Thursday, May 28, 2009

Homogeneity--how the fast one is pulled over us

I will take a break from analysing the data at Dave's preferred site to show you the homogeneity filter. This is a filter that tries to make the TREND of presumed bad sites fit the TREND of the presumed good sites. And it is this aspect of the editing process that makes me want to scream. A thermometer even at a bad site ought to have it's trend preserved, because the effects of the bad ought to be constant, or semi constant. Either that or throw the entire station out.

Thomas C. Petersen shows a couple of examples. The first is from the Journal of the American Meteorological Society.

He describes the procedure in a caption beneath a picture. Here is the picture and the caption below it.



“The homogeneity adjustments applied to the stations with poor siting
makes their trend very similar to the trend at the stations with good
siting.” THOMAS C. PETERSON, “EXAMINATION OF POTENTIAL
BIASES IN AIR TEMPERATURE CAUSED BY POOR STATION
LOCATIONS,” American Meteorological Society, Aug, 2006, p. 1078 fig 2


So, in the above picture they take a cooling station the red dashed line, and turn it into a warming station--by magic, we have warmed the world. In the second picture in the AMS article they take a slightly warming station and make it warmer. Caption below the picture



“Again, the homogeneity adjustments applied to the stations
with poor siting make their trend very similar to the trend at the
stations with good siting.” THOMAS C. PETERSON, “EXAMINATION OF POTENTIAL
BIASES IN AIR TEMPERATURE CAUSED BY POOR STATION
LOCATIONS,” American Meteorological Society, Aug, 2006, p. 1078 fig 3.

So, we no longer let the data talk we make it say what approved stations say.

In April of this year, Peterson presented a talk to the AMS in San Antonio, a wonderful place for a convention. He presented the homogeneity filter for Reno, Nevada which has some interesting lessons.



The upper chart shows the black, relatively raw data. That shows a 3 degree drop in 1937 or so (black line upper chart. They correct the data by making the temperature warmer from the 1940s through the present. That is the red curve. You can see that they have warmed the measured curve by up to 3 degrees C, which is about 5.4 deg F!.

They chose to keep the 1900-1937 data be the standard. In other words, they implicitly decided that the instruments used between 1900 and 1937 are better than the instruments used from the 1940s on. That does seem like the wrong decision, unless they think the world's technology is getting worse with time.

The other interesting thing is that they added half a degree to 3 deg C to the temperature in the 1990s. What they have decided is that Reno is too cool and it must be made warmer than the thermometers read. Obviously this takes a huge amount of hubris on the part of the editor who implicitly saying that he knows the temperature better than the thermometer does. So the editor raises the temperature, adding heat to the world and worrying those who don't understand these things.

What is the result of all this editorial sleight of hand? A warming of over half a degree for the US.

“The annual difference between the RAW and FILNET
record (Figure 2) shows a nearly monotonic, and highly statistically
significant, increase of over 0.05 [deg]C [per]dec. Our analyses of
this difference are in complete agreement with Hansen et al. [2001]
and reveal that virtually all of this difference can be traced to the
adjustment for the time of observation bias. Hansen et al. [2001]
and Karl et al. [1986] note that there have been many changes in
the time of observation across the cooperative network, with a
general shift away from evening observations to morning observations.
The general shift to the morning over the past century may
be responsible for the nearly monotonic warming adjustment seen
in Figure 2. In a separate effort, Christy [2002] found that for
summer temperatures in northern Alabama, the correction for all
contaminants was to reduce the trend in the raw data since 1930,
rather than increasing it as determined by the USHCN adjustments
in Figure 2.It is noteworthy that while the various time series are
highly correlated, the adjustments to the RAW record result in a
significant warming signal in the record that approximates the
widely-publicized 0.50 [deg]C increase in global temperatures over the
past century.”
Robert C. Balling and Craig D. Idso, “Analysis of adjustments to the United States Historical Climatology
Network (USHCN) temperature database, GEOPHYSICAL RESEARCH LETTERS, VOL. 29, NO. 10,, p. 1388

No comments:

Post a Comment