STI and other researchers examined how dozens of model inputs and choices affect uncertainty in smoke predictions. We found that improving model inputs produces more accurate smoke predictions. Our findings help land managers use fire to manage ecosystem health while also protecting people from harmful pollutants present in smoke.
Using the 175,000 acre Tripod Fire Complex in Washington state as a test area, we evaluated how the choices made at each step in the smoke modeling process affect predicted pollutant levels. Our study shows that the largest potential errors occur when estimating fire size and biomass available to burn. Because a series of models are run, the uncertainty is compounded as errors are passed to the next modeling step. Use of accurate fire information and data is critical to reducing uncertainty in the overall modeling chain, resulting in better planning that protects public health.