| Popis: |
Background Fire is an essential component in restoring and maintaining healthy forests. However, historic land use and decades of fire suppression has excluded fire from millions of forested hectares across much of the western United States, including Grand Canyon National Park (GCNP). Forest management at GCNP aims to reduce wildfire vulnerability by applying fire (both natural and planned ignitions) to remove forest vegetation and forest fuels and promote species diversity. However, the cost, complexity, and concerns associated with managing fire for resource benefit requires that fire managers utilize and implement locally relevant, science-based knowledge to strategically identify when and where to use fire to produce the greatest benefit. Observations from GCNP, topographic variation and weather were used to assess thresholds that affect burn severity and fire spread to identify conditions that may be beneficial or incompatible with multiple resource objectives. Results For ponderosa pine and mixed conifer forests, we developed burn severity and fire spread models by incorporating fire weather variables using gradient boosting machine learning on multi-day wildfires between 2000 and 2018. Elevation, wind, and vapor pressure deficit (VPD) were among the most influential across all models. Elevation was the most influential predictor for both the ponderosa pine severity and spread models. Wind and topographic roughness index were the most influential predictors in the mixed conifer spread and mixed conifer severity models, respectively. Using these models, prediction tables were generated to characterize expected burn severity and fire spread associated with common weather conditions, based on elevation, wind, and VPD. At lower elevations, a VPD threshold between low and moderate severity of 2.0kPa with winds between 8-10mph in both severity models was observed. Thresholds of VPD and wind decreased as elevation increased. In both forest types, average wind speeds greater than 4mph were found to be a threshold for fire spread. Greater fire spread was observed as winds and VPD increased. Conclusions Prediction tables can provide fire staff with a quick but comprehensive assessment of the relative likelihood of expected fire activity during changing weather conditions over the course of a fire event or season based on previous fire activity at GCNP. |
