Enhancing the reliability of hydrological simulations through global weather data assimilation in watersheds with limited data

Hydrological models are critical for water resources planning and management. The precision and reliability of the simulations hinge greatly on the accessibility and quality of available input data. Particularly in developing nations, the major challenge in modeling is the scarcity of fine-scale spatiotemporal input data, specifically precipitation. Remotely sensed weather data has been increasingly used in recent years. However, they possess bias compared to ground observations due to the nature of indirect measurement and may affect the simulated water balance. To address these limitations, we explored data assimilation techniques to improve the Global Precipitation Measurement product (IMERG) precipitation using limited ground observations. Multiple assimilation methods are applied by incorporating Linear scaling Correction Factor (CF) and Power Transformation Function methods (PF). The assimilated IMERG precipitation from the most effective method identified, is utilized in an eco-hydrological model, and the resulting stream flow simulations are validated against observed flow data. The findings indicate that assimilated precipitation enhances the monthly flow statistics in both CF and PF methods and also in conditional merged precipitation. An ensemble of hydrological simulations, outperformed those based on raw IMERG precipitation. Additionally, the hydrological simulations are compared with observed gauge precipitation data and the widely used Climate Forecast System Reanalysis (CFSR) dataset in data-limited watersheds. The simulations utilizing the assimilated IMERG dataset (NSE = 0.52) are comparable to gauge precipitation-based simulations (NSE = 0.61) and significantly superior to CFSR-based simulations (NSE=-0.2). These results highlight the potential of utilizing assimilated remote sensing data for hydrological modeling in data-limited watersheds, leading to improved simulation accuracy and reliability.