Soil geochemistry controls fire severity: A soil approach to improved understanding of forest fire consequences in southwest Montana.

Wednesday, 17 December 2014
Russell Callahan and Tony Hartshorn, Montana State University, Bozeman, MT, United States
Fire severity can be defined using satellite imagery to ratio mid (~2.2 um) to near (~0.8 um) infrared reflectance values. We examined how lithology and topography affected burn severity, and how post-fire soils data could be used to ground-truth burn severity at two sites in southwestern Montana. A burned area reflectance classification (BARC), lithology, and terrain attributes were used to predict burn severity for the Millie Fire, which was triggered two years ago by lightning and burned ~4,000 ha. Burn severity showed a strong dependence on lithology: the ratio of areas with high burn severity vs. low or moderate burn severities was 2.9 for gneiss (vs. 0.3 for volcanics). The high-severity burn area for the gneiss was larger than the volcanics, despite the latter lithology covering ~270% greater area (~2,600 ha). Aspect and elevation also influenced burn severity with lower severity at higher elevations (2,600-3,000 m) and higher severity at lower elevations (1,800-2,400 m). Southern and western aspects burned more severely than northern and eastern aspects. To clarify whether post-fire soil geochemical changes might predict ground-based estimates of fire severity, a lab experiment was carried out . We expected residual enrichment of trace metal concentrations, as soil organic matter (SOM) was combusted, which we quantified as loss on ignition (LOI). To test this approach, burned and unburned soils were sampled from the ~6000 ha Beartrap 2 fire, which also burned two years. We simulated differing fire severities on unburned soil using a muffle furnace factorially (duration [5, 15, 30, 45, or 60 minutes] x temperature [50, 100, 200, 300, 400, or 500ºC]). Consistent with expectations, unburned samples had a lower mean (±1SD) concentrations for 23 of 30 elements than field-burned samples. For example, barium concentrations ([Ba]) in unburned samples were (708±37μg/g), 16% lower than field-burned [Ba] (841±7 μg/g). Simulated burning yielded smaller [Ba] (732±9 μg/g). Of the 30 trace metals examined, barium explained the greatest fraction of variance in post-burn LOI (R2 =0.79); gallium explained slightly less variance (R2=0.67). Our results document the promise of post-burn soil geochemistry to indicate soil burn severity, which could complement vegetation-based and remotely sensed indices.