Guest post by John Goetz
As noted in the previous post, GISS has released their monthly global temperature summary for June, 2009. This month’s whopping anomaly of 0.63C is once again much higher than that of RSS, UAH, and even NOAA, which is the source of the GISS temperature data. Not only is the anomaly higher than the other metrics, but it is trending in the opposite direction.
Temperature data from 1079 stations worldwide contributed to the analysis, 134 of them being located in the 50 US states. Data from essentially the same few stations have been used for the past twenty-four months. Many, many hundreds of stations that have historically been included in the record and still collect data today continue to be ignored by GISS in global temperature calculations.
Once again, the bulk of temperatures comprising the present-day worldwide GISS average come from airports – in this case 554 airports, according to the NOAA metadata from the V2 station inventory. In the US, the ratio of airports to total stations continues to run very high, with 121 out of the 134 reporting stations being located at airports.
Why worry about airports? Aside from recent posts on this site documenting problems with airport ASOS equipment in the US, WUWT has also documented a number of equipment siting problems, notably the typical close proximity of the equipment to a tarmac heat sink. Airports can introduce a mini-UHI effect where one would otherwise not be found.
The NOAA metadata is not entirely accurate, and several stations located at airports are not noted as such. Some examples include Londrina and Brasilia in Brazil, Ely / Yelland in Nevada, and Broome in Austrailia. Those stations were easy to find because they had “airport” (or some variant) in the station name. A check of coordinates using Google Earth confirmed the airport locations.
Let’s examine the metadata a little further, shall we?
NOAA says that 345 of the stations it passes on to GISS are rural and presumably free of UHI influence. Fifteen of those stations are located in the US. However, only 201 of those rural stations are not located at an airport, and therefore presumably free of UHI effects (including tarmac heat sinks). In the US, only one of the fifteen stations is listed as both rural, and not located at an airport: Ely / Yelland in Nevada.
Doh!!! As noted above, that station is located at an airport – confirmed not just by Google Earth, but also by NOAA’s NCDC website as well! This means that all of the US temperatures – including those for Alaska and Hawaii – were collected from either an airport (the bulk of the data) or an urban location.
As for the rest of the world, some of the stations listed as being rural and not at an airport have metadata indicating they are located in an area of “dim” or “bright” lights. Filtering those out, we find a total of 128 stations that are rural, not at an airport, and “dark”.
Why are “dark” stations important? Recall that GISS uses dark stations to adjust for UHI in the urban stations. With only 128 dark stations available, none being in the US, it would seem this is an impossible task.
Fortunately, GISS adjustment rules allow old data to be used in adjusting new data. The older “non-reporting” rural weather stations continue to adjust reporting urban stations, even though the most recent two years of overlap is missing.
Thankfully, the algorithms are robust enough to calculate adjustments to the 100th of a degree even when data is missing.
I think we all know that July is heating up quickly, but in what way, seeing the continued stream of record low highs in the US according to NOAA is suggesting it’s not neccesarily because of high temps. in the US.
And for those reading the scare stories by the media, look, I can make alarmist statements too and top them, like
[alarmist]-emit all you want, by 2011 it will be so hot there will be no more life on Earth, enjoy your last two years, the beginning of thermal armeggedon will begin in less than a month and it’s too late to stop it[/alarmist]
.63. Hmmm. That’s gotta be near the record for the month, right?
I’m betting the shoe pounding is also at a record high over at RC…..
“”” David (11:32:54) :
“Reply: ……Also note that the majority of stations actually used did not report a June 2009 temperature. – John”
Isn’t that the problem? If there is a valid algorithm that can predict temperatures within an X mile radius, then I am sure that we would understand weather a lot better than we do. As it is, there can be a huge variance between a distance of ten miles. I can think of one such example from here in Ohio back in June where it was at least 5ºF warmer downtown than it was at my house. Cloud cover and precipitation were the main reasons for this, but how does an algorithm adjust for that, and why would you want it to? “””
You are talking about a simple failure to observe the standard laws of sampled data systems; notably the Nyquist sampling Theorem:
Any band limited continuous function can be reconstructed accurately from samples of that function; if and only if the continuous function is sampled at a rate at least twice the frquency of the band limit.
