GISTemp Trend Map Data Analysis

In my last post (which if you haven’t read yet you really need to or you won’t get everything I’m saying here) I showed how, at least on the GISS Map Maker program, that the 1880-2009 gridded trends have a couple of hidden blemishes. For most of the land area “trends” have interpolated data used as the basis for its trend not observed data through out the time period. This applies either back around 1880 when there was very little stations outside of the US and Europe or to the “Great Dying of Thermometers” starting around 1990. In some cases like Central Africa and portions of Central South America the data is “infill” or as I like to call it a SWAG in both 2009 and 1880. That’s right in those areas the entire trend of its “warming” for that grid point is a guess with very little observable data from those areas over that almost 130 time frame.

That’s fine and dandy for those areas where it is complete or mostly infill but what about the areas where there is a lot of historical data to compare to?

You would think that the infill would have no impact on those grid cells right? Why would they need to infill since they have actual observed readings right?

Well we can check that using the data provided by GISS from their Map Maker program. You see by turning down the infill to 250km you get closer to the actual observed data for those areas. Now if they have actual observed data for the entire or close to entire time period in question for a specific grid then there should be no influence from infill and the turn back from 1200 to 250 would not change the value of the trend. Sure there is still some infill since it does say 250km “smoothing” but you expect that to be  a small portion of what is left and not every grid box or even a majority of them to change.

Now before I show the results of the comparison between the trends for 1200km and 250 km infill lets lay out what we are talking about.

First on the Gridded Global map from the GISS site as shown at this link:

http://data.giss.nasa.gov/cgi-bin/gistemp/do_nmap.py?year_last=2010&month_last=1&sat=4&sst=0&type=trends&mean_gen=0112&year1=1880&year2=2009&base1=1951&base2=1980&radius=1200&pol=reg

Now from this is an option to get the data and from the following link you get the page for the data for the map in the above link:

http://data.giss.nasa.gov/work/gistemp/NMAPS/tmp_GHCN_GISS_1200km_Trnd0112_1880_2009/GHCN_GISS_1200km_Trnd0112_1880_2009.txt

Now when you are looking at the data you see a bunch of columns with numbers. The first 2 columns is the label GISS gives for each box on the gridded map. They start at the bottom of the map in the lower left hand corner and go left to right from there filling the boxes. You repeat this for each layer going up the map. These boxes are centered on the Longitude and Latitude of the center of each grid box. That Long/Lat reading is the next two columns of numbers. The next column of numbers is the temperature trend GISS believes is the actual change in temperature for that grid. A 9999.0000 means no data.

So this is how I handled the data:

Step one copy the data from that page into a Spreadsheet. Now I got rid of the grid labels since I didn’t need them, but kept the Long/Lat numbers since that acted as a way to line up the numbers for 1200km and 250km. First thing I noticed is that there is 16,200 individual grid boxes and of those only 9,232 is used by the GISS land analysis. The other 6,968 is taken up by the Had SST anomaly data and is not needed for this so is turned off and comes up 9999.0000. From there I went and turned down the map from 1200km to 250 km infill and pulled up the data. I then copied it into the spreadsheet program and then started eliminating every grid where there wasn’t an overlap of data, this cut the number of used grids down to 2,778. So there is less then 17% of all grids and only 30% of all land grids that GISS has observed data for in the 1880-2009 trend. Now of this 2,778 grids you would not expect the majority of them to have had their trends change by going from 1200km infill to 250 km infill, however the answer is that almost every grid trend changed.

Differences in Trend between 1200 km and 250 km Infill

Now how I did this was take the data from the 250km infill and subtract the data from the 1200km infill and this gives you the difference between the two and in the correct direction of change. (I almost pulled a Mann when I first did it and had the thing inverted). What you notice is that one very big area of change in the negative direction is at the far right end. The Long/Lat where those differences occur is for a strip of land up where northern Russia meets the Arctic Ocean and what you see is that with 1200 km infill you had a warming trend of 1.62 degrees change into .61 Cooling trend. Other areas you see the opposite occur example there is a strip where the grids had 2.61 degrees of warming and when the infill was ripped out it went up to 3.39 degrees of warming.

So the bottom line is that we do not know with any certainty what the real trend is for the globe because of GISS infilling.

Advertisements

One response to “GISTemp Trend Map Data Analysis

  1. vjones February 28, 2010 at 7:28 am

    Great work. We [hit blog link]have been working mostly with individual station trends from a database*. I’ve only put up on the blog a small fraction so far and not much written up yet on GISS outputs. Very clearly there are things going on at high latitude. I’ve started to become a bit obsessed with the grid square thing but am really only starting to work out the best way to tackle it. I really like your approach using GISS’s own tools.

    *which will be made available to all, just a bit of work to do still.

    It would be interesting to see your graph in 3D – Change vs Lat vs Longitude. Should be able to do this for you (email me: verityajones at gmail dot com)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: