Soil moisture and NDVI as factors to assess fire susceptibility – A case study in Greece

Forest fires in the Mediterranean region are becoming more frequent and more destructive. The areas affected show an upward trend, with the fire season extending from the early months of spring to the end of autumn and tending to affect areas at higher latitudes. These concerning findings highlight the need to develop susceptibility assessment methods capable of identifying areas vulnerable to forest fires. This study aims to detect potential precursory warning signals from soil moisture (SM) and vegetation productivity (NDVI) by examining 11 extensive and destructive fires that occurred in Greece during the years 2021, 2022, and 2023. We relied on previous studies suggesting that SM and NDVI indices are suitable variables to indicate alarming changes in spatial and temporal soil and vegetation conditions leading to wildfire occurrence. The potential precursor signal of SM was assessed using the Soil Moisture Active Passive (SMAP) L4 dataset as a wildfire occurrence factor. SMAP SM was compared to in-situ SM measurements in NE Greece and correlation values of r=0.78 were found, indicating the agreement between ground measurements and satellite SMAP L4 values. In addition, the USGS Landsat 8 data with a spatial resolution of 30 m was used to assess monthly NDVI values during the study period, which represent vegetation conditions. Both SM and NDVI time series were evaluated over a period of approximately two years prior to each wildfire event. The results of the analysis identified specific characteristics of the variables studied, which ultimately delineated areas sensitive to the onset of forest fires. NDVI values in all fire-affected areas ranged from 0.13 to 0.35, while mean monthly SM consistently showed negative anomalies in the preceding to wildfire occurrence spring, for all fire cases studied. With respect to NDVI, no specific temporal trends were observed in the affected areas, except for the expected seasonal variations during the pre-burn period. Finally, fire susceptibility maps were produced for the whole country, verifying the usefulness of remote sensing data in terms of SM and NDVI. All the results of this specific research, if properly used by the relevant authorities, can serve as a significant tool in the prevention of forest fires.