Evaluation of an agricultural area affected by a vegetation fire using remote sensing techniques

This study assesses the impact of a wildfire on an agricultural area using remote sensing techniques. Satellite images (Sentinel 2) were taken before the wildfire (bf) and after the wildfire (af). The indices determined before the fire (NDVI-bf, MSAVI-bf and NBR-bf) and the indices determined after the vegetation fire (NDVI-af, MSAVI-af and NBR-af) were analyzed comparatively. The index values ​​showed low variability before the wildfire, and moderate variability (CVNDVI-af = 25.74; CVMSAVI-af = 24.10), respectively high variability (CVNBR-af = 46.66) after the wildfire. The applied tests confirmed significant differences between the mean values ​​(t Test) and median values ​​(Wilcoxon Test) of the indices after the fire and before the vegetation fire (p<0.001). The calculated ratio of the indices presented sub-unit values, respectively NDVI-af/NDVI-bf = 0.406, MSAVIaf/MSAVI-bf = 0.433, respectively NBR-af/NBR-bf = 0.388. Dunn’s Post Hoc test confirmed significant differences between the indices at the evaluation times, before and after the fire. According to the recorded value (DNBR = 0.3332), the severity of the analyzed vegetation fire was classified as Moderate-low Severity. A mathematical model of quadratic regression and graphical models (3D, isoquants) described the variation of the NBR index in relation to the NDVI and MSAVI indices, in conditions of statistical safety. The results of the study are important for the promotion of remote sensing techniques in agricultural practice, as a useful tool in management decisions for agricultural areas affected by vegetation fires.