Close Window

A multi-platform technique to validate active fire detections from the GOES Imager

Ivan Csiszar, University of Maryland, Department of Geography, (Presenting)
Wilfrid Schroeder, University of Maryland, Department of Geography,
Jeffrey Morisette, NASA Goddard Space Flight Center,
Elaine Prins, Consultant in Environmental Remote Sensing Applications,
Christopher Schmidt, Cooperative Insitute for Meteorological Satellite Studies,

One of the goals of the LBA-ECO Phase III study LC-35 is to evaluate fire detections from the GOES Imager as part of the generation of a fused multi-sensor active fire product in the Amazon. The principles of the validation procedure used in this study were developed and tested in the Amazon within the Phase II project LC-23 for MODIS and ASTER on the Terra satellite. For the GOES Imager, the burning and non-burning areas within the pixel were mapped using near-coincident fire masks from Terra/ASTER and Landsat-7/ETM+ and detection probabilities were derived as a function of summary statistics of the high resolution fire pixels. To evaluate the impact of the time difference between the GOES and high resolution fire observations, we estimated the short-term change of the summary statistics at the scale of the GOES pixel, using pairs of fire masks from same-day imagery 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 and did not impact substantially the derived detection probabilities. This configuration of geostationary and polar data also allows for the evaluation of the scan angle effects of the detection performance. However, as ASTER and ETM+ fly on sun-synchronous platforms, further sensors will need to be considered to cover a fuller range of the local time of fire observations.

Science Theme:  LC (Land Use and Land Cover Change)

Presentation Type:  Poster

Abstract ID: 54

Close Window