Assessing the Reliability of Integrating global data with the GEOframe-NewAGE hydrological model

Ground-based precipitation measurements are essential for meteorological and hydrological studies; however, they face challenges such as spatial variability, limited coverage, and instrument precision. In contrast, global precipitation products derived from Earth observation (EO) and reanalysis datasets provide broader spatial coverage, making them particularly valuable in regions with sparse or inaccessible ground-based observations. EO data detect precipitation by utilizing electromagnetic signals across specific frequency bands to generate precipitation datasets. The reanalysis-based datasets combine model outputs with observational data to create comprehensive precipitation records. Despite their advantages, both EO and reanalysis datasets are subject to uncertainties from various sources and often exhibit significant biases when compared to ground-based measurements. These biases can introduce errors in hydrological modeling, affecting model calibration and ultimately compromising the accurate simulation of hydrological processes. This dissertation explores the integration of global precipitation datasets into hydrological modeling to improve the representation of water cycle dynamics within the catchment. The first chapter of this study introduces the application of global datasets in hydrological modeling, emphasizing their significance and associated challenges. The second chapter delves deeper into the application of hydrological modeling to simulate the complex interactions of the water cycle within the Upper Nera River Catchment in central Italy. Building on this foundation, Chapter 3 assesses the influence of the spatial and temporal patterns of global precipitation products on representing the water cycle in mountainous regions and investigates how their inaccuracies may impact hydrological model performance. Finally, Chapter 4 aims to improve the accuracy of global precipitation data by refining snowfall estimates through the assimilation of satellite-retrieved snow depth into the snow component of the hydrological model. Chapter Two of this study demonstrates that effective modeling solutions for complex carbonate basins can be implemented even with limited data availability. Given the vulnerability of these basins to drought and climate change, exploring alternative modeling approaches under such data constraints is crucial. A hybrid method combining time series analysis and reservoir modeling is proposed to capture the behavior of carbonate basins. Time series analysis estimates the contributing area and response time of the fractured carbonate system beyond the catchment hydrographic boundaries, with results aligning with previous field surveys in the literature. This information is used to develop a reservoir-based model with the GEOframe-NewAGE modeling system, which is validated against in situ discharge observations and compared to EO data for evapotranspiration and snow. Our findings demonstrate that flows from carbonate rock areas outside the hydrographic boundaries significantly influence the water budget of the upper Nera River. The storage capacity of the carbonate basin plays a critical role in sustaining river discharge during drought years. While drought is generally alleviated in individual dry years, its effects become more pronounced in consecutive dry years, resulting in slower recovery during multi-year droughts due to the time needed for precipitation to replenish depleted storage. This unique behavior renders these basins particularly vulnerable to the more frequent and severe droughts anticipated under future climate change. Chapter Three focuses on the reliable application of global precipitation data in hydrological modeling, specifically examining how biased spatial and temporal patterns in precipitation data affect model performance and uncertainty. The study utilizes the European Meteorological Observations (EMO) and Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) global datasets as inputs for the GEOframe-NewAGE hydrological model to simulate the hydrological processes of the mountainous Aosta Valley catchment in northwestern Italy. The uncertainty of the hydrological model, GEOframe-NewAGE, forced with global precipitation data is assessed using a proposed method called Empirical Conditional Probability (EcoProb). Although traditional performance metrics suggest similar outcomes for models using both global precipitation products (EMO and CHIRPS), the uncertainty analysis indicates less reliability when CHIRPS is used as the precipitation input. To leverage all available information, the spatial correlation of CHIRPS is combined with a subset of rain gauges using the EcoProb method to modify the EMO precipitation data. This approach integrates the strengths of both EMO and CHIRPS, which offer higher temporal and spatial correlation with ground observations, respectively, into a unified precipitation product. The resulting dataset, referred to as the EcoProbSet product in this study, outperformed both CHIRPS and EMO, reducing the uncertainty introduced into hydrological models compared to the original global datasets. Up to this point of the study, the correction/integration of global precipitation data has relied on a subset of precipitation gauges. To advance this further, Chapter Four of this study explores the potential to improve global precipitation data by assimilating EO-derived snow depth into the hydrological model. This chapter introduces a novel approach to characterize orographic precipitation in mountainous regions, using a particle batch smoother to incorporate Sentinel-1-derived snow depth retrievals in order to correct the EMO precipitation product over the Aosta Valley catchment. The results indicate that the corrected EMO precipitation aligns more closely with the kriging interpolation results, which were derived from 81 precipitation gauges distributed across the catchment and used as the reference in this study. Additionally, the considerable improvement observed during the winter period highlights the effectiveness of the snow depth assimilation approach in correcting precipitation estimates. The GEOframe-NewAGE model forced with the corrected EMO demonstrates greater accuracy compared to the model forced with the original EMO.