Things That Make You Go Hmmm Part 2
August 9, 2010Posted by on
Ok the last time I looked at the data I found that the adjustments made by NCDC to the Hanford and Lemon Cove records for Tmax and Tmin caused them to look like those stations do not support the conclusions of Dr. Christy’s 2006 paper, where as in the “Raw” data you see a huge warming trend in the Hanford Tmin, a slight warming trend in the Lemon Cove Tmin and both stations showing cooling trends in Tmax. Those trends go against what the computer models say should happen, matter of fact they are reversed. According to the models Lemon Coves Tmin should have a large warming trend because it is in the mountains and Hanford should not because it is in a valley. This was shown in the Snyder 2002 paper. So you can imagine my surprise that the new adjustment method that NCDC came up with in 2009 for USHCN adjusts the data in such away that the data all of a sudden matches what the models said back in 2002. The adjustments made were not small either, not when you are dealing with almost 3° F swing from a huge warming trend in the Valley station Tmin to a very small one of .45° F.
Now I was going to start looking at the USHCN records for the Merced/Yosemite pair, but I remembered something about overlap between USHCN and GHCN. Basically what it boils down too is that some USHCN stations are also in the GHCN stations. What makes that important is that GHCN still uses the old adjustment method that USHCN used to use, so if the Hanford/Lemon Cove pair is in the GHCN stations we can compare the old adjustment method to see if it gives us different results: ie does it agree with Dr. Christy’s paper? If not how much difference is there?
So I went and checked my copy of GHCN max and min files (Dec 2009 is the one I have unzipped atm) and yes Hanford and Lemon Cove are in there. So it’s time to put them into the spreadsheet and see what we got.
NOTE: GHCN records its data in Celsius where USHCN is in Fahrenheit, the temperature anomalies in the last post will not scale with the ones in this one. However the overall trends should not be affected, a warming trend should still be a warming trend in either scale and vice versa for cooling.
In Figure 1 we have the Tmax numbers for Hanford. You will notice that the data only goes up to 2005, that is because that is all there is in the GHCN file I got. Also note that I converted the USHCN Tmax adjusted data from Fahrenheit to Celsius and add it in.
Now here you see the GHCN Raw trend is almost -1.5° C over the 105 year period. The GHCN adjusted trend is down to a -.25° C, while the USHCN adjusted trend is about a .3° C of warming. So while the old adjustment method (First Difference) does lower the cooling trend it doesn’t flip it to a warming trend.
Now in Figure 2 we see the Tmax for Lemon Cove, with a “Raw” trend of about -1° C over the time period. The GHCN adjusted trend is almost 1.25° C and the USHCN adjusted trend is just a little over .25° C. Both adjusted methods change a cooling trend into a warming trend but the older method still used in GHCN is the more egregious here with an almost 2.25° C swing, where the newer method in USHCN reduces that swing to just a 1.25° C change (notice both stations with both adjustment methods produce swings that are greater than the IPCC backed warming trend of .7° to .8° C per century).
Now we move on to Tmin for both stations, starting with the Hanford Tmin in Figure 3
Here for the “Raw” we got a trend of 2° C over the time period. The GHCN Adjusted trend is about 2.25° C and the USHCN Adjusted trend is not even .5°C, closer to a .4° C trend, that is a huge adjustment from the raw.
So to recap the numbers for Hanford:
|Tmax Trend||Tmin Trend||Tmean Trend [Tmean=(Tmax+Tmin)/2]|
|Raw||-1.5° C||2° C||.25° C|
|GHCN||-.25° C||2.25° C||1° C|
|USHCN||.3° C||.4° C||.35° C|
So according to the numbers the Tmean trend should be the lowest for the raw with a strong daytime cooling trend and a strong nighttime warming trend. For GHCN a warming trend about .2° to .3° C over the global average with a low daytime cooling trend and a strong nighttime warming trend. For USHCN you got a small warming trend with a small daytime and nighttime trend. This can be seen in that my eyeball estimate is very close in Figure 4:
Now in Figure 5 we will look at the Lemon Cove Tmin:
For the Raw Tmin trend we get about .5° C of warming. For GHCN Adjusted we see a slightly higher trend of about .6 to .65° C of warming. Now for the USHCN Adjusted we get a trend of about 1.2° of warming, which is an increase of the trend by almost .75° C from the Raw trend and almost .6° C over the GHCN adjusted trend.
So a recap of the Lemon Cove numbers:
|Tmax Trend||Tmin Trend||Tmean Trend [Tmean=(Tmax+Tmin)/2]|
|Raw||-1° C||.5° C||-.25° C|
|GHCN||1.25° C||.6° C||.93° C|
|USHCN||.25° C||1.2° C||.73° C|
As we see there is a slight cooling trend in the Lemon Cove Tmean by the raw numbers with a strong cooling trend for daytime high temperature and a moderate warming trend during nighttime lows. GHCN adjusted shows a strong warming trend with a strong warming trend for daytime temperatures (an over 2° change from the raw numbers) and a moderate warming trend for nighttime lows. USHCN also has a strong warming trend which is very close to the adjusted global average, with a slight warming trend for daytime highs and a strong warming trend for nighttime lows
The GHCN and USHCN trends in the graph made by the v2 meanadj file and USHCN data appears to be almost exactly as my rough numbers in the table above, however the v2 mean file trend appears a little less than the numbers derived from the rough numbers in the table, only about a .1° to .15° cooling trend.
For Hanford (valley) the GHCN adjusted trends are close to the results of the Christy et al paper in that the Tmin warming that Dr. Christy pointed out is shown with a less of a cooling trend in Tmax trends. For Lemon Cove (Sierra) however it is even worse than USHCN. For Tmax. Christy et al 2006 found that there was no warming trend in either Tmax or Tmin for Sierra stations, but GHCN has an even higher warming trend then USHCN for Tmax, while almost no adjustment to the Tmin trend from Raw.
So once again the adjustments made to the dataset by NCDC are what causes the divergence from the Christy et al findings of no appreciable warming trend in either Valley or Sierra Tmax (both show a cooling in the paper), but a clear warming trend in Valley Tmin but no appreciable warming trend in Sierra Tmin.
now it’s off to look at the other pair and see if things work out the same, if it does there is 3 stations in Costa Rica I want to revisit and see if they are in the GHCN Tmax and Tmin files and see if the Valley/Mountain theme was just a Central Valley thing or if it will apply to the Beach/Mountain in Costa Rica.