We propose to investigate the predictability of north american coastal ocean conditions, including sea surface height, surface and subsurface temperature, mixed layer depth, currents, and…

The National Oceanic and Atmospheric Administration awarded a Cooperative Agreement to The Regents of the University of Colorado for $1.618 million on October 1, 2025, under the Cooperative Institute (Inter-Agency Funded Activities) program (CFDA 11.405) to investigate the predictability of North American coastal ocean conditions and extreme weather events.

The recipient will assess predictability of sea surface height, surface and subsurface temperature, mixed layer depth, currents, and stratification, with specific focus on tail probabilities of temperature and sea surface height associated with marine heatwaves and coastal inundation. Work includes evaluating the impact of data assimilation—including sea surface temperature, altimetry, Argo profiles, moorings, and temperature and salinity profiles—on reanalyses and subseasonal to seasonal hindcasts using 1-degree ocean simulations. The recipient will develop and compare linear inverse models and their variants, incorporating state-dependent noise and seasonal-cycle accounting, alongside hybrid forecast frameworks combining linear inverse models with deep learning techniques such as long-short term memory networks and convolutional neural networks. The research aims to inform strategic deployment of observational resources, guide development of high-resolution forecast systems, and develop a statistical forecast framework with potential transition to an operational product.

Performance occurs in Boulder, Colorado from October 1, 2025 through September 30, 2027. The Cooperative Institute program supports multidisciplinary research collaboration primarily between NOAA scientists and university partners, with awards typically available for up to five years based on fund availability.

We propose to investigate the predictability of north american coastal ocean conditions, including sea surface height, surface and subsurface temperature, mixed layer depth, currents, and stratification. Additionally, we aim to assess the tail probabilities of temperature and sea surface height associated with marine heatwaves and coastal inundation. The proposal leverages a combination of ocean simulations, reanalyses, and hindcasts to enhance understanding and improve forecasts from both dynamical and statistical perspectives. Key objectives include: assessing the impact of data assimilation, including sea surface temperature, altimetry, argo profiles, moorings, and other temperature and salinity profiles, on reanalyses and subseasonal to seasonal hindcasts (1-degree ocean). Evaluate the representation of coastal conditions, including the mean state, variability, and extremes, as well as their dynamical drivers, in both high-resolution and low-resolution model simulations. Developing linear inverse models (lims) and their variants, including those driven by state-dependent noise, those accounting for the seasonal cycle, and combinations of both, to enhance coastal ocean predictions and provide insights into predictability diagnostics. Constructing hybrid forecast frameworks, such as lims combined with nonlinear techniques (e.g., long-short term memory networks) and purely deep learning frameworks (e.g., convolutional neural networks integrated with long-short term memory networks), with the ultimate goal of designing an earth system emulator. The proposed research will provide valuable insights into future observational needs, aiding the strategic deployment of resources, which can, in turn, enhance operational forecast frameworks. IT will also provide guidance on the development of high-resolution forecast systems. The statistical forecast framework has the potential to transition into an operational product, providing forecasts with skill comparable to, or exceeding, that of more computationally intensive ocean dynamical models.