Using SEVIRI Geostationary Imagery for Active Fires Analysis and Burned Biomass Estimation: a Case Study in a Mediterranean Ecosystem

Fire is a frequent phenomenon in the Mediterranean region mostly caused by human activities. Every year several hundred thousands hectares are burned throughout European forests with consequences on ecosystems and local air pollution. Although being an important issue, there is still high uncertainty about global and regional estimates of biomass burning emissions. Current methods use burned biomass and emission factors to quantify the amount of trace gases released. Each parameter depends on ecosystem, fire type and burning conditions, making this phenomenon particularly difficult to assess. Because of the high spatio-temporal variability burned biomass is the most problematic factor. In order to improve current estimates of this parameter a new approach has been developed with successful application in southern Africa. It uses the radiant energy released during burning activity (Fire Radiative Energy, FRE) to retrieve the amount of burned biomass. In this study we want to assess the applicability of FRE to Mediterranean ecosystem using the geostationary satellite sensor SEVIRI (Spinning Enhanced Visible and Infrared Imager). Despite having a coarse spatial resolution, the instrument has essential characteristics which allow active fires observation and analysis. These characteristics are the high temporal resolution, with an acquisition time of 15 minutes, and a suitable spectral domain which includes infrared (Medium and Thermal) which is important for fire detection and FRE computation. A forest fire has been chosen as a case study to apply the FRE methodology and assess the derived results. The fire occurred in Italy in February 2005 affecting an area of 1600 hectares. In order to compare remotely sensed results of FRE, a field campaign has been carried out on the study area to collect ground data for burned biomass derivation During the field work biomass has been measured both on the burned area and on non-burned areas of the surroundings which showed similar characteristics to the fire affected site (i.e. vegetation type). The amount of burned biomass is derived as the difference between the biomass load in the non-burned area and the amount of unburned biomass on the fire affected area. Results show that more than 50% of the trees biomass has been consumed during burning. The fire detection capability of SEVIRI sensor is then assessed and results shown. Finally FRE methodology is derived from satellite images. Here preliminary results about burned biomass estimations from FRE application are shown. Future work will be the comparison of RS-derived data with burned biomass estimated from field measurements.JRC.H.7-Land management and natural hazard