Observation-Driven Estimation of Surface Water Balance Components from SMAP Measurements

With the availability of global satellite remote sensing observations of surface soil moisture, it is now possible to quantify important hydrological fluxes such as evapotranspiration (ET) and drainage. Furthermore, given the current level of accuracy of remotely sensed soil moisture, these fluxes can be estimated without the need for large-scale land surface or climate modeling. In this work, remote sensing data from the NASA SMAP mission, at 36 [km] scale, and gauge-based precipitation data over the US are utilized within an adjoint-state variation estimation method to obtain time-series daily estimates of ET and drainage. The approach uses only SMAP measurements and precipitation. Neither a hydrology or land surface model nor ancillary hydrologic data are used. ET estimates are compared to eddy covariance measurements from the AmeriFlux network and are shown to capture up to 70% of the in-situ measurements' annual variance. Similarly, Drainage estimates are compared to USGS streamflow measurements. On average Drainage under-estimates streamflow by 1–2 [mm day-1] with seasonal correlation (R2) varying between 0.52-0.77. These estimates close the surface water budget with only SMAP measurements and precipitation information.