Data assimilation of soil water flow via ensemble Kalman filter: Infusing soil moisture data at different scales

This paper assesses the value of multi-scale near-surface (0∼5 cm) soil moisture observations to improve state-only or state-parameter estimation based on the ensemble Kalman filter (EnKF). To the best of our knowledge, studies on assimilating multi-scale soil moisture data into a distributed hydrological model with a series of detailed vertical soil moisture profiles are rare. Our analysis factors include spatial measurement scales, soil spatial heterogeneity, multi-scale data with contrasting information and systematic measurement errors. Results show that coarse-scale soil moisture data are also very useful for identifying finer-scale parameters and states given biased initial parameter fields, but it becomes increasingly difficult to recover the finer-scale spatial heterogeneity of soil property as the observation grids become coarser. In state-only estimation, near-surface soil moisture data result in improvement for shallow soil moisture profiles and degradation for deeper soil moisture profiles, with stronger influences from finer-scale data. With the decrease of background spatial heterogeneity of soil property, the value of coarse-scale data increases notably. Soil moisture data at two scales with contrasting information are found to be both useful. By updating spatially correlated soil hydraulic parameters, deviated observations still contain considerably useful information for finer-scale state-parameter estimation. Eventually, by presenting a difference information assimilation method based on EnKF we successfully extract useful information from soil moisture data containing systematic measurement errors. The current study can be extended to consider more complex atmosphere input and topography, etc.