Satellite-derived inland lake dynamics : mapping, recovery, and water clarity impacts

Abstract

Lakes are essential components of the hydrological and biogeochemical cycles and are undergoing rapid changes worldwide in response to climate change and human activities, intensifying the water stress in many regions. Hence, it’s of great significance to explore the potential causes and consequences of the lake expansion/shrinkage to alleviate the widespread water crisis, which firstly requires accurate monitoring of the lake area changes. This thesis attempts to alleviate two major issues that hamper continuous and long-term lake area monitoring through satellite observations, including lake identification and image contamination. On such basis, this thesis also aims to explore what can be revealed in terms of the spatiotemporal lake variation patterns and the main drivers for such variations. The water quantity is further associated with water quality analysis by exploring the potential impacts of the lake area variations on water clarity changes in a geographically important region under climate change.

Accurate delineation of lake boundaries is the premise for long-term lake area monitoring. Here a new global lake dataset called GLAKES is developed, constituting 3.4 million lakes with maximum extent > 0.03 km2. Based on GLAKES, we discover a net lake area increase across all six continents analyzed during the past four decades, and identify the human regulation instead of climate warming as the most essential drivers for such variations. Besides, the identified lake area increase over time leads to higher lacustrine carbon emissions. Our findings also illustrate the emerging roles of small lakes in regulating not only local inland water variability, but also the global trends of surface water extent and carbon emissions.

Remote sensing images are frequently contaminated by clouds and other unfavorable conditions. Here an enhanced water occurrence threshold (EWOT) based reconstruction algorithm is proposed to alleviate such issue. This algorithm utilizes long-term historical water occurrence information to recover water pixels from contaminated images, and exhibits better performance than a classic WOT approach. Application on China fills the gaps of the current lake studies in this region through the first provision of the annual-based lake area time series record for 4936 lakes. We discover a substantial expansion of both the natural lakes and reservoirs between 2000 and 2019, along with profound intra-annual variations, despite divergent trends observed spatially.

Substantial lake expansion can have strong impact on lake water clarity under rapid climate change, which is especially the case for the Third Pole region. Here a MODIS-based Secchi disk depth (Zsd) retrieval model is developed and systematic spatiotemporal variation patterns of Zsd for 64 large alpine lakes in this region are first analyzed during the 2003-2018 period. Relevant analysis has also identified the lake expansion as the most important driver of the water clarity increase in this region.

Overall, this thesis has significant implications for the long-term continuous lake area monitoring and uncovers new scientific findings in this field. It also indicates the significance of incorporating both the water quantity and water quality into the comprehensive evaluation and management of inland freshwater resources.