Assessment of the robust satellite technique (RST) in real time detection of summer fires

Robust Satellite Technique (RST) has already demonstrated to be a suitable change detection scheme; recently it has been applied and tested, in a pre-operational way, in the context of forest fire monitoring. A Total Validation Approach (TVA) was experimented during a campaign - performed last year in Sicily (Southern Italy), in collaboration with the local Civil Protection authorities - for the assessment of RST capabilities in detecting forest fires. In this paper, preliminary results of the campaign (which is still ongoing) are presented focusing on the assessment of the RST capability to timely detect small fires. The 2008 Sicily campaign, performed by applying the RST scheme to images collected by 3 different satellite sensors onboard polar (NOAA-AVHRR, EOS-MODIS) and geostationary (MSG-SEVIRI) platforms, confirmed the capability of RST in detecting fires at a very early stages. In particular, by using geostationary sensors, it was successfully tested the capability of timely (within 15 minutes) detection of very small (less than 30m2) fires. Finally, thanks to direct observations by ground and by light aircrafts it was possible to perform a TVA controlling each individual RST detected thermal anomalies. Several of them - that traditional a-posteriori analyses would flag as “false alarms”, because not recorded by Forestry Service in their registers - were instead demonstrated to be related to actual fires (burning stubbles or other very small fires) as well as to other, actually existing, hot sources like chimney of industrial plants with variable thermal emission regimes. This confirmed the importance of a TVA for a reliable assessment of whatever satellite-based fire detection method.