Towards hyperresolution land data assimilation to monitor terrestrial water, energy, ecosystem, and hydrological disasters
Land data assimilation has contributed to improving the skill of land surface models (LSMs) to monitor and predict terrestrial water, energy, ecosystem, and hydrological disasters. Conventional LSMs are vertically 1-demensional models which assume that lateral water flows are negligible at their coarse resolution (>25km). Recently, 3-demensional “hyperresolution” land surface models (HLSMs) with a finer horizontal resolution (<1km) have been developed. HLSMs explicitly simulate surface-subsurface lateral flows, which brings new challenges in land data assimilation. I first introduce our recent advances in land data assimilation studies on a conventional LSM such as (1) uncertainty quantification, (2) parameter optimization, and (3) simultaneous estimation of both soil moisture and ecosystem dynamics by particle filtering. Then, I discuss the new challenges in hyperresolution land data assimilation. I performed a minimalist synthetic experiment where shallow soil moisture observations are assimilated into a HLSM by an ensemble Kalman filter (EnKF). I found that the explicit consideration of lateral flows induced by heavy rainfalls greatly contributes to efficiently using sparse soil moisture observations although the nonlinear dynamics of topography-driven lateral flows harm the performance of EnKF.
Sawada, Y. (2018). Do Surface Lateral Flows Matter for Data Assimilation of Soil Moisture Observations into Hyperresolution Land Models?. EarthArXiv doi:10.31223/osf.io/byguf.
Dr. Yohei Sawada Meteorological Research Institute, Japan Meteorologial Agency