Fuel Load Estimation
Fuel load and moisture content has a major impact on the resulting smoke plume by determining emission rates of a number of pollutants including PM2.5, (i.e., emission factors) and the initial trajectory of the plume (i.e., heat released affects plume rise). Thus, attempts to use the Bluesky framework are pointless unless fuel load can be approximated with some certainty. Unfortunately, fuel loads vary spatially and temporally across the Flint Hills. The primary factors that govern this variation are natural differences in edaphology and ecology (soils, vegetation composition, slope, etc), annual variations in climate (precipitation during the growing season), and differences in land management (grazing intensity, prior burn frequency). Fortunately, almost all the land subjected to prescribed burning is grazed, in fact the main reason for burning is to increase productivity of the grassland and increase the rate of gain of the cattle. Most grazed land is dominated by two warm season grasses, Big Bluestem (Andropogon gerardii) and Indiangrass (Sorghastrum nutans) along with a few other minor grass species. Thus, characterizing the species composition of the fuel load is not the challenge. The problem is predicting how much dead standing grass is present at the time of the burning and how it's distributed across the area slated for a prescribed burn. Preburn biomass varies greatly among years and is equally impacted by grazing and landscape form. Figure 8 shows preburn biomass collected at upland positions from grazed and ungrazed prairie near Manhattan Kansas. Similar data show how landscape position affects fuel loading (Figure 9). These data show that fuel loads can vary from 300 to 800 g/m 2 among years, a two or three-fold interannual difference. Furthermore biomass on grazed areas is about 0.6 of that on ungrazed land and additional spatial variation is caused by topographic position (upland vs. lowland) and burn history. Clearly, having site specific and year specific fuel load data is crucial if smoke emissions are to be predicted with any accuracy.
Fuel load analysis will be done in two parts; 1) a field component conducted on small scale, localized research sites to establish the factors controlling fuel load distribution and how these can be modeled and predicted using spectral data, and 2) a remote sensing component in which the fuel load model developed in part 1 is rescaled to predict fuel load over the entire Flint Hills study area.