Close Window

Towards multi-platform validation of active fire products from moderate resolution sensors in the Amazon

Ivan Csiszar, University of Maryland, Department of Geography, icsiszar@hermes.geog.umd.edu (Presenting)
Wilfrid Schroeder, University of Maryland, Department of Geography, schroeder@hermes.geog.umd.edu
Jeffrey Morisette, NASA Goddard Space Flight Center, Jeff.Morisette@nasa.gov
Elaine Prins, Consultant in Environmental Remote Sensing Applications, elaine.prins@ssec.wisc.edu
Christopher Schmidt, Cooperative Insitute for Meteorological Satellite Studies, chris.schmidt@ssec.wisc.edu

Validation of active fire products from moderate resolution sensors, such as MODIS, AVHRR or GOES Imager, requires the mapping of burning, smoldering and unburnt areas within the pixel using independent information. This can be achieved by using coincident observations from higher resolution space borne sensors, even if their band configuration is typically suboptimal for active fire mapping. Thirty meter resolution ASTER data on board the Terra satellite provide an optimum sampling configuration for the validation of the 1km Terra/MODIS active fire products, but only for a narrow range of near-nadir scan angles. Sensors from alternative platforms provide the potential to sample a wider array of observing geometries of Terra/MODIS or other moderate resolution sensors, but at the expense of a time difference between the two observations. To estimate the temporal change in summary fire statistics at the scale of the MODIS pixels, used as input parameters for validation, we analyzed 10 pairs of fire masks from same-day imagery over the Amazon from ETM+ and ASTER, flown on Landsat-7 and Terra ~ 25 minutes apart. We found that, while the progression of the fire front was observable, in some biomes the change in summary statistics was small. The sensitivity of the validation results to the input fire masks was analyzed by comparing MODIS fire detection probabilities as a function of ASTER- and ETM+ -derived summary statistics. This work is part of a new LBA-ECO Phase III study, where multi-platform validation techniques will also be used to evaluate fire detections from the GOES Imager.

Presentation Type:  Poster

Abstract ID: 28

Close Window