Investigating the impact of overlying vegetation canopy structures on fire radiative power (FRP) retrieval through simulation and measurement

Fire radiative power (FRP) retrievals are now routinely made from polar and geostationary instruments, providing a means to estimate fuel consumption and trace gas and aerosol emissions directly from remotely sensed observations. This study presents the first investigation of the impact of vegetation canopy structure (percentage canopy cover and leaf area index, LAI) on FRP retrievals, based on 3D radiative transfer model simulations. The Discrete Anisotropic Radiative Transfer (DART) model is used to simulate above-canopy observations made through 3D vegetation canopies with different structural arrangements, under which a centrally positioned uniform landscape fire is burning. The vegetation canopy is modelled in two ways, as an opaque structure and as a hybrid turbid medium. The percentage canopy cover in each simulated scene is varied between 5 and 95%, and the FRP retrieved above the canopy is found to decrease in proportion to percentage canopy cover when the canopy is opaque, a finding that is in agreement with a series of small scale outdoor measurements conducted to evaluate the realism of the simulations. However, when the canopy is modelled as a turbid medium, which is in some ways a more realistic representation of a real 'gappy' vegetation canopy, the degree of FRP interception occurring at any particular canopy cover decreases by ~ 14%, due to some fire emitted thermal energy being transmitted through the canopy gaps. The simulations also reveal the impact of canopy LAI on above-canopy FRP retrievals, reducing these by 6% when both canopy cover and LAI are low (5% and < 1.0 respectively), but by up to 92% when canopy cover and scene LAI are high (95% and ~8 respectively). We use the derived relationships between FRP interception and canopy structure, along with MODIS LAI and percentage tree cover data, to adjust 2004-2012 fire radiative energy (FRE) estimates calculated from FRP data collected by the geostationary Meteosat Spinning Enhanced Visible and Infrared Imager (SEVIRI) instrument. The adjusted annual FRE is on average 15% greater than estimated, and is characterized by low inter-annual variability as result of the majority of fire activity occurring in areas where percentage tree cover remains below 40%. Landscape burning occurs more frequently in areas of higher tree cover in southern hemisphere rather than northern hemisphere Africa, leading to a larger annual FRE adjustment (18.5% compared to 16.3%). This study illustrates the impact that canopy interception has on FRP for the first time at the satellite scale, and over Africa demonstrates a large but temporally consistent underestimation which can be accounted for using LAI and percentage tree cover metrics when estimating fuel consumption and atmospheric emissions from the FRP retrievals.