I’ve moved to WordPress: http://bobtisdale.wordpress.com/

Friday, June 26, 2009

Part 2 of Comparison of GISTEMP and UAH MSU TLT Anomalies

I’ve moved to WordPress.  This post can now be found at Part 2 of Comparison of GISTEMP and UAH MSU TLT Anomalies
###############
Or The Comparison of GISTEMP Land Surface Temperature Anomalies and the UAH MSU TLT Anomalies for the Same Land Surface Areas

INTRODUCTION

In the first part of this post, Part 1 of Comparison of GISTEMP and UAH MSU TLT Anomalies, I illustrated that global GISTEMP had a higher linear trend than UAH MSU TLT--nothing earth-shattering there. But that post also divided the globe into 8 subsets and compared the linear trends for those smaller areas of the globe. In this post, we’ll look at Land Surface Temperature and TLT anomalies for each of the continents to further illustrate any differences between the GISTEMP and UAH MSU TLT data.

AREAS FOR THE COMPARISONS

The KNMI Climate Explorer website provides access to GISS Land Surface Temperature data that is separate from SST data. The UAH MSU TLT data available through KNMI, however, is not separated into land and ocean subsets. And due to the shape of the continents, there is no way to gather the TLT anomalies over the complete continental land masses without also getting TLT anomalies over portions of the oceans. Therefore, for this comparison of Land Surface Temperature anomalies and the TLT anomalies over those same land areas, I subdivided the data into smaller areas of the continents to minimize any influence from ocean data.

Figure 1 illustrates the global grids used in the comparisons of GISTEMP Land Surface Temperature and UAH MSU TLT anomalies. The coordinates are listed on the graphs that follow. Also, an area of Siberia seems to have had elevated surface temperatures in recent years (though it did not make an appearance in the most current map [May 2009] of Surface Temperature Anomalies, Figure 1). I’ve attempted to capture that “Siberian Hotspot” in the area enclosed in purple.

http://i41.tinypic.com/i1k28l.png
Figure 1

CONTINENTAL SUBSECTION COMPARISONS

Figure 2 illustrates the Australian GISTEMP Land Surface Temperatures and UAH MSU TLT anomalies for the same land mass area. The GISTEMP linear trend is negative (-0.011 deg C/decade), while the UAH MSU TLT linear trend is positive (0.11 deg C/decade).
http://i40.tinypic.com/20h7f3t.png
Figure 2
########

For the Central North American data, Figure 3, the GISTEMP linear trend of 0.201 deg C/decade is also less than the UAH MSU TLT linear trend of 0.224 deg C/decade.
http://i40.tinypic.com/nget8k.png
Figure 3
########

The Northwestern North America datasets, Figure 4, include most of Western Canada. The UAH MSU TLT linear trend (0.256 deg C/decade) is greater than the GISTEMP linear trend (0.169 deg C/decade).
http://i43.tinypic.com/f44ljb.png
Figure 4
########

It is unfortunate that the UAH MSU TLT data is not separated into ocean and land data on the KNMI Climate Explorer website. I did look for other websites where I could download the separate land and ocean TLT anomalies, but if one exists, it eluded me. The comparison of European land surface temperature and TLT anomalies required that I whittle down of the area covered to minimize any influence of ocean data. Refer again to the map in Figure 1.

The “Eurasian Strip” data, Figure 5, should at least capture the “core” temperatures for Europe. The 0.475 deg C/decade linear trend for GISTEMP is higher than the UAH MSU TLT linear trend of 0.426 deg C/decade.
http://i40.tinypic.com/k6ija.png
Figure 5
########

Travelling east, the shape of Asia allows a comparison of a much larger land surface area. The GISTEMP linear trend (0.438 deg C/decade) in the Asian comparison, Figure 6, is significantly higher than the UAH MSU TLT linear trend (0.237 deg C/decade).
http://i42.tinypic.com/10ylzjc.png
Figure 6
########

For the “Siberian Hotspot,” Figure 7, the difference in the trends is less that the difference for the Asian datasets with the larger surface area. The GISTEMP linear trend for the "Siberian Hotspot" is 0.451 deg C/decade, while the UAH MSU TLT linear trend is 0.317 deg C/decade.
http://i43.tinypic.com/mw38mq.png
Figure 7
########

Figure 8 illustrates the significant difference in the linear trends for the South American data. The UAH MSU TLT linear trend (0.097 deg C/decade) is much less than the GISTEMP linear trend (0.237 deg C/decade). The GISTEMP linear trend is more than twice that of the UAH MSU TLT data.
http://i41.tinypic.com/16idyc7.png
Figure 8
########

