My post How not to measure temperature, part 51 was also cross posted over at Climate Audit, and has created quite a stir when Atmoz, who is at the University of Arizona, tried to demonstrate that the temperature spike shown in the GISS data at Lampasas, TX, was not due to the relocation next to a building and asphalt parking lot, but rather some problem with GISS algorithm to do homogeneity adjustment to the data.
Steve McIntyre had doubts and posted a tongue in cheek rebuttal where he blamed the problem on UFA’s (Unidentified Faulty Algorithms). It seems reasonable given the fact that Arizona is already the center of surface measurement weirdness given the parking lot weather station operated by the Atmospheric Science Department of the University of Arizona.
Enter serendipity. Warren Meyers’ son Nicolas, has been actively surveying Arizona stations for his school science project. My inbox had a new station from him today, Miami, AZ. So I decided to take a look at it.
As is typical when an MMTS sensor gets installed by NOAA/NWS to replace the traditional Stevenson Screen, it got closer to human habitation, and in this case, a LOT closer. Too close I’d say:
So I though I’d take a look at the raw GISS temperature plot for Miami, AZ to see if the move would show a spike, it did:
From NCDC’s MMS database, they have a map showing station moves. This is at the Magma Copper Mine in Arizona, and the station used to be further away from the administration buildings near the pit:
Seeing a similar scenario to what occurred in Lampasas, TX, where a rural station was moved from a cooler location to a much warmer one, I decided to do the same sort of comparison on the GISS temperature plots as I did before:
RAW GISS DATA:
HOMOGENIZED GISS DATA:
HOMGENIZED GISS DATA OVERLAID ON RAW DATA:
Notice that after the GISS homogeneity adjustment, the past temperatures go down, with the present acting as a hinge point, thus making the slope of the temperature trend rise. The new slope is purely artificial, and appears to be an artifact of data adjustment by NASA GISS on this rural station. This is the second instance of this happening, the first being seen in the GISS Lamapasas, TX data adjustment for homogeneity.
In both cases, the abnormal spike coinciding with a station move near the present time remains in the record, and that is what the homogeneity adjustment is supposed to catch and remove as I understand it.
In a comment on the subject, Steve Mosher offers an explanation:
In Hansen 2001 Hansen says he uses nightlights to determine
if a station is Rural in the US and population everywhere else.
Miles city population is less than 10K which makes it rural,
BUT, nightlights ( satellite imagery taken in 1995)
indicates a brightness factor for Miles of 26! effectively making it urban.
I concur. There appears to be a flaw in the GISS nightlight methodology and adjustment algorithm. I look forward to seeing GISS investigate, and if this problem is indeed verified, a dataset correction.