Objective

Wildfires are a top concern for Interior Alaska military land managers, yet officials lack accurate predictive tools to manage the risk. U.S. Army Garrison Alaska (USAG Alaska) manages 1.6 million acres of land in Interior Alaska, which is vulnerable to the effects of wildfire, and Alaska has experienced three of four of the highest-acreage fires on record since 2000. Warming air temperatures are projected to increase the length and severity of the Alaska fire season, requiring more resources to fight and manage this expanding threat. Due to these worsening risks, USAG Alaska training land managers need geospatial tools to inform land management practices and prioritize limited resources across a vast and inaccessible domain. One critical input for wildfire forecasting is characterization of fuel sources. Artificial intelligence (AI) has proved to be an efficient tool for characterizing large remote-sensing datasets. The objective of this two-year effort is to use existing AI technology to identify, map, and characterize different wildfire fuels across Interior Alaska military lands at higher resolution than current wildfire forecasting inputs and make the new dataset discoverable on existing cyberinfrastructure platforms, such as the Permafrost Discovery Gateway and Defense Installation Spatial Data Infrastructure Portal. This technology will use existing AI models and geospatial datasets to generate updated fuel maps for refined wildfire forecasting and to help manage resource allocation.

Technology Description

This project will use existing workflows that contain AI technology to efficiently map high-resolution and high-precision wildfire fuels over Interior Alaska Army training lands. Existing geospatial datasets will be used to create a “fuel loads” training dataset, so that the AI model can learn to map fuels solely from satellite or airborne optical imagery. The product will be much higher-resolution (1 m) than existing fuel maps (30 m) and can be repeated efficiently as fuel loads change over time. The training dataset will include high-resolution satellite optical imagery, light detection and ranging datasets, and updated ecotype classifications across the training lands from SERDP project RC18-1170. The technology will apply emerging AI techniques to map spatially precise wildfire fuels across Interior Alaska training lands. Project success constitutes: 1) soldiers and training land managers engaging in the development of the AI fuel mapping model and discovery tools; 2) soldiers and land managers using produced fuel maps to inform wildfire management and infrastructure decisions; and 3) identifying a path forward for developing an AI model for wildfire forecasting using fuel map inputs produced during this effort.

Benefits

There are numerous financial and operational benefits associated with having the ability to better predict the location and severity of wildfire activity due to improved fuel characterization. Currently, USAG Alaska spends $1.3 million on external wildfire management per year, in addition to resources spent internally on wildfire management. Many of the management techniques deployed require direct human involvement to identify fuel types and prioritize DoD assets when a fire ignites. This is costly and can overwhelm the system due to the vast amount of vulnerable land. This effort will develop and deploy an AI-based fuel mapping model to provide a first-of-its-kind one-meter-resolution map of wildfire fuel types across Interior Alaska military lands. This data product will provide land managers with an updated, high-resolution input to wildfire forecast models. The technology will support land managers in identifying vulnerable locations and allocating resources more efficiently, reducing the risks and costs (financial and human) of fighting wildfires. Under current climate predictions, wildfire will be an ongoing and increasing threat to USAG Alaska soldiers, missions, and property. Better tools, such as the technology used for this project, may save lives, DoD assets, and ensure mission-readiness.