Mapping forest microclimates using Sentinel 1 Radar data: insights and complementarity with LiDAR and optical data
Résumé
Forest ecosystems are well known to provide microrefugia for many species in the context of climate change, due to their buffered and more stabilized microclimatic conditions compared with macroclimatic fluctuations outside forests. Forest structure is a major driver of microclimate resulting in spatial heterogeneity in managed forests. Spatializing the buffering effect of forest canopies on understory temperatures requires spatially contiguous predictors, which can be derived from remote sensing. ALS and satellite optical data have already proven their interest to map forest microclimates. However, no study has yet been tested the potential of SAR and especially Sentinel 1 data that allow intra- and inter-year spatiotemporal analyses of forest structure. The objective of this study is therefore to test the potential of SAR, combined with ALS and optical data, to map forest buffering capacity. We derived metrics describing both the vertical and horizontal heterogeneity of the canopy over the year 2021 using Sentinel 1 and 2 time series and ALS data that cover the French state forest of Blois. These metrics were then related to understory microclimate data collected across 52 plots during the leaf-on period in 2021. Model selection was performed to identify the best metrics to predict microclimate. Finally, the best model was used to map temperatures across the whole forest. Results showed that SAR-derived metrics performed well combined with ALS and optical data, with models explaining up to 90% of variance in microclimate effect. This study highlights the potential of SAR data and its complementarity, especially with optical data, to map forest thermal environment at fine spatiotemporal resolutions.