Improved Fire Activity Time Series by Accounting for Variation in Viewing Geometry

Robust spatially and temporally resolved observations of fire activity are essential to prediction of the important air quality and climate effects of biomass burning smoke. Satellite detection of fires with the coverage and frequency necessary for regional and global monitoring relies on weather-based satellite sensors with limited sensitivity to fires. This sensitivity varies as a function of viewing geometry, which results in a systematic artifact in fire detection time series data generated from polar orbiting satellites. We demonstrate how this variability can be characterized and corrected. A corrected time series is generated with our methods to provide a spatial and temporal description of fire activity that improves correlation with higher-resolution observations. This method is applied to MODIS and VIIRS active fire detections, and evaluated by comparison with geostationary fire detection products and other data. The implications of our results for effective combination of polar and geostationary fire detections are also described.