Beschreibung:
In Mediterranean forest ecosystems affected by severe fires, a fine-scale spatial evaluation of soil status can provide relevant information on ecosystem recovery. In the present study, we evaluated the potentiality of combining multispectral imagery with different spectral and spatial resolutions to estimate soil properties sensitive to burn severity. We conducted a study in a heterogeneous area located in the northwest (NW) Iberian Peninsula (Spain) and affected by a wildfire occurred in August 2019. Soil burn severity was measured one month after fire in 34 field plots of 50 cm × 50 cm, using an adapted protocol of the Composite Burn Index (CBI). In each plot, we collected one soil sample at 0-3 cm depth to analyse three soil properties potentially sensitive to fire: mean weight diameter (MWD), soil moisture content (SMC) and soil organic carbon (SOC). Then, post-fire imagery was collected from the Sentinel-2A MSI satellite sensor (10 and 20 m spatial resolution), as well as from a Parrot Sequoia on board a UAV (spatial resolution of 0.50 m). A Gram-Schmidt (GS) image sharpening technique was used to increase the spatial resolution of Sentinel-2 bands from 20 m up to 10 m and to fuse these data to UAV information at 0.50 m. We developed a machine learning decision tree to determine the sensitivity of soil parameters to discriminate soil burn severity categories. The relationship between the most fire-sensitive soil properties and the reflectance values (UAV, Sentinel-2 and fused UAV-Sentinel-2 images) was analysed by means of support vector machine (SVM) regression models. All the considered soil parameters decreased their value with severity, but SMC, and to a lesser extent SOC, discriminated at best among soil burn severity classes (overall accuracy of 0.77; Kappa= 0.55). Reflectance values derived from the fused UAV-Sentinel-2A images were the finest predictors of SOC content (R 2 = 0.52; RMSE= 7.83). This study highlights the advantages of combining satellite and UAV images to produce spatially and spectrally enhanced images useful for estimating main soil properties in heterogeneous burned areas where emergency actions need to be applied