainties used as threshold as part of a station classification scheme. Visually inspecting a certain station and applying a classification to it on that basis does not necessarily mean that its measurement error corresponds to the assigned classification.

The satellite trends mimic each other very closely, no matter which “team” processes the data, and the differences in trend between them is LESS than 0.10C.

0.10C per what unit time? Paying attention to units is important here.
Trend 1979-2015 (K/decade):

UAH TLT 0.114
RSS TLT 0.123
mean    0.118
range   0.009

Data sources:
http://vortex.nsstc.uah.edu/data/msu/v6.0beta/tlt/uahncdc_lt_6.0beta4.txt
http://data.remss.com/msu/graphics/TLT/time_series/RSS_TS_channel_TLT_Global_Land_And_Sea_v03_3.txt

The surface temperature data does not even match in trends, let alone mimic each other, and in the case of the land based surface temps, 92% of those stations show a difference of more than 1C!!! Or in other words- 92% of the land surface records show a difference in trend that is MORE THAN 0.10C.

Apples and oranges again. As I explain above, measurement error of a single station is not trend uncertainty.
Trend 1979-2015 (K/decade):

NCDC      0.155
BEST [1]  0.156
GISTEMP   0.165
HADCRUT4  0.165
BEST [2]  0.172
mean      0.162
range     0.023

[1] Global Average Temperature Anomaly with Sea Ice Temperature Inferred from Water Temperatures
[2] Global Average Temperature Anomaly with Sea Ice Temperature Inferred from Air Temperatures
Data sources:
https://www.ncdc.noaa.gov/cag/time-series/global/globe/land_ocean/p12/12/1880-2015.csv
http://berkeleyearth.lbl.gov/auto/Global/Land_and_Ocean_complete.txt
http://data.giss.nasa.gov/gistemp/tabledata_v3/GLB.Ts+dSST.txt
http://www.metoffice.gov.uk/hadobs/hadcrut4/data/current/time_series/HadCRUT.4.4.0.0.monthly_ns_avg.txt
Going back to your argument: The satellite trends mimic each other very closely, no matter which “team” processes the data, and the differences in trend between them is LESS than 0.10C.
Using that logic, I could say that GISTEMP and HADCRUT4 mimic each other very closely even though they’re produced by different teams using different source data and processing methods, and also note that they only differ from the mean trend of all surface series by 0.003 K/decade whereas UAH TLT and RSS TLT each differ from their respective mean by +/-0.005 K/decade. Therefore, by the same logic you apply in your argument, GISTEMP and HADCRUT4 are the “best” choices.
But I can’t really make that argument on that basis because comparing the trends to each other does not necessarily indicate the structural uncertainty of the temperature reconstructions themselves. This argument is also reliant on which surface temperature reconstructions we include.
The better way is to go to the literature and start by reading what the providers themselves have to say about the estimated uncertainties of their own products.