Evaluating the influence of hydrologic signatures on hydrological modeling using remotely sensed surrogate river discharge

In recent studies, a surrogate river discharge model (SRM) has been proposed that uses remote sensing data, termed henceforth as the Surrogate River discharge (SR), instead of river discharge in hydrological model calibration. SR is calculated from the relative values of target and reference reflectance obtained from satellite observations and thus does not have a defined unit. Hence, to ensure physical consistency, the model calibration mandates the integration of a hydrological feature to translate the non-dimensional SR to a quantified river discharge estimate. In this context, an estimate of the mean catchment river discharge (QM) is proposed as a suitable signature. This variable is derived from the Budyko relationship, which only uses two climate average data: precipitation and potential evapotranspiration. This allows the SRM calibration to depend primarily on remotely sensed data and climate data. However, the model's sensitivity to the hydrologic signature approximation is generally unknown and should be characterized for improved probabilistic predictions in ungauged basins. Therefore, this study evaluates the impact of the hydrologic signature on SRM. The study first identifies the sensitivity of the SRM to the QM accuracy. Subsequently, the SRM integrated with the Budyko relationship (SRM^B) was compared to the Australian Landscape Water Balance model (AWRA-L) using real catchment data on 49 selected catchments of various sizes, topographies, and climates to assess its uncertainty. Then the investigation is subsequently broadened to include 384 catchments in Australia to analyze the sensitivity of SRM with respect to classified SR quality, as assessed through a range of hydrological signature estimation methods. To summarize, while the efficacy of SRM is contingent upon SR quality and mean flow accuracy, incorporating the Budyko framework yields effective mean flow estimations, bolstering prediction reliability. Given its efficient calibration, SRM demonstrates potential for broader application in hydrological studies.