Challenges and progress towards multi-scale hydrologic data assimilation

Subsurface moisture and temperature, and snow/ice stores exhibit persistence on seasonal to interannual time scales. This persistence has important implications for the extended prediction of climate and hydrologic extremes. However, errors in forcing, parameterization, and physics, accumulate in modeled land surface stores, which leads to future errors in water and energy partitioning. This has motivated the development of land surface data assimilation methods, which constrain land surface simulation models by forcing them primarily by observations, and by using observations of land_surface storages to realistically constrain model evolution using data assimilation techniques. This development: (1) improves understanding of the time and space variability of hydrological and energy budgets, (2) mitigates land surface parameterization and observation errors through continuous simulation-observation intercomparison, and (3) improves the initialization and dynamics of land surface states in numerical weather prediction models, for more realistic weather and climate predictions.