RUTI USA part 1




1) Basic results for US contiguous 48 states
2) TOBS vs. COTS,  Hadcrut adjustments to USA temperature data.
3) US temperatures, a geographic perspective


1) Basic results for US contiguous 48 states

Fig1. CLICK to enlarge. USA, Contiguous 48 states, locations of 830 temperature stations used in the present analysis. Most urban stations are not included, and as always in RUTI, Urban location is determined using Google maps. In addition, temperature series used generally cover both the warm peak 1930-40 and several years after 1990.
Red stars: Temperature level 1998-2010 warmer than 1930-40.
Blue stars: Temperature level 1998-2010 colder than 1930-30.

Given that these data are the best temperature data available in the world, its a good oportunity to study the influence of geography on temperature trends. The present writing will also show Hadcrut temperature trends compared to unadjusted GHCN .
“COTS” is a reason for warmer trends in Hadcrut data (as are TOBS adjustments) and this “COTS” will be shown and explained because it appears to be a significant factor causing the  warmer trend for hadcrut data.
Baseline for temperature trends in this article is 1961-1990.

First, lets take a look at the overall RUTI result based on 830 mostly long temperature series. 

Fig2. CLICK to enlarge. For each 5x5 grid areas of similar temperature trend where identified and trends weighted according to the size of the areas represented. The numbers on the graphs are average of the 12 years 1998-2008 compared to 1930-40.

Finally 5x5 grids where summed up, weighted according to area of land. Result:

Fig3. USA temperature trend derived from 830 stations.

Fig 4: Since a RUTI ground rule is to seek unadjusted data when reasonable, lets compare with the older, less adjusted temperature trend for USE made by NASAs Jim Hansen, 1999.

The similarity RUTI vs Hansen 1999 is obvious.

Fig5. This kind of similarity (RUTI vs. Hansen) is hardly coincidence and perhaps methods used to some degree are similar. The stunning similarity may allow considering RUTI USA data as a “Hansen 99” extended with one more decade.

Table: 1 and 2

2) TOBS vs. COTS,  Hadcrut adjustments to USA temperature data.

Two more sets of data was collected and calculated like the unadjusted GHCN data above.

1) Hadcrut station data,
and then
2) The Unadjusted GHCN data for stations chosen by Hadcrut.

Since Hadcrut stations normally are GHCN stations too, I found it interesting to compare those stations picked by Hadcrut before their adjustments (Unadjusted GHCN) and then after adjustments.


Fig6. The RUTI temperature trend (closely resembling the Hansen 99 dataset) has colder trend than the Hadcrut stations, which is of course no surprise.

You can see the differences even easier by setting year 1900 to approximately zero (Base period now 1895-1900):


Fig7. In general we can see that the 2 series from stations chosen by Hadcrut – red and black – are rather similar, while the RUTI (or Hansen 99) appears to be significantly colder. Still, this is no news, no surprise.
Or  is it?


Fig8. By comparing slopes from the trend lines fig6, we can estimate that:

 Most of the extra heat trend from Hadcrut stations seems to originate from “COTS” – “Choice Of Temperature Stations” and only a third from adjustments like ”TOBS”.

The black trend and the blue trend are from the very same dataset, unadjusted GHCN. The reason that makes the black trend much warmer than the blue (RUTI/Hansen) is that the temperature stations chosen by Hadcrut has significantly more heat than the bulk of Unadjusted GHCN data.

So first, Hadcrut makes a    Choice Of Temperature Stations   resulting in significantly warmer temperature than the bulk of unadjusted GHCN data, and THEN they believe that these should be even warmer trended, and adds their warm adjustments, TOBS etc.

How come the stations that Hadcrut has picked from GHCN have a much warmer trend even before Hadcrut starts adjusting?


Fig9. List of all Hadcrut stations for contiguous USA. Red numbers are population in thousands. Normally stations with more than 50.000 inhabitant are classified as Urban, stations with 10-50.000 thousand inhabitants are classified as “surburban”, and below 10.000 inhabitants we talk about rural stations.
Most of the stations chosen by Hadcrut stations are not included in RUTI analysis due to obvious urban locations.

The unusual and overwhelming abundance of long temperature series for USA from GHCN does not make it easier to understand this urban approach by Hadcrut.

In addition, several Hadcrut stations from rural or small urban areas are located on the coast, and thus should only represent a small coastal area.

Summa: The much warmer USA trends found in Hadcrut data is not mostly due to warm-adjustments that some may find relevant, no, the bulk of the added warming seems simply to originate from the choice of temperature stations used by Hadcrut, “COTS”.

Fig10. Finally a compare Hansen 2008 with the above results from Hadcrut stations. It seems that the Hadcrut-stations resemble Hansens temperature data for USA 2008.

3) US temperatures: A Geographic perspective.
In RUTI: Coastal temperatures the coastal temperature trends and some temperature trends from coastal facing elevated locations are shown to be simlar, and thus likely to be related due to marine origin:

Fig 11. The red coastal zone and the elevated brown zone facing the Marine air winds shows similar temperature trends (typically warmer temperature trend than non coastal areas).

