Global spatiotemporal dynamics of inland water body storage during the satellite altimetry era

Abstract

Lakes and reservoirs are essential components of the global hydrological cycle. They function as sentinels of climate change, and provide indispensable water supply, energy generation, aquatic ecosystems for meeting agricultural, industrial, and domestic needs. Despite the importance of lakes and reservoirs to natural systems and human societies, their water storage dynamics are poorly understood on a global scale, especially in remote and harsh environments such as the Tibetan Plateau. The challenge has been even amplified by the currently declining global network of gauging measurements. This knowledge gap prevents a fundamental understanding of the terrestrial water cycle and surface water availability, and thus impedes effective water resource management. Recent regional evidence has suggested that water-abundant regions show increasing trends in surface water extents whereas water-limited regions exhibit decreasing trends. However, whether this divergence also holds true for the storage of open water bodies across the entire land surface remains unknown. A synergy of fine resolution optical imagery and level measurements by satellite altimetry has the potential to accurately deduce lake/reservoir storage variations. However, the existing estimates of lake/reservoir storage changes are mostly limited by the number/size of studied water bodies and the low temporal frequencies in time series, primarily owing to atmospheric contaminations in satellite imagery and inadequate spatiotemporal coverage in altimetry. This dissertation aims to mitigate the above technical limitations through several remote sensing method advancements, and to leverage multi-mission satellites to investigate the recent lake/reservoir storage dynamics and their implications for the regional and global water cycles. Specifically, this dissertation is orientated to three overarching questions: 1) how has the lake water storage across the landlocked inner Tibetan Plateau changed over the recent decades and what was the dominant driver? 2) Can we improve the methodology of lake/reservoir area time series mapping in order to better enable a temporarily consistent monitoring of water storage dynamics in the world? 3) Did the recent trends in global open-surface water storage exhibit a divergence between arid and humid regions, and if so, what does this divergence indicate for the global water cycle and water management? The above questions were addressed in three chapters. Chapter 2 provides a comprehensive and spatially explicit quantification of lake storage changes across the inner Tibetan Plateau since 2002, through a synergy of satellite imagery and freely-available digital elevation models. By further incorporating satellite gravity observations to a water balance model, the dominant driver of this regional lake storage variation was attributed to net precipitation, i.e., precipitation minus evapotranspiration, rather than glacial melting. Chapter 3 introduces a novel method that substantially improves the temporal coverage in lake/reservoir area mapping, by effectively recovering inundation areas from contaminated spectral images with a relative error of 2.2%. Inclusion of recovered water areas from contaminated images improved the monthly coverage for 400+ lakes and reservoirs in the world by an average of 43%. Chapter 4 combines this novel mapping method and a constellation of satellite altimeters to reveal monthly storage changes in 1000+ major lakes/reservoirs worldwide from 1992 to 2018, the entire satellite altimetry era thus far. The finding confirms a multi-decadal divergence in global open-surface water storage between endorheic (arid/semiarid) and exorheic (more humid) regions. This divergence indicates an amplified contrast between fresh and saline water abundance, and was partially induced by the increasing water impoundment behind dams. Due to the divergence, global lakes and reservoirs slowed down the observed sea level rise by only ~1% during the past nearly three decades.