Satellite remote sensing and modeling of the hydrosphere for understanding terrestrial water cycle dynamics at different scales
Seyoum, Wondwosen Mekonnen
MetadataShow full item record
Water resources are important to both society and ecosystems. However, humans put pressure on water resources with stresses that are likely to be exacerbated by the change in climate. Nonetheless, the lack of continuous data availability and inadequate monitoring networks has been a challenge to the scientific community. Recent advancements in satellite-based hydrology have demonstrated hydrologic variables can be measured from space with sufficient accuracy at limited regional and global scales (GRACE’s spatial resolution is 200,000 km2). Therefore, research on the enhancement of the utility of satellite products in understanding and monitoring the water cycle at local scales (with size of 5,000 km2) is necessary, especially to complement studies in the absence or malfunctioning of in-situ observations. This dissertation sought to (1) estimate the spatial and temporal variation of hydrologic fluxes and storages at different scales using satellite remote sensing data, (2) assess the efficacy of publically available data (e.g. satellite remote sensing data) on our ability to predict/understand the terrestrial water cycle and the implications for water management, and (3) measure the relative effect of human-induced (e.g. abstraction) vs. climatic variability on the terrestrial water cycle. Moreover, the potential of multi-source datasets and integrated approaches for predicting the variability were evaluated. The work presented in this research has been conducted using a combined approach of processing and interpretation of satellite data, numerical modeling, analysis of in-situ data, and statistical and geospatial analysis in an effort to overcome data paucity. The results demonstrated the capability of GRACE at measuring water storage variations on a regional scale based on results from a robust integrated hydrologic model. Further, merging GRACE data with other data sources in an ANN (Artificial Neural Network) model reproduced the observed TWS (Terrestrial Water Storage) and groundwater storage anomaly at local scales. This downscaled product also replicated the natural water storage variability due to climatic and human impacts. Finally, the relative impact between humans vs. climate variability was distinguished and measured in Ethiopia using an integrated approach that can be transferable to similar settings. The implications utilizing satellite data for improving local and regional water resources management decisions and applications are clear. This is especially true with areas lacking hydrologic monitoring networks.