Transforming Wildfire Detection and Prediction Using New and Underused Sensor and Data Sources Integrated with Modeling

Wildfire management relies upon prompt detection of new ignitions and timely anticipation of a fire’s growth as influenced by local terrain, fuel characteristics and condition, and weather, notably wind. Recent advances include the CAWFE modeling system, which couples a numerical weather prediction model optimized for modeling fine-scale airflows in complex terrain with fire behavior algorithms, capturing how the fire “creates its own weather”, and ingests spatially refined (375-m pixels) satellite active fire detection products from Visible Infrared Imaging Radiometer Suite (VIIRS), igniting simulated fires ‘in progress’. Assuming regular fire mapping data, this allows an accurate forecast of fire growth for the next 12–24 h; sequences of these simulations may maintain a reasonable forecast of fire growth from detection it until it is extinguished. However, accurately anticipating a fire’s growth is a difficult forecasting challenge because of accumulating model error, stochastic processes, and intervention (i.e. firefighting), data may be missing, and some conditions are inherently less predictable. Here, we develop and apply algorithms (steered by other data) to distill new and existing (but underutilized) sources of data on wildfire detection and mapping, develop and apply algorithms to integrate asynchronous data on wildfire detection and monitoring with coupled weather–wildland fire models, and assess the improvement in wildfire detection time and forecasted fire growth. We investigate the use of additional datasets from satellites with fire detection algorithms and adjacent non-utilized passes of VIIRS to enhance simulations of the 2015 Canyon Creek Complex. These additional asynchronous data allowed 1–3 h earlier fire detection by remote sensing, allowed forecasts to begin and be delivered that much earlier, and introduced several additional simulations into the cycling forecast of a three-day fire growth period. By updating the anticipated fire growth forecast more frequently, these supplemental simulations improved fire growth forecast and compensated where standard, scheduled observations were missing or obscured by clouds.