Potential Application of Cosmic-Ray Neutron Sensors to Infer Precipitation Rates

Abstract

More studies are finding utility in using cosmic-ray neutron sensors (CRNS)s to understand hydrological processes within the environment. This study explores estimating precipitation by analyzing CRNS derived soil moisture time series data. A simple bucket model is employed to estimate precipitation from changes in soil moisture. Two soil moisture time series were simulated by iteratively weighting Hydrus-1D soil moisture profiles to estimate soil moisture observed by the CRNS. The simulated soil moisture time series were used to parameterize the model and better understand the system. The Kling-Gupta Efficiency metric was used to evaluate site performance where a value larger than 0.9 is desired. The model was evaluated at four field sites: Manitou (CO), Santa Rita Mesquite (AZ), Kendall-Walnut Gulch (AZ), and Silver Sword (HI). Due to apparent noise in time series data, temporal aggregation was used to improve performance. The 24-hour aggregated Kling-Gupta Efficiency metric for all sites was 0.92, 0.93, 0.87, and 0.62 respectively. The field results demonstrated some limitations using CRNS derived soil moisture due to environmental noise and uncertainty in measured precipitation. Further field investigations are needed to improve the performance of this method and better understand the nature of the precipitation measured. This method may be useful where direct measurements of precipitation are unavailable, unreliable or poorly represent the required scale.