Another Problem with GISS Infilling?
November 1, 2010Posted by on
Over on WUWT Steven Mosher has a very interesting post up about problems with the latitude and longitude that NCDC gives in it’s GHCN dataset. What he has found if that the coordinates given by NCDC for different temperature monitor stations are off and not just by a little bit, sometimes it’s by as much as 300km from their actual locations. He shows how this error is then transmitted from GHCN into the GISS analysis via the GISS nightlights adjustment for UHI. One example he shows is that the temperature station is at an airport out among the sands of a desert. Now this should cause GISS to find nightlights and also to list the station as airport, but it doesn’t. According to GISS that station comes up dark and is listed as desert. Why? Well the WMO (World Meteorological Organization) has the coords listed for the airport but GHCN has the coords 300km away in the middle of the desert with nothing surrounding it. This will obviously skew the numbers for that station in the GISS analysis,but how much is not known at the moment and we do not know how many other stations are incorrectly placed (some listed as dark when they should be lighted and vise versa) and how much the global numbers are affected. While I have tried to explain it adequately it is better to see the post it self which includes Google Earth shots that drive the point home before I move on to the next part. Here is the link to the post:
Now I didn’t read all the comments on WUWT yet but a thought occurred to me: If this error causes this problem in GISS could it cause another type of problem? My answer is yes.
If you haven’t read up on the GISS analysis you might not know that they divide the planet up into Grid squares and they use a formula to do a weighted average of all readings inside that grid cell to come up with the Grid’s temp anomaly. This is what you see in the GISS maps expressed in colors and what is used to help come up with the GISS Global Temperature Anomaly. Now one step in that process is they have a limit on distance, they state in their reference material that they start from the center of the Grid box and only use stations within 1200km of it. Now if GISS has stations up to 300km off from where they actually are at, then they could also be placing stations into the wrong Grid box, therefore that Grids numbers are wrong and so is the one that station actually does belong in. How much that is not known at the time or even if one of these misplaced stations does wonder from its correct grid box.