Quantifying Uncertainties of Remote Sensing-Based Rainfall and Evapotranspiration Products for Groundwater Recharge Estimation in the Tikur Wuha Watershed, Rift Valley Lakes Basin, Ethiopia
No Thumbnail Available
Date
2024-08
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
Groundwater recharge estimation is important for better groundwater management and decision making. Despite advancements in methods for groundwater recharge estimation and the availability of different methods, the results acquired are still variable and uncertain. Due to its hydrogeological formation, slight recharge variability in the Hawassa basin causes significant groundwater level variability. Therefore, this study focused on investigating groundwater recharge estimation uncertainty caused by input data in the Tikur Wuha watershed. Additionally, this study aimed to investigate the effectiveness of rainfall data assimilation from ground observations and climate hazards group infrared precipitation (CHIRP/S) satellite rainfall estimates (SREs) to decrease the uncertainty of recharge estimations. Accordingly, using a conditional merging procedure, two versions of resampled and spatially bias-corrected CHIRP estimates were merged with ground measurements. Additionally, the physically based fully distributed hydrological model WetSpa was used to simulate 30,000 possible combinations of parameters (i.e. randomly generated through Monte Carlo simulation stratified by Latin hypercube sampling (LHS)) for the three model setups. The M1 model setup was developed based on the rainfall measurements obtained from rain gauge stations scattered in and around the Tikur-Wuha watershed in Ethiopia, and the M2 model setup was developed using bias-corrected SREs of CHIRP merged with relevant ground station records. In contrast, the M3 model setup was executed using version 3.8a GLEAM evapotranspiration. One hundred best-performing parameter combinations were selected for each model setup to generate spatial recharge statistics and assess the resulting uncertainty in the recharge estimates. The results of the applied performance measures (i.e., seven) on the corrected and merged CHIRP SREs show that the percentage of detection (POD) and percent volume error (PVE) improved. Moreover, over the sparsely populated western part of the Lake Hawassa basin, the bias-corrected and conditionally merged CHIRP SREs outperformed the estimates obtained by CHIRPS. However, the devised multistage bias correction was limited in considering dry-day events during bias correction, which affected the bias correction performance of the CHIRPS product. On the other hand, the results of the uncertainty assessment revealed that enhanced spatial recharge estimates can be produced through improved CHIRP-based SREs. The replacement of ET estimates using ground meteorological records with the GLEAM dataset reduced the Cv value by 54% compared to M2. However, uncontrolled irrigation water uses and total recharge from
irrigation fields scattered across the Tikur-Wuha watershed were not considered in the study, which is a limitation of the study. Finally, future research should concentrate on methods of fusing to understand the benefits of various approaches and produce more precise rainfall records. Additionally, future studies should consider the contribution of irrigation water to the total recharge of the watershed to analyze recharge uncertainty.
Description
Keywords
Quantifying Uncertainties, Remote Sensing-Based Rainfall, Evapotranspiration Products, Groundwater Recharge Estimation, Tikur Wuha Watershed, Rift Valley Lakes Basin, Ethiopia