AI-Informed High-Resolution Fuel Mapping

Overview

The U.S. Army Engineer Research and Development Center (ERDC) Cold Regions Research and Engineering Laboratory (CRREL) seeks advanced artificial intelligence (AI) solutions for mapping high-resolution fuel characteristics in cold regions using variable datasets. Desired solutions would deliver new laboratory capability to enhance this research by integrating AI technologies with pre-existing datasets, improving the land cover models that feed wildland fire predictions.

Project Objective

The primary objectives of this Individual Program Requirement (IPR) are to:

• Develop an AI model that will create fuel maps across interior Alaska training lands using multi-modal datasets
• Derive a canopy height model from remote sensing and LiDAR datasets over interior Alaska training lands for training and validation of the AI model
• Develop a model with a feature-extraction backbone
• Publish the AI model to the Permafrost Discovery Gateway
• Coordinate communication between researchers and the wildland fire community

Project Manager

Cold Regions Research and Engineering Laboratory (CRREL), U.S. Army Engineer Research and Development Center (ERDC)

Background

The Department of Defense (DoD) has a clear need to protect its land and assets against damage from wildland fire. The DoD owns millions of acres of land where wildland fire is a hazard. Artificial intelligence (AI) solutions are needed to improve wildland fire forecast models over such large areas of land. AI is an emerging field that will revolutionize how decisions are made across the military and will produce a wide range of technologies and capabilities. AI has the potential to drastically improve the DoD’s ability to manage their natural resources and save costs by identifying the locations most susceptible to environmental damages.  

The DoD has funded a two-year project at ERDC-CRREL to develop AI methods for wildland fire fuel mapping based on pre-existing remotely sensed and ground-truthed data.

Requirements

ERDC invites pre-proposals that contribute to one or more of the objectives to enable a publicly available multi-modal AI model for mapping wildfire fuel across the interior Alaska training lands. This model must take multi-modal inputs including remotely sensed optical imagery, LiDAR, and land cover datasets. One input to the model and separate product will be processing of LiDAR to derive a canopy height model. The model must use a feature-extraction backbone and be published to the Permafrost Discovery Gateway. Frequent communication between project partners and the wildland fire forecasting community is required.

ERDC recognizes that individual entities typically may lack in-house capabilities to address all objectives. Multiple awards may be considered if individual offerors submit proposals addressing some but not all of objectives. Though not required, ERDC prefers unified proposal addressing all objectives through partnership (subcontracting) with other entities. All proposals should clearly identify which objectives are being addressed in the proposal using the nomenclature (A, B,…).

Note: Submissions should NOT include confidential or proprietary details.

Estimated Government Funding Profile

Up to $400,000 for the full requirement

*Multiple solutions may be awarded

All resultant contracts will be firm-fixed price. All items, technologies, and services (including research and development) procured via this CSO are treated as commercial. The Contracting Officer must determine the price fair and reasonable prior to award.

Estimated Period of Performance

Estimated up to 24 months for completion

Desired End State

The desired end state is the completion of a multi-modal AI model for wildfire fuel mapping across interior Alaska training lands. The model will be published to the Permafrost Discovery Gateway for community use to be rerun as conditions change. The outputs, model, and all code shall be provided to U.S. Army Garrison Alaska for incorporation into wildland fire management.

Evaluation Criteria

Submissions will be reviewed based on the criteria described in ERDC’s CSO Solicitation document by ERDC subject matter experts. Submissions may be shared as appropriate with other ERDC stakeholders. The government has the authority to decline all submitted proposals. The government does not plan to engage in the debrief process outlined in FAR part 15 but will provide feedback to unsuccessful offerors as appropriate and at its discretion.

Notional Project Schedule

Proposed project milestones include:

October 7, 2024	Project Announced, Submissions Open for Phase I
October 18, 2024	Question Period Ends, FAQs Finalized
October 31, 2024	Submissions Close for Phase I (Deadline Extended)
October 31-November 18, 2024	ERDC Evaluation Period for Phase I
November 25, 2024	Participants Selected for Phase II are Notified by ERDC
December 16, 2024	Phase II Full Proposal Deadline
January 7, 2025	Phase II Full Proposal Evaluation Period
March-April 2025	Anticipated Phase II Awards

*Dates may vary to accommodate project team and participant availability.

Project Security Classification

Unclassified