Above is a map showing the amount by which GFS has upgraded or downgraded its 15-day accumulated rainfall (forecast period starts from tomorrow).

The amounts by which it's upgraded or downgraded are averaged over the past 2 runs (any more and the more accurate increasing/decreasing trends in the more recent runs tend to be masked by the ones in the older runs without a complex technique to overcome it.... any less and potential one-off erratic jumps in a single run tend to be a problem). Ideally I would've liked to have used whatever number of consecutive runs have consistently been going for an increasing or decreasing trend but it's very hard to do with the limitations of the GrADS programming language.

It's designed to give an indication of which areas might hold more promise and which areas might turn out to be fizzers but the inherent problem with trying to extrapolate trends in consecutive model runs is that there's no guarantee that any upgrades or downgrades in recent runs will necessarily continue in future runs (sometimes they can even reverse).