GISS Map Maker Data Compared to GISS LOTI Data
September 7, 2010Posted by on
While playing around with the GISS Map maker program, I had a thought: “How accurate is the Map Maker’s estimated Global Mean to the one in their Land Ocean Temperature Index (LOTI)?”. So I went ahead and started making a spreadsheet by bringing the most up to date LOTI data into it first. You can get the NASA GISS LOTI data here: http://data.giss.nasa.gov/gistemp/tabledata/GLB.Ts+dSST.txt
After that I went to the GISS Map Maker Program ( http://data.giss.nasa.gov/gistemp/maps/ ) and started making maps. I set the parameters to Annual Dec-Nov, GISS analysis with Had/Reyn SST with the 1951-80 baseline with the 1200km smoothing. From there you just go year by year making maps for each year. In the upper right hand corner is the estimated global mean anomaly for that year. From there I just copied it into the spreadsheet for that year. I then repeated the process for the 250km smoothing option. This allows us to see the differences in the yearly global mean anomalies for both 1200km and 250 km compared to the LOTI and if there is any differences between each trend.
First lets compare the Map Maker 1200km data to the GISS LOTI which is also calculated with a 1200km “smoothing” as GISS puts it. As can be seen there is a little divergence between the two where there should be none. This translates into a small divergence in the trend between them. The GISS LOTI looks to the old mark one eyeball to be a trend of about .74°C and the Map Maker 1200km to be slightly less at about .73°C.
While that difference between those two is interesting, it is not as interesting as the differences between them and the Map Maker 250km trend and exactly where the 250km data diverges from the two 1200km “smoothed” products the most. First the trend for the 250km product is about .66°C to .67°C which is a slight drop in trend compared to the other two, however its where the divergence that occurs that is the most interesting: Between the years 1881-1894 and 2002 to the present. Now why is that interesting? It has to do with the actual numbers of stations used in the GHCN database for those years. Click the link to the following Graph to get a close up picture: http://chiefio.files.wordpress.com/2009/11/thermometer-records-by-year.gif . You can also see that same graph from the NASA GISS stations data page here (but smaller): http://data.giss.nasa.gov/gistemp/station_data/
What you see in those graphs is that the total number of stations in the 1881-1894 time range is very similar to what we see today, but where the number of stations used in the database grows over time until the 1950’s, 60’s and 70’s from 1881 the number has been decreasing recently, the “Great Dying of Thermometers”. This touches on something that has been debated on in the blogosphere on whether the drop in stations has effected the trends in the data by the way the data is analyzed. We can now answer that question for the NASA GISS dataset with an affirmative. It is clearly shown in the above graph of GISS data that the more infilling there is, the more warming there is. This also shows that Dr. Hansens assertion that temperatures correlate over 1200km distance is false as well. Unlike others that try to test this with theoretical models, simulations and toy worlds this was done using the real world data as adjusted by NASA GISS and their GISTemp program. However it can also be shown that there appears to be a threshold to where it effects the data, as shown the divergence is only really noticeable when the number of stations have been at their lowest.