Assessment of Aeolian Activity in the Bodélé Depression, Chad: A Dense Spatiotemporal Time Series From Landsat-8 and Sentinel-2 Data

Abstract

There are several hotspots of dust production in the central Sahara, the Bodélé Depression (BD) in northern Chad is considered the largest source of aerosol dust worldwide, with the fastest Barchan dunes that migrate southwesterly. Less is known about the complex patterns of dune movement in the BD, especially on a short time scale. Time-series inversion of optical image cross-correlation (TSI-OICC) proved to be a valuable method for monitoring historical movements with low uncertainties, high spatial coverage, and dense temporal coverage. We leveraged ∼8 years of Landsat-8 and ∼6 years of Sentinel-2 data to capture the dune migration patterns at BD. We used TSI-OICC, creating four independent networks of offset maps from Landsat-8 and Sentinel-2 images, and forming three networks by fusing data from the two sensors. We depended on the multi spatial coherence estimated from Sentinel-1 interferograms to automatically discriminate between the active and stagnant regions, which is important for the postprocessing steps. We combined the data from the two sensors in areas of overlap to assess the performance of the fusion between two sensors in increasing the temporal scale of the observations. Our results suggest that dune migration at BD is subject to seasonal and multiyear variations that differed spatially across the dune field. Seasonal variations were observed with migration slowing during the summer months. We estimated the median for velocities belonging to the same season and calculated the seasonal sliding coefficient (SSC) representing the ratio between seasonal velocities. The median SSC reached a maximum value of ∼2 for winter/summer, while the ratios were ∼1.10 and ∼1.35 for winter/spring and winter/autumn, respectively. The seasonal variability of the temporal patterns was strongly supported by the wind observations. Between (1984–1998 and 1998–2007) and (1998–2007 and 2013–2021), decelerations in dune velocities were observed with percentages of ∼4 and ∼28%, respectively, and these decelerations were supported by a deceleration in wind velocities. Inversion of time series provides dense spatiotemporal monitoring of the dune activity. The fusion between two sensors allows condensing the temporal sampling up to a weekly scale especially for locations exposed to contamination of high cloud cover or dust.