Opportunities and challenges in using catchment-scale storage estimates from cosmic ray neutron sensors for rainfall-runoff modelling

Abstract

Adequate characterization of catchment storage dynamics is crucial in hydrological models, yet scale-representative storage measurements are rare. Recent developments in Cosmic Ray Neutron Sensor (CRNS) technology and monitoring networks provide a powerful source of more scale-appropriate soil moisture data for many modelling applications. However, the potential in rainfall-runoff modelling is undeveloped. Here we present the first application of CRNS data in conceptual rainfall-runoff modelling and explore this potential in the context of a mixed-agricultural landscape in Scotland. We deployed and calibrated a CRNS in a heterogeneous soil-land use footprint over a ∼3-year period. In this generally wet environment, the CRNS shallow sensing depth and relatively high neutron count uncertainty were identified as major challenges. However, given the better spatial coverage (up to 14 ha) and ease for maintenance, CRNS was thought to represent the simplest approach for long-term monitoring of managed mixed-agricultural sites. We used CRNS-derived, as well as single point-scale estimates, of near-surface soil storage (SNS) to explore their characterisation of storage dynamics at the catchment-scale. Inter-comparison using linear regression showed that SNS related well to catchment-scale storage dynamics, however this relationship was stronger for CRNS (R2 = 0.91) compared to point-scale derived estimates (R2 = 0.76). Based on this, we evaluated the effect of using the CRNS and point scale derived SNS data to constrain storage estimates controlling runoff generation in a common rainfall-runoff model (HBV-light). Including CRNS or point-scale field SNS data alone in model calibration was especially useful for intermediate and wet periods. A combined model calibration using discharge and either SNS storage estimates provided a better representation of catchment internal dynamics, additionally reducing uncertainty during low flows. In the context of mixed-agricultural landscapes in humid environments, this study showed the potential of using CRNS over point scale data (in terms of representativeness for single point data and practicality for point sensor networks) to characterise the catchment storage-discharge relationship and inform hydrological modelling.