Enhancing hydrological models with remote sensing: a review of products, techniques, and uncertainties

Distributed hydrological models are crucial for flood prediction, drought analysis, and water resource monitoring. They are typically calibrated using streamflow observations at the watershed outflow to determine the best parameter values within their common ranges. These models are then applied to analyze management and climate scenarios. However, accurately representing hydrological complexities is challenging due to limited knowledge, data availability, and imprecise measurements. Uncertainties in these models arise from parameters, model structure, calibration processes, and data, especially in regions with scarce data. Consequently, hydrological models require extensive hydro-meteorological data for calibration and validation, which can be costly and time-consuming. Recently, remote sensing techniques advanced hydrological modeling by providing regular sampling of essential variables like precipitation, soil moisture, and evapotranspiration. However, thanks to technological advancements, numerous global and regional remote seeing products for the same variable have become freely available. These products vary in their algorithms, approaches, spatial and temporal resolutions, leading to diverse datasets for the same variable. Therefore, different products can perform differently in terms of parameter estimation, model robustness, and water balance predictions within the same area. However, each product may introduce biases or uncertainties, necessitating modelers to assess their performance and carefully choose the most suitable product for their study objectives. This research reviews commonly used remotely sensed products and the techniques and approaches for integrating them into distributed and semi-distributed hydrological models. Additionally, this review examines the uncertainties associated with different existing products and their performance within hydrological models.