Abstract:
Soil moisture controls the exchange of water and energy between the land surface and the atmosphere and exhibits memory that may be useful for climate prediction at monthly time scales. Large-scale observations of root zone soil moisture, however, are not routinely available. Assimilation of surface soil moisture observations (for example from satellites) into a land surface model (LSM) can yield improved estimates in the root zone.

Land surface models, however, differ significantly in their representation of subsurface soil moisture processes. Therefore, the propagation of surface information into deeper soil layers depends on the LSM that is used in the assimilation system. Here, we use the Land Information System (LIS) data assimilation testbed, an interoperable framework for sequential data assimilation that enables the integrated use of multiple LSMs, observations types, and data assimilation algorithms. In a suite of experiments we assimilate synthetic surface soil moisture observations into four different LSMs (Catchment, Mosaic, Noah and CLM) using the Ensemble Kalman Filter and investigate the impact of subsurface physics on the skill of soil moisture assimilation products.

Our results suggest that assimilation of surface soil moisture generally provides improvements in the root zone estimates. The magnitude of improvements in the root zone is highest if the true subsurface physics has a strong correlation between the surface and root zone, such as in the case of Catchment LSM, regardless of which LSM is used in the assimilation system. For northern-hemisphere summer conditions, we also find that the average skill improvements through data assimilation across different truth scenarios are comparable regardless of which model is used in the assimilation system. When evaluating year-round improvements, the use of CLM in the assimilation system yields more limited improvements than use of Catchment, Mosaic, or Noah. This suggests that a simpler LSM is a safer choice for assimilation, unless the use of a more complex physical model is justified. The improvements in the soil moisture products through assimilation was found to be sensitive to the geographic location, landcover, and soil type.