ROSES 2025: A.9 User-Centered Applications with Large Earth Foundation Models

The NASA Research Opportunities in Space and Earth Sciences (ROSES) program element A.9, titled “User-Centered Applications with Large Earth Foundation Models,” supports the development of decision-support tools utilizing Prithvi Foundation Models (FMs). These large Earth FMs are advanced AI models co-developed by NASA and IBM and trained on significant Earth science datasets. The solicitation aims to harness the power of Prithvi-Earth Observation (EO) and Prithvi-Weather and Climate (WxC) models to create scalable, efficient, and impactful tools for decision-makers in sectors such as insurance, infrastructure, transportation, and real estate. NASA intends this funding to stimulate the integration of end-user-driven needs with the computational power of these models to support real-world decisions. Funded projects must address specific decision-making activities and include an identified end user as a member of the project team. Applications must start at Application Readiness Level (ARL) 3 or higher and demonstrate advancement of at least one ARL per year. While projects may last up to three years, preference may be given to proposals that last no more than two years and aim to advance two ARLs. Projects must clearly identify the user, their need, and how the proposed tool will support or enhance decision-making using the Prithvi FMs. Proposals must also describe how the tool or application can be scaled horizontally to reach more users or vertically to improve its operational impact. The opportunity is open to all institutions, with some limitations. NASA IMPACT team members already funded to work on Prithvi cannot receive further funding or serve as collaborators. Non-U.S. institutions may apply under a no-exchange-of-funds basis. Federal agencies outside of NASA and non-JPL FFRDCs are ineligible for funding. Eligible U.S.-based institutions include universities, nonprofits, for-profits, and state and local governments. The inclusion of diverse, interdisciplinary teams—comprising Earth scientists, computer/data scientists, and social scientists—is strongly encouraged. Proposals must be submitted through NSPIRES and follow strict dual-anonymous peer review (DAPR) guidelines. Key proposal components include an anonymized technical document (with sections on decision-making activities, project description, management plan, scalability, and sustainability), an Open Science and Data Management Plan (OSDMP), a redacted budget, and a separate “Expertise and Resources Not Anonymized” document. Review criteria will focus on the scientific and technical merit, relevance to Earth Science to Action (ES2A) goals, and the strength of the proposed team and resources. The total annual program budget is up to $2 million, with an estimated eight new awards to be made. Funding per project is encouraged to stay under $250,000 annually. Projects must allocate funds for mandatory site visits, project meetings, computational costs, and reporting. The anticipated start date for awarded projects is approximately eight months after the release of the solicitation. Required reporting includes quarterly updates, annual progress reports, and a final performance report summarizing results, challenges, and lessons learned. A pre-proposal webinar will be held approximately three weeks after ROSES release, with details posted on the NSPIRES program page. For questions, the primary contacts are Argyro Kavvada (argyro.kavvada@nasa.gov, 202-213-7891) and Tsengdar Lee (tsengdar.j.lee@nasa.gov, 202-251-9863). Technical inquiries about the Prithvi models should be directed to Rahul Ramachandran (rahul.ramachandran@nasa.gov). This funding opportunity is designated under funding opportunity number NNH25ZDA001N-EAFM and is considered highly relevant to NASA’s Earth Science objectives as outlined in the NASA Science Plan.

Ensure the proposal complies with DAPR and includes a named end-user organization.