Machine Learning (ML) of Forest Stand Metrics to Quantify Carbon Storage | W81EWF 23 SOI 0004

Opportunity Information: Apply for W81EWF 23 SOI 0004

The Department of Defense, through the U.S. Army Corps of Engineers (ERDC), offered a discretionary research opportunity (Funding Opportunity Number W81EWF 23 SOI 0004) to develop and demonstrate machine learning methods for estimating core forest stand metrics from remotely sensed data in order to better quantify forest carbon storage. The central idea is to bridge global-scale data resources with local-scale decision needs by using modern ML models, trained on large environmental and forest inventory datasets and run on high performance computing systems, to produce high-resolution estimates that can be applied down to a specific area such as a single DOD installation. The work is grounded in the premise that accurately characterizing forest habitat type and forest volume, derived from measurements like tree height, diameter, and density, provides a strong basis for understanding and mapping carbon stocks.

The project emphasizes turning traditional forest inventory measurements into scalable, remotely derived products. Rather than relying only on field plots, the effort aims to show that metrics such as tree diameter and stand density can be inferred effectively using ML applied to remotely sensed observations, supported by extensive training data and relevant environmental covariates. A key feature of the approach is the integration of numerous variables (for example, site conditions and other environmental drivers) with forest inventory datasets to capture forest growth patterns and successional conditions, improving the reliability of carbon-related inferences at fine spatial scales.

The expected work is structured as an initial year followed by up to two optional years, with the note that the government does not anticipate overlapping periods of performance. During the initial year, the awardee is expected to assemble a capable technical team, select one or more initial study areas, and build a proof of concept that demonstrates novel methods for quantifying basic stand metrics. In parallel, the team must compile a repository of forest inventory data sourced from national and international partners, then validate the accuracy of the resulting prototype metrics, establishing an early evidence base that the ML-derived estimates are credible when compared against independent reference data.

Optional Year 1 is oriented toward scaling and refinement. The project would expand to larger or additional study areas, improve and tune the initial methods, and potentially broaden the inventory-data repository as needed to support the new geography. The work also requires validation of the larger-area stand metric outputs, focusing on prioritized areas of interest, to show the approach remains accurate beyond the initial proof-of-concept footprint. A further deliverable in this year is a peer-reviewed journal article co-produced with ERDC researchers that documents the methodology and its application to forest stand metric estimation.

Optional Year 2 focuses on consolidation, final performance testing, and broader dissemination. The awardee would conduct a final accuracy assessment and refine methods if accuracy targets are not being met, then produce additional peer-reviewed publication(s) with ERDC integrating conclusions across all study phases. The effort also calls for public seminars to communicate findings, signaling that the government is looking not only for a technical prototype but for results that can be reviewed, replicated, and adopted by a wider community.

The opportunity was offered as a cooperative agreement under the Science and Technology and other Research and Development activity category (CFDA 12.630). It anticipated a single award with an award ceiling of $150,000. The posting lists eligibility broadly as "Others" with further clarification expected in the full eligibility text. Overall, the government is seeking applicants who can combine domain depth in forestry and carbon storage with practical field data capabilities, strong experience assembling and harmonizing forest inventory databases at national to global scales, and demonstrated ability to develop novel ML approaches for characterizing forest attributes from remote sensing data. The ideal team would also bring advanced computing capacity suitable for training and evaluating ML models on HPC infrastructure, since the project depends on computationally intensive workflows to deliver high-resolution, validated forest stand metrics that inform carbon stock estimation at local scales.

• The Department of Defense, Dept. of the Army -- Corps of Engineers in the science and technology and other research and development sector is offering a public funding opportunity titled "Machine Learning (ML) of Forest Stand Metrics to Quantify Carbon Storage" and is now available to receive applicants.
• Interested and eligible applicants and submit their applications by referencing the CFDA number(s): 12.630.
• This funding opportunity was created on Mar 23, 2023.
• Applicants must submit their applications by May 22, 2023. (Agency may still review applications by suitable applicants for the remaining/unused allocated funding in 2026.)
• Each selected applicant is eligible to receive up to $150,000.00 in funding.
• The number of recipients for this funding is limited to 1 candidate(s).
• Eligible applicants include: Others (see text field entitled Additional Information on Eligibility for clarification).

Apply for W81EWF 23 SOI 0004

[Watch] Creating a grant proposal using the step-by-step wizard inside the applicant portal: