Global glacier mass change by spatiotemporal analysis of digital elevation models

Overview & Synopsis

The world's glaciers are shrinking rapidly, with impacts ranging from global sea-level rise and changes in freshwater availability to the alteration of cryospheric hazards. Despite significant advances during the satellite era, the monitoring of the mass changes of glaciers is still hampered by a fragmented coverage of remote sensing estimations and a poor constraint of the errors in related assessments. In this thesis, we present a globally complete and resolved estimate of glacier mass changes by spatiotemporal analysis of digital elevation models. We first develop methods based on spatiotemporal statistics to assess the accuracy and precision of digital elevation models, and to estimate time series of glacier surface elevation. In particular, we introduce a non-stationary spatial framework to estimate and propagate multi-scale spatial correlations in uncertainties of geospatial estimates. We then massively generate digital elevation models from two decades of stereo optical archives covering glaciers worldwide. From those, we estimate time series of surface elevation for all of Earth's glaciers at a resolution of 100,m during 2000--2019. Integrating these time series into volume and mass changes, we identify a significant acceleration of global glacier mass loss, as well as regionally-contrasted responses that mirror decadal changes in climatic conditions. Using a large amount of independent, high-precision data, we demonstrate the validity of our analysis to yield robust and consistent uncertainties at different scales of the spatiotemporal structure of our estimates. We expect our methods to foster robust spatiotemporal analyses, in particular to identify sources of biases and uncertainties in geospatial assessments. Furthermore, we anticipate our estimates to advance the understanding of the drivers that govern glacier change, and to extend our capabilities of predicting these changes at all scales. Such predictions are critically needed to design adaptive policies on the mitigation of cryospheric impacts in the context of climate change.