Thursday, February 4, 2010

A Note On UAH's High January Temperature

UAH just posted the satellite readings for January, and the result is the hottest January since UAH temperatures started in December, 1978. This has caused a bit of discussion in the skeptic community. I think a lot of people are thinking “With all the cold we had last month in the northern hemisphere, how could this be the warmest January in the last 30 years? Is it possible the satellite readings are wrong?”

Yes. It’s possible the satellite readings are wrong.

As it travels around the Earth, the AMSU (the instrument used to read what becomes UAH temperature anomalies) reads 30 scans in the direction perpendicular to its orbit. Two of these scans are directly below the satellite and the each of the other 28 get progressively farther and farther away from the satellite in both the port and starboard directions. The further away from the satellite, the larger the error in its temperature readings.

Note that I didn’t say possible error. There’s no question at all these other 28 readings are wrong. The amount of error can be as high as 30 degrees Kelvin.

These errors are corrected via software. For each channel, the software checks the two neighboring channels and previous readings. This procedure is called a “Limb Adjustment”. I have a short discussion of this in the “Adjusting Footprint Brightness” section of the post I made here, and you can find a full technical discussion of the issue here.

Unfortunately, the Limb Adjustment uses statistical techniques to determine the adjustment values for each footprint, not the hardcoded technique I discussed in my blog post. It’s possible these statistical techniques fail in the face of unusual weather events like what we saw this January.

But without having access to the source code, I can’t tell if these unexpectedly high readings are due to a statistical error in the software calculations or are actually correct.

Some Useful Climate Code
The Limb Adjustment of AMSU-A Observations: Methodology and Validation

No comments:

Post a Comment