While the rainfall maps that have been produced are a good start, it is necessary to do adjustments that will produce more accurate maps. The adjustments to the maps need to take into account new information obtained from new data or historical data. The question is what is a good methodology for making such adjustments?

We propose a methodology that takes an existing map and makes geographically variable relative adjustments based on new data.

Note, for each ground measurement, we will have a measured monthly rainfall amount, and an amount which is presently estimated by current rainfall maps. The relative difference is the relative adjustment at the rainfall station location. Now for any pixel location, we propose taking a weighted average of the relative adjustments at nearby rainfall stations in order to determine the adjustment at that pixel location. We propose using a 1/(distance-squared) weighting for taking the average adjustment for nearby stations.

The averaged relative adjustment is then applied at each pixel value to produce the new adjusted rainfall map.