Fine tuning FireMapper imagery for post-fire plant recovery

Background/Question/Methods

Over 16,000 ha of mixed age chaparral growing over topographically diverse southern California terrain were burned by the Esperanza Fire during Santa Ana conditions in October 2006. Esperanza traversed North Mountain Experimental Area and adjacent lands where fire history and research had been documented for about 100 years. The Forest Service’s airborne FireMapper multi-spectral remote-sensing system was available to map at high resolution the fire’s spread and radiant-energy release. Subsequent flights during 2006, 2008, 2009, and 2011 documented the appearance of the post-fire landscape and changes with vegetation recovery in reflectance of red, near-infrared and short-wave infrared light and in radiometric temperature. These data have been applied with ground-level monitoring and historical records to advance our understanding of fire behavior and vegetation recovery as a function of fire severity. Ground-level monitoring involved measurement of plant cover by species along transects within 64 20x20m² plots within and adjacent to the burn. Because the landscape was highly dissected, improving the relationship of species composition to FireMapper spectral values required working with vegetation measurements at meter to meter rather than at the plot level. Consequently, in 2011, sixteen plots were selected to re-sample groundcover at consecutive intervals along five evenly spaced 20m point-intercept transects.

Results/Conclusions

FireMapper images in plots that contained an abundance of broad leaf species (manzanita and oak) clustered separately (in image scatter grams of red vs. near-infrared values) from plots in which chamise (a small leaf species) was predominate. A much higher degree of accuracy was obtained for the location of each of the plots and the ends of the sampling transects when multiple sets of GPS-collected coordinates were averaged. Preliminary analysis shows that vegetation recovery can be tracked in sequential FireMapper images and that general species categories can be delineated on the FireMapper multi-spectral imagery. Better resolution for interpreting the spatial relationship of species composition to FireMapper spectral values was achieved at the meter compared to the plot level. Individual plants and landscape features could be identified in many cases. Further processing and analysis is underway to improve the correlation of the FireMapper images to exact land features. Once refined, land managers may find that fine tuning FireMapper imagery to a species specific spectral identity could prove advantageous for resource management applications such as monitoring disturbed site recovery, locating and tracking populations of important or sensitive plant species, and expediency when surveying plants in large, remote or inaccessible areas.