Friday, March 30, 2012

Predicting the UAH Anomaly

I was going to write about this month's raw satellite temperature measurements for the lower troposphere from UAH, and "predict" what the final anomaly for March would be. But I thought I'd better check something first.

At first glance, March seems to be warming up more than recent years, based on the raw data:


So you might say, sure, La Nina is nearly over and we expect apparent global warming to resume again and so on. But then I made this graph of the monthly averages for the raw UAH temperatures, versus that after all the adjustments for satellite drift, etc.:

which shows no correlation. You might think, well, maybe there's some correlation in there, which changes after they do some special adjustment every 1.5 to 2 years:


but that's just too much guesswork. So UAH really means their caveat at the bottom of the page:
The global-average data displayed on this page have only limited quality control, can undergo unannounced changes, and so should only be used as a general guide. Official, quality-controlled global lower-tropospheric temperatures, using more extensive processing procedures (and possibly different satellite instruments) are updated every month and are available at: http://vortex.nsstc.uah.edu/data/msu/t2lt/
So, with the understanding this is for entertainment purposes only (as they say), my prediction guess for the March LT anomaly is:

UAH global LT March anomaly = +0.12°C

Be sure to read this post by Roy Spencer, which explains some of the complications.

3 comments:

thefordprefect said...

I've been logging the "discover" data for a time and did this post some time ago:
http://climateandstuff.blogspot.co.uk/2011/11/just-how-good-are-satellite-derived.html

Spencer says the discover page is virtually worthless:

Roy W. SpencerDec 21, 2011 03:59 PM

If you would have read the disclaimer at the Discover website, you would have realized you can't compare the daily, automated, quick-and-dirty data averages there with the fully intercalibrated, quality-controlled UAH dataset that we update every month.

The Discover website data are meant to give a rough idea of how the latest month is shaping up compared to the same calendar month a year earlier, that's all.

To which I replied
thefordprefectDec 25, 2011 05:54 AM

Roy The data is a couple of degrees adrift from satellite to satellite. This is not good! Why bother even giving the data if it so far out to be worthless?

Although there is a statement ofn the discover page about accuracy it says nothing about adjustments. It says nothing about where similar but corrected data can be obtained.

All-in-all this is a poor answer to my criticisms.

If you can correct one series from sattelites , why not all. Or are the corrections too complex for a computer!!!

Please give a link to the adjustments made and data for similar altitudes that are approved by yourselves.

========

How can AMSU data be better from the same data?

Does Spencer, or one of his minions, individually fettle the data by hand? If a computer is used then why cannot the same corrections be applied to the "discover" data?

David Appell said...

Ford: Interesting link, thanks.

I don't think for a second that UAH (or anyone) fiddles the data. I think they adjust it as best then can. Satellites drift, for one thing, in a way that isn't predicatable. One such factor are solar storms that temporarily expand the atmosphere, which creates drag.

thefordprefect said...

Apologies!

Fettle does not necessarily imply falsifying.

I used the term to suggest correction of data by some undisclosed amd manual means.

If the discover data were adjusted the same way (alorythmically - how the height of satellite has changed, atmospheric changes etc.) there would be no problem!