Recognition of whole-landscape changes due to extreme rain events in a hyperarid desert

Abstract

Infrequent extreme rain events in extraordinarily dry environments can dramatically shape the landscape. Here, a change-detection method is demonstrated that uses Landsat 8 OLI scenes to document land surface modification across thousands of square kilometers due to two extreme rain events which impacted the Atacama Desert. Images for 4 dates, in 3 date-pairs, are analyzed using a Change Vector Analysis (CVA) technique which delivers a measure of the intensity of change (Intensity-CVA). In 2015, an extreme rain event affected the southern Atacama and one in 2019 affected the northern Atacama. In this work, we show with the 2015 event the efficacy of CVA to identify areas of mass transference by favorable comparison to published Interferometric Synthetic Aperture Radar (InSAR) permanent coherence loss for areas located in nearly flat terrains. The efficacy of Intensity-CVA was also demonstrated for the 2019 rain event, where it identified mass transference and mineral transformations. To clarify the physical processes behind the Intensity-CVA response, for specific areas with known types of surface materials we conducted Principal Component Analysis (PCA) on quadratic difference of Surface Reflection (SR) values of Landsat 8 OLI scenes for multiple dates and also analyzed Intensity-CVA trends. These two analyses reveal that, among locations with strong Intensity-CVA change, remotely identified changes are dominated by soil moisture variations associated with active alluvial deposits, playa lakes and evaporitic units verified by PCA even 60 Km away from the focus of the rain. Secondarily we detected salt crystals growth several months after the rain. This approach, validated in the Atacama, successfully identified the consequences of extreme rain events at the scale of the entire landscape. This result opens the possibility to use a long-lived sensor, such as Landsat, to examine geomorphological surface changes through time in arid environments in a context of climate change.