Fig12. One example,  Left: For NE USA we saw that the “blue” non-coastal areas have significantly different temperature trends in comparison with the red coastal areas. Right: At the same time, the elevated “brown areas” had mostly same trends as the red coastal stations, even though the brown areas are much longer away from the coast line.
Stations located at higher elevation, but not located on the ocean side of the hills, share temperature trends with the other non-coastal areas.

This means, that in more fragmented mountain areas with many peaks, valleys and geographical nuances, you can have areas with a complex mix of warmer and colder trends often quite close to each other.

Example of geographical influence on temperature trends: SE USA:


Fig 13. Here we have just used the blue and red colours as described under fig1, but the geographical impact on trend is clearly reflected: Not only coastal stations deviate from the bulk of stations, blue cold trended. In the center of the image, we see the Appalachian Highlands. In several cases this highland is accompanied by warm trended stations, and more, these are all located on the east side of the Appalachian Highlands, that is, facing coastal winds.

Fig14. As shown in numerous examples in RUTI: Coastal temperature stations, Hill sides facing ocean winds normally have temperature trends that deviates from their surroundings, here a typical situation from Himalaya, a weather prognosis showing that these phenomena obviously are very well known. Notice also, that even several thousands of kilometres away from the coast, the mountains facing the ocean acts differently than its surroundings.

The general impression from fig 1 is:
Whenever we have an area of land not on or near mountains peaks, nor on or near coasts or colder water streams, it seams that we have the colder temperature trends, the “blue stars”.
That is, any piece of land not influenced by water (or ice) changes seems to have colder temperature trend from the 1930´ies until today.

“The Canadian boarder effect”: All along the Canadian boarder it seems that warmer trends are much more abundant than in other locations. This is appears to be due to presence of mountains and mostly water near the boarder, a local phenomena.
However, In RUTI: Coastal station we could see a clear connection between marine (island) temperature trends, and then the coastal trends. Why do the Lakes of USAseem to have an effect?
My guess is, that these lakes in warmer periods releases more moist to the air, and thus dims the air, and avoids some of the strong warming taking place even near by. A “Water effect” ? This is just a guess.

Fig15. “Water effect”: This water effect may also explain some local warm trends like in the centre of fig9: Leach Dam, Pokedama Dam, Sandy L Dam and Pine River Dam? Obviously the exact location of a thermometer may have influence on such a water effect.

Fig 16. When we look at a big chunk of the central USA - with not too many disturbances - it becomes clear, that the rising Rocky Mountains to the left have a massive influence on temperature trends. Just like it was shown for Himalaya, a “mountain coast line” on the Atlantic ocean facing side of the Rockies is accompanied by higher frequency of warm trended stations than the lower areas. Again, see fig 5 for the mechanism.

However, the Rocky Mountains are more complex than this:

Fig17. The Rocky Mountains also face the Pacific ocean trends from west, obviously, and “worse”: We have several localities within the Rocky Mountains with even higher elevation, that then creates new higher “Mountain coast lines”, marked purple. In addition, we can have river stations in valleys close to warm trended mountains areas which also shows warm trend etc. etc.

All in all, it seems that USA shows that we can have cold trended and warm trended areas side by side with just short distances, not only near the coasts.

Fig18. CLICK to enlarge.
I took the average of cold trended stations and average of warm trended stations for each area to show to what degree these two trends can persist in each others near presence.
For the 5x5 grids located in the Rocky Mountains, some shows limited differences while others shows that differences in cold and warm trended areas can exist very well indeed within the Rocky Mountains. As shown in RUTI: Coastal temperature stations, the differences near cost lines can be massive.

!! This has the consequence, that one should take very much care using one station as target when adjusting another , not only near coasts, but also in areas with hills and mountains. !!

EX, the 45-50N / 110-115 grid, data shown is 5 yr avg:

Fig19. I find amazing that local trends in some areas can survive so strongly.
The text on the graph: “Colder 10 st 50%” means, that there are 10 cold trended temperature stations, and they represent an area of approx 50% , and thus, the cold trend should be weighted 50% for the overall USA numbers, no more.


Fig 20. One feature from the mosaic on fig 12 is the remarkable warming of the Western Rocky Mountains, 1880-1905, supported by many of these 5x5 grid average graphs. At the same time, in the Central USA we had a very cold period 1880-1905.

The impact on temperature trends for elevated temperature stations as mentioned has a strong conection to the marine winds, see fig 11. However, for example mountains with ice and glaciers also have other mechanisms that can affect temperature trends. And so does the valley river stations just downstream from such mountains.

The effect of mountains has been shown several times in the RUTI project and appears to be a global effect just like the coastal effect on temperature stations.

Lüdecke et al has made a general world side test of the elevation effect on temperatures:
Fig 21. From fig 7 of Lüdecke et al 2011
Lüdecke shows the effect of station elevation on heat trend 1906-2005. Lüdecke et al are thus confirming the RUTI results. Using the elevation alone as variable however, includes many stations on elevated sites that are not facing ocean winds, because they are located “behind” other mountains that face ocean winds. More flat elevated valley areas will mostly show trends a little similar to the lower non-coastal trends, but in the Lúdecke graphic all elevated sites are counted in.
This do give the important information that we have a world wide issue with regard to elevation, but in fact, the problem may be a little underestimated here.

Part 2 ! Coming Up.
Here focus will be on the 5x5 grid cases and the adjustments done …


Powered by Website Baker