If the continuous function is sampled at a rate of 2B, but contains an out of band signal component at a frequency B+b, the reconstructed signal will contain an aliassing noise signal at a spurious frequency of B-b; and since that signal is now inside the band limit; no filtering can remove it; it is a permanent corruption of the recovered function.
And in the case where the function contains a signal component at a frequency of 2B or higher, then the aliassed signal is at B-B or zero frequency (or less), and that zero frequency component is simply the average value of the function; which now becomes completely unrecoverable.
The GISS temp data; being based on a min-max thermometer reading gives no more than two samples in a 24 hour period, which is the bare minimum for a 24 hour period signal component (the diurnal temperature cycle) but it is only sufficient, if that temperature cycle data is time symmetric; like a pure sinusoid. Since the daily average of min-max is reported; then the factor of two undersampling is already present, so the average cannot be recovered; and any harmonic variation which would make the signal time assymmetric, or any higher frequency cycling such as that due to clouds moving by, makes even the temporal data corrupt. But when you look at the spatial sampling component (the 1200 km BS or even the 250 km), it is clear that the function is undersampled by orders of magnitude; so it is inherently impossible to accurately compute the average of the recovered but aliass corrupted signal.
Sorry GISStemp; your output is simply the result of applying your algorithm to your data. It in no way relates to anything to do with Planet Earth; since yourt signal sampling regimen is insufficeint to properly recover the original continuous function; you are simply publishing GISStemp; which is not any mean global temperature or anything else of physical importance to climate.
“Frank Lansner (06:10:04) :
OT:
BBC blog refelcting on frontpage, Spectator, “Global warming is a myth”:
http://www.bbc.co.uk/blogs/climatechange/2009/07/anyone_seen_the_front_page.html
Havent read it yet..”
Interesing PoV’s at that blog, some in need of serious education however.
Broome Airport Western Australia. For all intents and purposes the Airport is practically in the middle of the town (a great planning coup). That could be a case of a double whammy. UHI plus AHI.
Dan (11:11:40) :
New York cental park and Newark Int. Airport compared. (together with giss adjustment)
http://img379.imageshack.us/img379/5074/newarknycprawgiss.jpg
Les Johnson (14:18:53) & alex verlinden (12:40:35):
I’ll volunteer to buy some. Anyone who buys one, I’ll match it, up to 5. That would make a total of twenty, a good start.
The dataloggers would need to be calibrated at a certified lab, with standards traceable to N.I.S.T., or the inevitable critics will question the accuracy. That would add a little to the cost. But it’s basically a good idea. Real time info, archived online, would make for a few sweaty brows at GISS.
I used to live about 1 and a half miles outside of JFK, right under the flight path. In winter, any snow or ice that accumulated would disappear with a day or two. My parents house was about 10 miles from JKF and they would have ice for weeks. I enjoyed the UHI effect the airport gave me. Meant very little shovelling and I almost never needed to use salt on my walkways.
That being said, I thought the AGW hypothesis’ signature would appear in the tropsphere. If that’s the case, why are we concerned with land temps other than how they relate to the troposphere? What am I missing?
“The GISS temp data; being based on a min-max thermometer reading gives no more than two samples in a 24 hour period, which is the bare minimum for a 24 hour period signal component (the diurnal temperature cycle) but it is only sufficient, if that temperature cycle data is time symmetric;”
Has anyone done 24 Hrs logging temps every 5 mins so as to compare an integrated daily temp against a min/max average?
Seems to me a balmy night, blistering day and a CuNim hailstorm round the sensor for 30 mins then heating into a balmy evening again, gives a very short cold spike which is not representative.
