Application of a Unified Land Model for estimation of the terrestrial water balance

Approximately 60 – 80 % of precipitation is returned to the atmosphere through evapotranspiration (ET) on global average, making it a key component of the surface water budget. In-situ measurements of ET are sparse and cannot be readily interpolated over large areas given heterogeneity in land cover. Furthermore, in situ ET measurements can be subject to large measurement errors. Here, we seek to evaluate a Unified Land Model, ULM, which is a merger of the Noah land surface scheme used in NOAA's weather prediction and climate models with the Sacramento Soil Moisture Accounting Model, used by the National Weather Service for operational streamflow prediction. Our goal is to estimate regional-scale water balances, and to compare the estimates with independent ET and streamflow observations over a set of large continental U.S. river basins and their interior sub-basins. This work is motivated by two objectives, first to quantify the evaporative component of the terrestrial water balance, and second to evaluate the large-scale prediction skill of ULM. The experiments consist of comparing ET estimates from: (i) an atmospheric water balance, (ii) satellite based estimates of ET, and (iii) ULM, forced with the same precipitation data used in the atmospheric water balance.