Soil moisture assimilation in urban watersheds: A method to identify the limiting imperviousness threshold based on watershed characteristics

Data assimilation is well suited for merging observed and simulated values to improve hydrologic forecasts. Typically, observation data that is assimilated into hydrologic models can be obtained from in situ gauges, radar, and remote sensing. In an urban environment, the benefits of soil moisture assimilation are not clearly defined. This work aims to identify the feasible imperviousness range at which it is advantageous to assimilate remotely sensed soil moisture to improve hydrologic forecasting in an urban watershed. A synthetic experiment was set up to simulate the retrieval of soil moisture onto a spatial grid and assimilate it into a hydrologic model. Sub-catchments were set up such that they would represent areas within that retrieval grid with varying levels of development. Multiple rainfall events were simulated with forecasts of up to 12 hours for each sub-catchment. The results of these simulations indicate that when areal average soil moisture is assimilated into an urban sub-catchment model which has a level of impervious that exceeds a threshold value, there is a decrease in model forecast performance, indicating that soil moisture assimilation is no longer beneficial. A quick way to determine the imperviousness threshold was then derived using a modified NRCS-CN method which matches the results of the synthetic experiments. This methodology was then further tested using real-world urban watersheds and shown to be a valid approach that can be used to quickly determine whether soil moisture assimilation would be beneficial for the watershed of interest. In general, this work has shown that hydrologic forecasting in an urban watershed can benefit from soil moisture assimilation even when imperviousness reaches 65–75%. Additionally, this work found a quick way to validate further if soil moisture assimilation is beneficial using some simple watershed characteristics.