Search

Increasing impacts of wildfires on Western US air quality highlights the need for forecasts of smoke emissions based on dynamic modeled wildfires. This work utilizes knowledge of weather, fuels, topography, and firefighting, combined with machine learning and other statistical methods, to generate 1‐ and 2‐day forecasts of fire radiative energy (FRE). The models are trained on data covering 2019 and 2021 and evaluated on data for 2020. For the 1‐day (2‐day) forecasts, the random forest model shows the most skill, explaining 48% (25%) of the variance in observed daily FRE when trained on all available predictors compared to the 2% (<0%) of variance explained by persistence for the extreme fire year of 2020. The random forest model also shows improved skill in forecasting day‐to‐day increases and decreases in FRE, with 28% (39%) of observed increase (decrease) days predicted, and increase (decrease) days are identified with 62% (60%) accuracy. Error in the random forest increases with FRE, and the random forest tends toward persistence under severe fire weather. Sensitivity analysis shows that near‐surface weather and the latest observed FRE contribute the most to the skill of the model. When the random forest model was trained on subsets of the training data produced by agencies (e.g., the Canadian or US Forest Services), comparable if not better performance was achieved (1‐day R2 = 0.39–0.48, 2‐day R2 = 0.13–0.34). FRE is used to compute emissions, so these results demonstrate potential for improved fire emissions forecasts for air quality models. Plain Language Summary: Increasing wildfire smoke is undoing decades of air quality progress, yet air quality forecasts often miss the most intense smoke events. This is because forecasted smoke is released at constant rates whereas the rate of smoke release from real wildfires varies in time. In this work we teach a machine learning algorithm to predict the daily change in fire heat output, a quantity that is used to calculate wildfire emissions. The machine learning algorithm uses information regarding weather, fuel moisture and amount, and firefighting efforts to make its predictions. We also tried to predict the daily change in fire heat output using only weather information but found the machine learning method to be more successful. Many federal agencies have their own ways of tracking fire weather and fuel moisture, and in this paper, we show that we can apply machine learning to the data from any of several agencies and get the same level of forecasting skill. Key Points: Random forest models trained on weather, fuel, and firefighting data surpass persistence and weather‐based methods to predict fire energyRandom forest models beat persistence across states, for most days in the 2020 fire season, and across levels of fire severityFire weather and latest fire energy predictors add most skill to the random forest; models using agency weather data perform similarly [ABSTRACT FROM AUTHOR]

Predicting the evolution of burned area, smoke emissions, and energy release from wildfires is crucial to air quality forecasting and emergency response planning yet has long posed a significant scientific challenge. Here we compare predictions of burned area and fire radiative power from the coupled weather/fire‐spread model WRF‐Fire (Weather and Research Forecasting Tool with fire code), against simpler methods typically used in air quality forecasts. We choose the 2019 Williams Flats Fire as our test case due to a wealth of observations and ignite the fire on different days and under different configurations. Using a novel re‐gridding scheme, we compare WRF‐Fire's heat output to geostationary satellite data at 1‐hr temporal resolution. We also evaluate WRF‐Fire's time‐resolved burned area against high‐resolution imaging from the National Infrared Operations aircraft data. Results indicate that for this study, accounting for containment efforts in WRF‐Fire simulations makes the biggest difference in achieving accurate results for daily burned area predictions. When incorporating novel containment line inputs, fuel density increases, and fuel moisture observations into the model, the error in average daily burned area is 30% lower than persistence forecasting over a 5‐day forecast. Prescribed diurnal cycles and those resolved by WRF‐Fire simulations show a phase offset of at least an hour ahead of observations, likely indicating the need for dynamic fuel moisture schemes. This work shows that with proper configuration and input data, coupled weather/fire‐spread modeling has the potential to improve smoke emission forecasts. Plain Language Summary: Predicting wildfire growth and smoke behavior is an important and historically difficult task. Here we compare a coupled fire‐weather model to current practices used in air quality forecasting in their ability to predict daily wildfire spread and fire intensity for the 2019 Williams Flats Fire. We also evaluate the fire‐weather model against high‐resolution satellite and aircraft measurements in ways previously unexplored. We find that including fire‐fighting efforts in our model makes the biggest improvements toward accurate wildfire forecasting. When we account for fire‐fighting efforts, fuel moisture maps, and increased fuel loads, our coupled fire‐weather model significantly outperforms simpler methods with reasonable computational time. Despite this, modeled fire intensity tends to lead observations by at least an hour. Optimal configuration options and avenues for forecasting improvements are discussed. Key Points: Adding containment lines into our wildfire model made the biggest difference in improving accuracy of daily burned area predictionsAccounting for containment, fuel moisture content maps, and increased fuel density consistently increased prediction skill above persistenceModeled fire radiative power leads observations by at least an hour, but with similar accuracy to prescribed diurnal cycles [ABSTRACT FROM AUTHOR]