Towards an Enhanced Active Fire Product from VIIRS

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 19 January 2010: 3:30 PM
B313 (GWCC)
Ivan Csiszar, NOAA/NESDIS, Camp Springs, MD; and E. Ellicott, L. Giglio, and C. O. Justice

The VIIRS instrument on board the NPP and NPOESS satellites will provide radiometric measurements that offer useful information for the detection and characterization of active fires. The current baseline algorithm for the Fire Application Related Product uses primarily the moderate resolution 4 Ám band (M13) and the 11 Ám band (M15) measurements, aggregated from the native resolution observations into pixels according to a scheme aimed at maintaining a near-constant spatial resolution along the scanline. In this work, we explore the potential of developing an enhanced active fire product, taking full advantage of the native spatial resolution and radiometric information provided by the VIIRS sensor. We aim to characterize uncertainties resulting from the on-board aggregation of potentially saturated native pixel data, and thus minimize ambiguity. The work includes the analysis of simulated VIIRS radiances for hypothetical and real fires. Initial analysis of simulated hypothetical VIIRS radiances indicates an approximately two-fold decrease in the size of minimum detectable size from unaggregated data for typical fire temperatures at nadir. For real fires, we use MODIS and coincident 30m fire observations from ASTER to generate accurate VIIRS-like scenes in the presence of active fires. For non-fire VIIRS pixels spectral information from ASTER and MODIS is fed into spectral and geometrical transformation algorithms, using existing VIIRS SDR simulation schemes. For VIIRS pixels containing fire, radiances are directly simulated using the fire characteristics estimated from the ASTER data, taking into account the VIIRS sampling geometry. For algorithm development and testing, this procedure is performed for representative samples of different fire regimes. A VIIRS-specific contextual detection algorithm is run over the simulated VIIRS scenes and algorithm performance is evaluated.