The attached figure is based on an analysis of 17 climate models, showing average changes (%) in four different drought variables for the end of the 21st century under RCP 8.5 (the plausible high warming scenario) 
Brown areas are getting drier (less precipitation, soil moisture or runoff/streamflow compared to today), blue areas getting wetter
Areas that are NOT hatched are regions where the models generally agree on the sign and magnitude of the change, areas where we have the greatest confidence
The first thing you may notice is that these all look very different! Patterns of change for precipitation vs soil moisture vs runoff are different
In some areas, it even looks like they are going in opposite directions!
For example, look at the Central United States. You can see hatching in the precipitation figure, which means the models don't agree on what will happen (some models=more precipitation, some models=less precipitation)
But now look at soil moisture for the same area, either at the surface (most important for shallowly rooted vegetation, like crops) or deeper down (more important for things like trees with very deep roots)
All the models general agree that there will be significant declines in soil moisture over this area, despite no robust change in precipitation
