What was done
The authors compared soil moisture simulations made by eleven different models within the context of the Second Global Soil Wetness Project (a multi-institutional modeling research activity intended to produce a complete multi-model set of land surface state variables and fluxes by using current state-of-the-art land surface models driven by the 10-year period of data provided by the International Satellite Land Surface Climatology Project Initiative II) against real-world observations made on the top meter of grassland and agricultural soils located within parts of the former Soviet Union, the United States (Illinois), China and Mongolia that are archived in the Global Soil Moisture Data Bank.

What was learned
Guo and Dirmeyer report that "simulating the actual values of observed soil moisture is still a challenging task [our italics] for all models [our italics]," noting that "both the root mean square of errors (RMSE) and the spread of RMSE across models are large [our italics]," and that "the absolute values of soil moisture are poorly simulated [our italics] by most models [our italics]." In addition, they find that "within regions there can be tremendous variations [our italics] of any model [our italics] to simulate the time series of soil moisture at different stations."

What it means
How serious are these large errors and tremendous variations? It would appear they are very serious, based on a number of explanatory statements made by Guo and Dirmeyer.

First of all, the two researchers say "the land surface plays a vital role [our italics] in the global climate system through interactions with the atmosphere." Second, they state that "accurate simulation of land surface states is critical [our italics] to the skill of weather and climate forecasts." Third, they write that soil moisture "is the definitive [our italics] land surface state variable; key [our italics] for model initial conditions from which the global weather and climate forecasts begin integrations, and a vital factor [our italics] affecting surface heat fluxes and land surface temperature."

In consequence of what "those in the know" thus describe as large errors and tremendous variations in what they readily characterize as vital, critical, definitive and key elements of state-of-the-art land surface model simulations of soil wetness (which is a pretty basic parameter), it would appear that little faith should be placed in what they portend about the future.