Assessing the potential for evaluation of wildland fire models using remotely sensed data—Summary proceedings from a U.S. Geological Survey workshop in 2024
On September 19, 2024, the U.S. Geological Survey (USGS) held a virtual workshop titled “Potential for Evaluation of Fire Models with Remote Sensing Data Workshop” to assess the feasibility of using remotely sensed datasets to evaluate next-generation wildland fire behavior models. Remote sensing and fire modelling experts gathered to: (1) assess the suitability of a variety of classified, commercial, and publicly available remotely sensed datasets for advancing fire model evaluation; (2) develop ideas on how to integrate remotely sensed data products with fire model inputs and outputs; and (3) identify any barriers and limitations to performing an evaluation of next-generation fire models. The USGS National Civil Applications Center, USGS Earth Resources Observation and Science Center, and USGS Fort Collins Ecosystem Science Center presented information on remote sensing datasets for three Arizona wildfire case studies. The development teams of the Fire Dynamics Simulator and QUIC-Fire fire behavior models presented their models and current evaluation methodologies. Interspersed with these presentations were discussions regarding how to expand current wildfire remote sensing data collection efforts beyond operational needs to assist in future fire modeling. Workshop participants agreed that several of the remote sensing datasets have potential for wildfire model evaluation. However, participants also identified several barriers and complications to performing a model evaluation including key gaps in wildfire datasets; uncertainties related to model fire-atmosphere reinitiation; lack of ground truthing and atmospheric correction of remotely sensed datasets; and differences in spatial, geolocation, radiometric, and temporal resolutions between the datasets and models. Further, the absence of standardized methodologies for image interpretation, poor understanding of sensor capabilities and limitations, and a lack of automation also hinder model evaluation efforts. Based on feedback from this workshop, USGS fire modelers are considering a project to address the uncertainties related to fire model reinitiation and encouraging fire practitioners to collaborate with remote sensing experts on wildland fires to improve data collection for a broader community of practice. Additionally, multiagency efforts are in development for a comprehensive cross-sensor validation and ground-truth campaign to test spatial, spectral, and geolocation sensor capabilities, determine limitations, and identify observational gaps for future sensor development and acquisition.