The difference between the GISTEMP and UAH MSU TLT data for Northern Africa, Figure 9, is also substantial. The linear trend for the UAH MSU TLT data is 0.094 deg C/decade, but the GISTEMP linear trend is more than 3 times greater at 0.306 deg C/decade.
http://i43.tinypic.com/if1oh5.png
Figure 9
########

As illustrated in Figure 10, the UAH MSU TLT linear trend for Southern and Central Africa is a relatively low, only 0.031 deg C/decade. The GISTEMP linear trend, on the other hand, is 0.276 deg C/decade.
http://i40.tinypic.com/1hb5sm.png
Figure 10
########

For the last comparison, I used the “SoPol” land data from the UAH MSU TLT webpage:
http://vortex.nsstc.uah.edu/data/msu/t2lt/uahncdc.lt

Similar to the difference shown in the comparison of the land plus ocean data for the Antarctic, the GISTEMP Antarctic Land Surface Temperature shows a positive trend (0.055 deg C/decade) while the linear trend for the UAH MSU TLT data is negative (-0.113 deg C/decade). Refer to Part 1 of Comparison of GISTEMP and UAH MSU TLT Anomalies for a further discussion of the differences in the Antarctic data. It’s primarily a presentation of the unusual timing of the mid-1990s rise in the GISTEMP data. That rise and fall is not a result of the 1997/98 El Nino. It precedes the 1997/98 El Nino.
http://i40.tinypic.com/707vuo.png
Figure 11
########
SOURCE

The GISTEMP Land Surface Temperature and UAH MSU TLT data (with the exception of the Antarctic Land TLT data) are available through the KNMI Climate Explorer website:
http://climexp.knmi.nl/selectfield_obs.cgi?someone@somewhere

7 comments:

Anonymous said...

So how do UAH and RSS get their data for USA 48? Your method misses out a few states!!

I compared your area with UAH USA 48 data. They are a bit different but not significantly so and the slopes are:

using 0.0019x -0.3101 KMNI data
using 0.0019x -0.1779 UAH data

this was smoothed data using Excel's moving average using interval 12

UAH USA 48 unsmoothed give a equation of:

0.0021-0.2562 for Dec 1978 to May 2009.

Hmmmm - so your slopes are /yr?

.0019 x12 = 0.0228

Since I duplicated your results this time the Lat/Long on your charts are the correct way to enter them in KMNI? Were you consistent on all charts?

Steve H :]

did I send this X2?

Bob Tisdale said...

Steve H: UAH and RSS have the data separated for oceans and land. KNMI doesn't.

Nice to see the linear trend of the UAH USA 48 agreed with abridged KNMI dataset. I didn't think to check it.

The first values in the trend equation kicked out by EXCEL (the slope) are annual values. So the decadal trend values would be 10 times that. (My error in the last post.) No multiplying by 12.

Using EXCEL to do the smoothing "justifies" (Can't think of a more approprite word) the smoothed curve to the right-hand side of the chart. It shifts the timing and makes it difficult to identify timing of the variations, especially when you use 121-month filters on the AMO or ENSO.

You asked, "Since I duplicated your results this time the Lat/Long on your charts are the correct way to enter them in KMNI? Were you consistent on all charts?"

For latitudes, "n"S are negative values and "n"N are positive. For longitudes, "n"E are positive and "n"W are negative.

Am I consistent? Always. That's the only way the similar site for SST data, NOMADS, will work.

And no, you didn't send it twice.

Regards.

Anonymous said...

Bob

I want to pass something by you which has interested me for some time. Your 2 recent posts were very timely for me because I wondered if taking particular sections of the world was possible and if it was might it change my perception of Global Average Temperatures.

I had noticed that the UAH (and RSS)USA 48 had shown quite a decrease in temperature since the El Nino of 1997/8. Therefore, I split the chart into two time periods - Jan 1979 to Oct 1997 and Nov 1998 to present. This essentially takes out the El Nino "bump".

For the US this results in a +2.52 Deg C/Century (DCC) Linear Trend (LT) Pre El Nino and a -3.24 DCC LT Post El Nino for UAH.
The figures are 1.68 and -4.2 for RSS.

The LT for Jan 1979 to Present is 2.52 DCC (UAH) and 2.28 DCC (RSS)

I have posted this elsewhere and have been met with silence or disinterest (but no scorn).

I just did this for Canada (my country) using 128W/55W Long; 49N/70N Lat. I was motivated by looking at the Arctic result of your first post.

The results were:

-0.6 DCC Pre El Nino and -3.12 DCC Post El Nino.

