Spatiotemporal change of water level in ungauged high-altitude tropical lakes: a DInSAR approach

Abstract

<p>Spatiotemporal characteristics of physical responses of lakes to external and environmental changes are still largely unknown due to the consistent lack of monitoring of water level and corresponding changes in water storage in lakes. Understanding these changes is a fundamental step in advancing regional management of natural and anthropogenic systems that depend on the water resources of lakes. As an illustrative example, we here report a case study involving lakes of the headwater topical Andes mountain range, which, despite guaranteeing water security to millions of downstream inhabitants, still remain significantly ungauged. We present a novel evaluation of the potential of Differential Interferometric Synthetic Aperture Radar DInSAR techniques for the spatiotemporal analysis of patterns of water level change in lakes such as the ones comprising these ungauged high-altitude lake systems. Time series of Sentinel-1B data for the years 2017 and 2018 were used to generate continuous interferograms representing water level changes in twenty-four lakes of the Cajas National Park, Ecuador. The relation of these water level changes with climatic and topographical factors were analyzed to validate the methodology, and determine any patterns of change and response to climatic drivers. We found relatively high Pearson correlation coefficients between regional precipitation and water level change as estimated from the interferograms. Furthermore, we found an important negative relationship between water level change, as obtained from the DInSAR phase, and lake surface area. The study revealed a spatial trend of this correlation in terms of the altitude of the lakes at the basin scale; that is, lower correlation values were found in the headers of the basins, whilst higher correlation values were found at lower basin altitudes. The results of the present study demonstrate the potential of DInSAR techniques based on Sentinel-1 data for the monitoring of hydrologic changes in open water surfaces, and the possible validation of the DInSAR results with precipitation when gauged water level data is missing. These results are a basis to propose monitoring strategies in ungauged high-altitude lake systems in regions with similar data gauging constraints. Future work will encompass the integration of ongoing water level gauging for further validation of the herein depicted lake water level estimation approach.</p>