I am confused about the purpose of ARP, as you can see.
Part of the problem, as presented in the study, is that the rain comes localized in thunderstorms. So one farmer may get wet and his neighbor not. Hence the need for increased resolution beyond what is available from satelite data and the current software. I believe our study is using the WRF model, said to work better with weather conditions reported within 50 km of each other which apparently does not exist in rural Africa. Denoting a rainy season and a dry one, which I am sure they already know, is not predictable enough to know when and what to plant when the rain is unlocalized. Sort of like ketsup - it doesn't come for a long time and then you get it all.
Simulation or no, you seem to run into the old stockbroker caveat - past performance is not a predicter of future performance
Will studies of localized records for four years tell whether Africa is drying out? I wonder. If this study is looking at overall trends, how does it help the farmer in the now. Not to say it isn't important to know if Africa is drying out.
To the neophyte like me, it would seem that the GRAF machine is a better solution to the farmer's problem, if it is affordable through IBM. Sort of like vaccine for Ebola. I hope that what WCG did will lead to medicines for Ebola.
If and when there is a monthly update, I would like to see a discussion of the future of simulation with the possibility of direct calculation on the horizon. The GRAF does seem to rely on input which may not be there. Perhaps the researchers have already taken this into account.