Unprecedented remote sensing data products from before and after the Rim Fire, Sierra Nevada, California

Background/Question/Methods

In August 2013, a major fire in Yosemite National Park and the Stanislaus National Forest burned 104,131 hectares (257,314 acres) before extinction on October 24^th, 2013 (http://inciweb.nwcg.gov/incident/3660/). One of the unique attributes of this fire is the unprecedented amount of before and after remotely sensed data available particularly at fine spatial (~3-30 m) and spectral resolutions spanning the electromagnetic spectrum from visible to thermal infrared. This work presents analyses of processed data products for these unique remote sensing datasets. In particular, these data provide a unique opportunity to address questions such as: how do satellite broadband and airborne narrowband data compare? Can we improve understanding of fire-fuel dynamics by cross-referencing spectral data of fuel condition, type, amount, and moisture content to that of LiDAR, which provides accurate estimates of fuel type and structure?

We employ data from four airborne campaigns including: hyperspectral sensors (1) Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) and (2) MODIS/ASTER Airborne Simulator (MASTER) which cover nearly 90 percent of the Rim Fire before ignition and full coverage post-fire, the (3) Airborne Snow Observatory (ASO), which utilizes LiDAR and an imaging spectrometer to produce data products useful for determining when, where and the amount of runoff from snowpack, and lastly (4) high resolution (~1m) resolution LiDAR that covers 5,000 hectare pre-ignition and the full fire including a two kilometer buffer after the fire. We present methodologies for processing atmospherically and topographically corrected data products such as surface reflectance and indices. Additionally, we compare the ability of an index to distinguish between burned and unburned areas using a separability metric, and how spectral data compare to LiDAR data.

Results/Conclusions

In general, many indices have high separability (M>1), which is probably because of the high severity nature of Rim Fire. Although, the separability between burned and unburned areas is easily distinguished by satellite spectral data, the high spectral and spatial resolution of the airborne datasets provide a unique opportunity whereby studying this megafire as a case study provides both a model for future research of before and after fire analyses using hyperspectral data when it becomes available on satellites, and also the ability to cross-reference remote measurements collected from different technologies to assess megafire characteristics.