Smokey: A good start.
Now, we just need to find someone to ram rod this. Someone with meteorlogical experience; and preferably one who is familar with Surface Meterological Stations…..
30 or so years ago, I took off from Chicago OHare airport in a snow storm. Takeoffs were delayed as they cleared snow off the runway. It was snowing hard. My plane sat waiting to takeoff for at least an hour.
I watched truck after truck go to a big hole in the concrete and dump its load of snow into the hole. Then I saw blue flames coming out of the hole. They were dumping snow onto a gigantic gas burner.
It must have been putting out prodigous amounts of heat to melt the volume of snow being dumped in. I hope they weren’t measuring temperatures anywhere near it.
“They were dumping snow onto a gigantic gas burner.”
Airport heavy-duty industrial version of the home barbeque.
An insightful post, as usual, by John Goetz.
Airport records are subject to a double whammy of distortion: a) urban encroachment and b) growing infrastructure and traffic that generates heat locally. Cities tend to grow out toward the airport, rather than away fom it. What may have been cool pastures just a few decades ago turns into a town surrounding the airport. What has troubled me for some time is the steady progression towards airport sites in the GISS global compilations and the increasing failure to update truly rural stations. While everbody is pointing fingers at the SST component this June, the land-only anomaly is even higher. I would suggest a study that determines the greatly diminishing proportion of truly rural records actually employed by GISS during the last few decades.
History of Broome Airport
http://www.broomeair.com.au/corporate/history/
For a reasonably uncorrupted site I would recommend Meekatharra.The station is ~800mtres from the Airport infrastructure and 400mt from the end of the strip.
Although the station moved to the airport in the 1950’s , at that stage there was no tarmac even on the major highway through town.
http://maps.google.com.au/maps?hl=en&q=meekatharra%20airport&um=1&ie=UTF-8&sa=N&tab=wl
http://www.bom.gov.au/weather/wa/wa-observations-map.shtml
“George E. Smith (17:13:17) ”
A couple of additional points.
To precisely reconstruct the analog (continuous-time) temperature record at a point in space, the sampled temperature values must be perfect — i.e., they cannot be quantized to a fixed set of values — like the nearest tenth of a degree centigrade.
Since it is extremely unlikely that the time intervals between “minimum/maximum and maximum/minimum temperature states” are the same, the sampling times are not uniform — which means that even if the highest frequency component in the analog temperature record at a fixed location were less than one cycle per day (which itself is next to impossible), the analog temperature record can be precisely reconstructed only if the times of the temperature samples are perfectly known.
In addition to being a function of time, real-world temperature records are also functions of position (latitude/longitude/altitude). To reconstruct over a volume of space a precise instantaneous spatial analog temperature record from a set of spatially diverse temperature samples taken at a single instant in time, (a) the sampling locations must be precisely known, (b) the analog temperature variations must be “bandlimited” in a three-dimensional spatial sense, and (c) the “sampling spatial locations” must obey a spatial three-dimensional sampling rule similar to the Nyquist sampling rate for real time-sampled-data.
If the objective is to precisely reconstruct a four-dimensional (three spatial dimensions and one time dimension) analog temperature record (i.e., a temperature record that is continuous in both space and time) from a sampled temperature record (sampled with respect to both space and time), the “sampling rules” will be both complex and a function of the four-dimensional “frequency” content of the actual analog temperatures.
Bottom line, I agree with you. Anyone who claims the existing temperature records provide anything but a coarse and ill-defined estimate of “average” temperature is simply wrong.
For a reasonably uncorrupted site I would recommend Meekatharra.The station is ~800mtres from the Airport infrastructure and 400mt from the end of the strip.
Thats where the metadata says it is, but there is no sign of it on Google Earth. And why would anyone put a station so far away in the scrub. It looks like the meta data is wrong.
http://www.bom.gov.au/climate/cdo/metadata/pdf/metadata007045.pdf
Warning: non-scientist alert!
