I am assuming that this study is limited to sub saharan africa, an area of africa that might expect to have some rainfall.
In many ways "sub saharan africa" is a political term and not climatic; but I can find a line called "the transition zone" on the internet. Taking that into consideration, I estimate the area of sub saharan africa to be about 1/2 the total area of africa by eyeballing a geographic map without political divisions. Thus 1/2 of 30,370,000 sq K = 15,185,000 sq K. Possibly more, if my estimates are off.
I understand that each task that we crunch represents 250 sq K (which I shall call a "tile") and that each result represents a simulated time span of 48 hrs over that area. Also, I understand that each result is returned for a subsequent simulated 48 hrs repeatedly until a total of 1 years worth of data simulated for that 250 sq K unit has accmulated.
Therefore, the area of study should be divided by each tile's area to discover the number of tiles that would represent sub saharan africa. (15185000/250=60740 tiles)
We are crunching about 1600 units/day. At that rate, it will take 38 days to process each tile in the set once. (60740/1600)
The work on each tile represents 48 hrs. There are 182.5 such periods in a year. Currentlty therefore it will take 182.5 X 38 days or 6935 days or 19 years to complete the project.
This seems impractical.
Perhaps the initail paucity of task release is related to the construction of the physical atributes represented in each tile. That will ease in the re-iteration phase because they shouldn't change.
It will be interesting to see how the project progreses. It would also be interesting to know the are under consideration.