Influence of land surface temperatures, precipitation, total water storage anomaly and fraction of absorbed photosynthetically active radiation anomaly, obtained from MODIS, IMERG and GRACE satellite products on wildfires in eastern Central Italy

Forest fires are increasingly frequent and pose a risk to the entire ecosystem and also to subsequent hydrogeological risks that may be amplified. Italy and the Mediterranean area are increasingly affected by fires and in recent years they have dried up abruptly due to extreme climatic conditions that increase the number of continuous days without rainfall, as well as the duration and intensity of heat waves that increase the risk of wildfires. In this context, it is essential to implement countermeasures in order to better plan the territory, by means of monitoring tools that can guarantee optimal coverage of the areas under investigation, assessing risk conditions. This research was carried out using remote sensing products such as Moderate-resolution Imaging Spectroradiometer (MODIS) and Integrated Multi-satellitE Retrievals for GPM (IMERG) by recording daily values of land surface temperature (LST), precipitation, total water storage anomaly (TWSA) and the fraction of absorbed photosynthetically active radiation anomaly (FAPAR), to explain the annual variance in the number of fires and the amount of area affected by fire in eastern Central Italy. The statistical technique adopted was multiple linear regression (MLR), which identified the most influential variables in defining the annual number of fires, Summer LST showed a partial correlation values of 0.85, followed by precipitation in April (-0.07) and November (-0.38), June TWSA (−0.68), March FAPAR (0.78) and June FAPAR (−0.12), in addition to a low value of collinearity between the variables. The model obtained with MLR resulted in an 84% explanation of variance, a result inferred from the adjusted R-square. For the burned area, the same variables were involved but produced a different outcome, explaining 60% of the variance. This suggests potential for future predictive scenarios using more suitable variables to assess fire spread.