This project addresses known data needs for the 1- to 10-year time horizon between weather forecasts and climate model projections for actionable information that can be used for climate risk decision-support. Resolving this Weather-to-Climate Continuum (W2CC) gap is a grand challenge for both civilian and military operations and infrastructure planners. For example, Department of Defense (DoD) activities ranging from theater security, to force readiness, to asset design should account for potential conditions and extremes that could impact installations, missions, and operations worldwide within the next 10 years. This rapid response Electric Power Research Institute, Inc.-National Science Foundation National Center for Atmospheric Research (NSF NCAR) project systematically reviews state-of-the-science W2CC approaches and objectively evaluates a hierarchy of forecasts from the perspective of military needs. Specifically, the project team will develop forecast datasets for regions of DoD interest and conduct an evaluation in support of customized user guidance for DoD climate resilience efforts. The project has been scoped to deliver results within 20 months, including benchmarks that demonstrate performance relative to current practices, enabling DoD to take advantage of cutting-edge methods and datasets that are available or at a very high readiness level.

Annual to decadal forecasting is a quickly emerging prediction science area spanning the time domains of seasonal forecasting and century-long projections produced by traditional climate models, in which the latter tools are run in a novel configuration for decadal forecasting. This rapid response project leverages available simulation results from emerging methods and further enhances them with machine learning (ML)/artificial intelligence (AI) post-processing approaches to bias correction and downscaling, including NSF NCAR’s analog ensemble approach and a vision-based deep learning architecture. The core technology developed is a suite of decadal forecasts derived from a hierarchy of methods ranging from simple to more computationally complex. The relative skill of these approaches will be evaluated in terms of diagnostic quantities (e.g., skill and uncertainty of meteorological variables) and documented level of user effort and cost for specific locations and hazards. Regional datasets will be integrated into the Defense Climate Action Tool (DCAT) and findings communicated to DoD users through a multi-layer knowledge transfer approach.

This project has been designed to provide a rapid response that meets DoD needs for near-term climate information by leveraging readily available forecast products enhanced with frontier methods. The products will 1) advance DoD knowledge of available W2CC approaches; 2) assess forecast skill of emerging AI/ML technologies compared to current benchmarks; and 3) provide levels of user guidance and technology transfer. These products will enable DoD to develop guidance for representing near-term climate conditions and extremes in infrastructure planning and risk management procedures for the 1 to 10-year time horizon, which otherwise would be subject to known W2CC data gaps. Further, the project should enable ESTCP to begin connecting shorter-term weather forecasts with multidecadal climate projections, guided by the manifold experience across weather forecasting and climate impacts studies. The long-standing experience translating cutting edge science to military, energy sector, and user community partners, along with the collaborative co-production model, will ensure high-quality information transfer to DoD resulting in improved forecast performance and reduced costs.