IDENTIFYING LANDSCAPE CHANGES DUE TO EXTREME RAIN EVENTS IN THE ATACAMA DESERT AND THEIR GEOMORPHIC CONTROLS: INSIGHTS FROM UPPER RIO SALADO AND SIPUCA WATERSHEDS

In the Atacama Desert, one of Earth’s driest places, the extremely slow pace of change makes it difficult to evaluate the processes of changes. Here, we analyze the impacts of two extreme rain events, focused on the erosional sectors of two Atacama watersheds. The upper Rio Salado (uRS) watershed was affected by the March 2015 event and the Quebrada Sipuca (QS) watershed was affected by the January 2019 event. The geomorphological literature anticipates spatial variability of the degree of landscape change according to slope, altitude, flow accumulation, Strahler order, stream gradient and M-Chi. The trends of change in these parameters should correlate with the hydraulic behavior of the watershed, and hence with erosion or deposition. To identify the landscape changes, we used intensity Change Vector Analysis (iCVA) applied to a pair of Landsat 8 OLI images taken a year apart. The most important changes recognized are in the floors of valleys. Each landscape parameter was separated into 10 different classes based on the value of each pixel. Each class provided a spatial mask of the iCVA and then we analyzed the upper 20% of iCVA values. The only two parameters that presented similar behaviors between watersheds were the Strahler order and the altitude, showing that most of the change happens at low elevations and that changes are more intense in the main branches of the stream. The latter result agrees with published field observations reporting the impacts along the upper Rio Salado. The other parameters have opposite trends or are invariant to the iCVA, such as M-Chi. This may be due to the geomorphological differences of the basins, or spatially variable rainfall, or different type of change such as deposition and erosion. In the future, we will explore InSAR and differential DEM approaches to characterize changes in these and other watersheds. We anticipate that more statistical approaches will help to identify the main drivers of the landscape change.