Comparison of a Soil Moisture Surrogate (Keetch-Byram Drought Index) with In-Situ Measured Soil Moisture as a Predictor of Wildfire Danger in Oklahoma

Due to the past unavailability of soil moisture measurements, national and state wildfire danger assessments have often used the Keetch-Byram Drought Index (KBDI) as a surrogate for soil moisture. Originally developed in the 1960s for the forested regions of the southeastern United States, KBDI attempts to model soil moisture conditions by estimating the net effects of precipitation and evapotranspiration. But with the increasing availability of in-situ soil moisture measurements, such soil moisture surrogates may no longer be necessary. In Oklahoma, for example, soil temperature and soil moisture conditions are recorded continuously by the Oklahoma Mesonet, which includes more than 100 stations across the state and has a record length spanning more than 20 years. Recently, significant connections between soil moisture and the occurrence of large wildfires (≥ 1000 acres) in Oklahoma have been found using data from the Oklahoma Mesonet, suggesting that measured soil moisture may be a promising alternative to KBDI in wildfire danger assessments. Given that global wildfire potential is projected to increase throughout the century, with some of the most dramatic increases in wildfire potential expected in the United States Great Plains, any improvement in wildfire preparedness gained using measured soil moisture could result in significant decreases in property damage and loss of life.

Up until recently, the relative merits of measured soil moisture and KBDI as indicators of wildfire danger have been largely unknown. This presentation will specifically address this knowledge gap by comparing the utility of measured soil moisture versus KBDI for wildfire preparedness. The three major objectives of our research were: (1) to identify relationships of KBDI and FAW (fraction of available water capacity), an index calculated from in-situ measured soil moisture, with 34,939 growing and dormant-season wildfires; (2) to compare relationships between each of the two indices and wildfire probability for 501 large (≥ 1000 acres) growing-season and dormant-season wildfires; and (3) to quantify relationships between KBDI and FAW for each season in Oklahoma. Neither KBDI nor FAW accurately represented dormant-season wildfire danger, with wildfires ≥ 300 acres occurring across nearly the entire range of each variable. During the growing season, however, we found that a smaller percentage of wildfires ≥ 300 acres occurred under extreme levels of KBDI than under equivalent levels of FAW (66% vs. 81%), and a logistic regression model based on FAW correctly classified more growing-season days with large wildfires than the KBDI model (84% vs 79%). Furthermore, while FAW represented soil moisture in near real-time, KBDI responded more slowly to soil drying and recharge, with the result that FAW provided about 10 days earlier warning of extreme wildfire potential for the 10 largest growing-season wildfires in our study. After discussion of the research results, a daily operational product will be presented that was developed out of this research and is currently on the OK-FIRE wildland fire management system of the Oklahoma Mesonet. In summary, our research shows the importance of in-situ measured soil moisture and argues for the replacement of KBDI with FAW in growing-season wildfire danger assessments for regions having available soil moisture measurements and having similar climate and vegetation to that of Oklahoma.