Several lifetimes ago, I spent some time in the air force, mainly in hot and uncomfortable places (and in the tropics too). At that time, the air force always mounted the airfield weather station at the top of a wooden tower, about twenty feet above ground level – presumably to remove bias caused by ground temperature.
All of the pictures I have seen recently that show civilian weather stations show them close to the ground.
Imah sittin here wonderin whay th’ar be a diff’rnce?
“Thats where the metadata says it is, but there is no sign of it on Google Earth. And why would anyone put a station so far away in the scrub. It looks like the meta data is wrong”
http://maps.google.com.au/maps?hl=en&q=meekatharra%20airport&um=1&ie=UTF-8&sa=N&tab=wl
If you zoom in the station is just around the corner where the Airport road leaves the Murchison downs road.
It is about 4-5km from town.
Oh please, how dare you think that there is a heat Urban Island effect!?!?!? Remember that is why we have algorithms to account for such silly non-sense. I mean seriously a twelve degree difference is nothing hard to account for lets see that really just means we have to raise the surrounding area by 12 degrees! See Problem solved. Sheesh, come on people this is science, don’t question our methods there is a consensus after all.
>>>Broome Airport Western Australia. For all intents and
>>>purposes the Airport is practically in the middle of the town
>>>(a great planning coup). That could be a case of a double
>>>whammy. UHI plus AHI.
And there is also a b***** great sea breeze there.
Broom’s temperature can go up to the high thirties in the early morning, and then at 8 or 9am there will be a sudden, great rush of air and the temperature will go down by 10 or more degrees (centigrade). (this change can happen in less than a minute)
How can you determine the temperature of such a variable place? Surely, you should take a reading further inland, away from the sea breeze.
Flanagan, (07:11:54)
Frank Lansner, (16:33:47)
“Here is the difference between GISS and RSS since 2002:
http://www.klimadebat.dk/forum/vedhaeftninger/gisshotterthanrss.gif
I dare say there is a trend (!)”
I made some graphs (not shown here, i regret) and i did the following findings:
1. From 1979 to June 2009, the average difference between (RSS+UAH)/2 and GISS anomalies is 0.248 C (GISS 0.248 C more than the average between RSS and UAH). This difference can be explained (different base).
2. From 1979 to June 2009: The trend of the difference between (RSS+UAH)/2 and GISS is y = 0.0002x. This means that in this period, GISS showed 0.07 C more warming than the average between UAH and RSS.
3. Since 2002, this trend is rising. It amounts to y = 0.0016x. So GISS gives 0.144 C more warming during this 90 months.
4. Since 2008, we are witnessing large fluctuations in the differences (0.145 C (February 2009) to 0.592 C (June 2009).
In 2008, there was a similar discussion on this site as we have now. Taking into account the large differences since 2008, i think we have to wait the end of 2009 to draw conclusions. One point is clear to me: the more the difference between GISS and RSS/UAH is rising, the less the GISS figures become credible. How long will this trend go on like this?
And there is also a b***** great sea breeze there.
That’s true of much of coastal Australia.
In Perth we call it the Freemantle Doctor.
In Sydney, they call them Southerly Busters. Confusing cold fronts from the south with sea breezes in summer.
ralph ellis (23:44:26) :
>>>Broome Airport Western Australia. For all intents and
>>>purposes the Airport is practically in the middle of the town
>>>(a great planning coup). That could be a case of a double
>>>whammy. UHI plus AHI.
And there is also a b***** great sea breeze there.
In Perth, at The WACA, they call the sea breeze ‘The Freemantle Doctor’. It has been very welcome to many an England team toiling in many a long day in the field in the Boxing Day test match.
OT:
Unisys SST, change of colours..
Now anomaly of + 0,1K is indicated with DARK RED..
http://weather.unisys.com/surface/sst_anom.html
and -0,1 K is indicated by yellow.
To get a bluish color you have to come below -2 K..