But is it about the average state of the reservoir or the extreme case of catching (all) the water you get during unlikely but intense weather events?
I think a correct reading of the article, which is perhaps too oblique on this point, is the loss of water to runoff where it is not in a watershed with a major reservoir, and not able to infiltrate into the aquifer. Add some runoff, such as agricultural runoff, is heavily polluted, complicating the matter even more.