Wednesday, April 20, 2022

Global projections for TempLS temperature reporting.

A couple of years ago I described here my way of implementing equal area projections for global data, and in particular the Mollweide and Robinson projections. Robinson would not have been a high priority, except that GISS ofers it for its temperature projections, so I can use that for comparison.

Events intruded, and I didn't implement it for routine use immediately, and when I later did, then of course mission creep crept in, and I thought of ways to use the pixel mapping methods for the actual integration (not that I am lackimg options).

However, it really should be done. The rectangular lat/lon method that I use conveys the necessary information in an easily understood way, and I will still be using it for some internal purposes, but it overweights behaviour near the poles visually. I calculated. Here is a table of the % by pixel of colors in my plots for March 2022, shown also below. Color 11 is the warmest (>4°C)

Color: 2 3 4 5 6 7 8 9 10 11
Lat/Lon0.190.551.855.110.3913.5124.7421.6313.638.41
Robinson0.220.662.175.5510.6814.1828.0123.1311.224.18
Mollweide0.230.652.235.5610.6813.9528.5423.2311.153.77

March was a month in which the Arctic was very warm, and you can see that there were more than twice s many of the warmest pixels in the lat/lon projection (approx 8% vs 4%). The other projections are close to the true % on the sphere.

I plan to show the three projections in a single frame, with navigation buttons. Robinson will show initially, but you can cycle to Mollweide or lat/lon using the buttons below.








0 comments:

Post a Comment