The Jan 1979 to Present result was + 3.72 DCC !! Is the reality really this differnt?

Your first post seems to show other areas of the world (Aus, SA, Africa even the Arctic) that have the same result. I haven't had time to analyse them yet in this way mainly because of the Lat/Long mixup.

It seems to me that the 1997/8 El Nino has really skewed the UAH and RSS results for certain areas of
the world. Some areas of the world -Europe, Asia, Antarctica may march to a very different drummer (AMO?). For me this really calls into question whether a global temperature has any real meaning and whether or not some areas dominate over others for a period of time.

Am I going a bridge too far here?

Steve H :]

Bob Tisdale said...

Steve H: The following is a graph the UAH TLT Anomalies for the “USA48”. I broke the data into three periods per your comment above.
http://i43.tinypic.com/zsx2rm.png

The first thing to note: Dividing the data that way doesn’t eliminate the El Nino “bump”. Due to the time lags involved, the Nov 1998 start month captures the direct (though lagged) impacts of the 1997/98 El Nino.

The second thing, isolating and removing the 1997/98 El Nino signal from the USA48 data is difficult. Note the secondary spike in 1999/2000. It’s clearly an aftereffect of the 1997/98 El Nino that occurred toward the end of the 1998/99/00 La Nina. And it would have a significant impact on the trend.

Global temperature is an easier dataset in which to isolate the direct impacts of ENSO events, yet there are debates about the proper year to begin post 1997/98 El Nino trends. Lucia, in her posts at TheBlackboard, uses multiple start dates.
http://rankexploits.com/musings/2009/hadley-may-anomaly-rose-0002c-from-april/

This is why I tend to steer away from trends that don’t cover the entire term of the dataset. I don’t mind, though, using average temperatures before or after the 1997/98 El Nino to show the step changes in temperature caused by that ENSO event. In fact, if you were to use the “search blog” feature above, you’d find quite a few posts with those step changes. Lots of different datasets.

You wrote, “Some areas of the world -Europe, Asia, Antarctica may march to a very different drummer (AMO?).” The AMO should have noticeable impacts on North America and Europe, and on Asian winters. It should have little visible effect on the Antarctic.

Anonymous said...

Bob,

Thanks for the chart and comments.

I did take the global El Nino "bump" and used that to get the dates. I have used them to try not to cherry pick and be consistent. As you say there are regional differences and I find these interesting in themselves. It is a problem in what to use as the El Nino period.

Ya, I know that Antarctica is not influenced by AMO. I added Antarctica as an afterthought and didn't notice the bad construction until after I'd posted. I was too lazy (and it was sack time)to change it.

So I'll spend some of the weekend looking at this from various viewpoints. Did you use anything different to get the raw GISS data as opposed to UAH?

Again - thanks for your time. I get back to you if I have more questions about this weekend efforts.

Steve H :]

Anonymous said...

Bob

Just a note:

Today WUWT posted the Alan Carlin report. Just wanted to point out that Figure 2-8 pg 43 and Figure 2-13 Pg 50 (with explanations) illustrate what I am interested in investigating.

It's just an idea, not even a hypothesis. It will be fun to watch where the data goes in the next 5 to 10 years. At least its testable within a reasonable amount of time!!

Andrew said...

Bob-Very interesting analysis! I notice that one of the largest discrepancies is in Africa. A recent paper by John Christy and colleagues documented some apparent problems with the, er, "official" data in East Africa including land-use effects on the Minimum temperatures:
Christy, J.R., W.B. Norris and R.T. McNider, 2009: Surface temperature variations in East
Africa and possible causes. J. Clim. 22, DOI: 10.1175/2008JCLI2726.1.

In his recent EPA submission there was a graph which looks strikingly like some of your figues:
http://icecap.us/images/uploads/EPA_ChristyJR_Response_2.pdf
See page 5.

Cheers.

Donations

Tips are now being accepted.

Comment Policy, SST Posts, and Notes

Comments that are political in nature or that have nothing to do with the post will be deleted.
####
The Smith and Reynolds SST Posts DOES NOT LIST ALL SST POSTS. I stopped using ERSST.v2 data for SST when NOAA deleted it from NOMADS early in 2009.

Please use the search feature in the upper left-hand corner of the page for posts on specific subjects.
####
NOTE: I’ve discovered that some of the links to older posts provide blank pages. While it’s possible to access that post by scrolling through the history, that’s time consuming. There’s a quick fix for the problem, so if you run into an absent post, please advise me. Thanks.
####
If you use the graphs, please cite or link to the address of the blog post or this website.