Meteosat SEVIRI Fire Radiative Power (FRP) Products from the

4 Wooster, M.J., Roberts, G., Freeborn, P. H., Xu, W., Govaerts, Y., Beeby, R., 5 He, J. , A. Lattanzio, Fisher, D., and Mullen, R. 6 7 8 1 King’s College London, Environmental Monitoring and Modelling Research Group, 9 Department of Geography , Strand, London, WC2R 2LS, UK. 10 2 NERC National Centre for Earth Observation (NCEO), UK. 11 3 Geography and Environment, University of Southampton, Highfield, Southampton 12 SO17 1BJ, UK. 13 4 Fire Sciences Laboratory, Rocky Mountain Research Station, U.S. Forest Service, 14 Missoula, Montana, USA. 15 5 Rayference, Brussels, Belgium. 16 6 MakaluMedia, Darmstadt, Germany 17 18 19 Abstract 20 Characterising changes in landscape fire activity at better than hourly temporal 21 resolution is achievable using thermal observations of actively burning fires made 22 from geostationary Earth observation (EO) satellites. Over the last decade or more, a 23 series of research and/or operational 'active fire' products have been developed from 24 geostationary EO data, often with the aim of supporting biomass burning fuel 25 consumption and trace gas and aerosol emission calculations. Such "Fire Radiative 26 Power" (FRP) products are generated operationally from Meteosat by the Land 27 Surface Analysis Satellite Applications Facility (LSA SAF), and are available freely 28 every 15 minutes in both near real-time and archived form. These products map the 29 location of actively burning fires and characterise their rates of thermal radiative 30 energy release (fire radiative power; FRP), which is believed proportional to rates of 31 biomass consumption and smoke emission. The FRP-PIXEL Product contains the full 32 spatio-temporal resolution FRP dataset derivable from the SEVIRI imager onboard 33 Meteosat at a 3 km spatial sampling distance (decreasing away from the west African 34 sub-satellite point), whilst the FRP-GRID product is an hourly summary at 5° grid 35 resolution that includes simple bias adjustments for meteorological cloud cover and 36 regional underestimation of FRP caused primarily by under-detection of low FRP 37 fires. Here we describe the enhanced geostationary Fire Thermal Anomaly (FTA) 38 detection algorithm used to deliver these products, and detail the methods used 39 generate the atmospherically corrected FRP and per-pixel uncertainty metrics. Using 40 SEVIRI scene simulations and real SEVIRI data, including from a period of 41 Meteosat-8 'special operations', we describe certain sensor and data pre-processing 42 characteristics that influence SEVIRI's active fire detection and FRP measurement 43