Prediction, Sensitivity, and Dynamics of Subseasonal To Seasonal Phenomena Diagnosed Through Linear Inverse Models, Their Adjoints, and Numerical Weather Prediction Models

“Subseasonal to seasonal (S2S) predictability faces a unique set of forecast challenges related to initialization, parameterization, and development of model bias around which the forecast state must evolve. Furthermore, a wide range of phenomena exist on S2S timescales that require realistic evolution of the forecast state over a wide range of geographical regions and physical processes (e.g. tropical intraseasonal variability, low-frequency mid-latitude wave dynamics, and downward propagation from the stratosphere), which presents a major challenge for numerical weather prediction (NWP) models that aim to forecast out to S2S time scales. Additionally, targeted forecasts of different specified extreme events (e.g. blocking, heat waves, cold snaps) may be associated with very different sensitivities to any of these processes. A general framework is needed for analyzing and evaluating the role of initial conditions, model physics, and development of model bias, in S2S predictability of specific extreme events. The project will utilize linear inverse models (LIMs) that are derived to infer the dynamics of the atmosphere and ocean from a set of observed states; these states can come from analyses (which are largely free of model bias) or from forecast model states (which may contain biases around which a NWP model state must evolve). The feasibility of the LIM for studies of low- frequency variability has been demonstrated in previous studies.

This project will advance predictive capacity for S2S forecasting by (i) developing LIMs separately around the analysis state (analysis-LIM) and around NWP forecast states that include the effects of the evolved model bias (forecast-LIM); (ii) analyze the role of initial conditions, model physics, and development of model bias in event-specific subseasonal prediction; (iii) test findings using LIM-informed data-denial initialization experiments in a NWP model; and (iv) blend NWP and LIM strengths to maximize predictive skill on S2S time scales. The combination of tasks in this project compose a general framework that addresses the unique challenges outlined above, and informs priorities for NWP model development related to S2S prediction.

This project is directly relevant to the goals of Competition 2 in the FY16 MAPP funding opportunity, through understanding the predictability and potential to advance prediction of specific phenomena occurring on S2S timescales. The project develops a framework that can be used to investigate how prediction of S2S phenomena is influenced by specific initial states, coupling between different model components and regions, and model physics (and associated evolution of model bias). The framework will be applied to specific S2S prediction efforts and will be provided in real-time for modeling groups to gain familiarity with the process. Our long-term goal is to develop an operational approach for a priori determination of “state-dependent” subseasonal forecast error. The PIs will also contribute to a MAPP Task Force by providing a product that can be used in real-time to diagnose S2S predictions from a variety of different NWP modeling efforts. Finally, this project is also relevant to NOAA’s Next Generation Strategic Plan for a “Weather-Ready Nation”, through increasing understanding of predictability of potentially hazardous extreme weather, as well as phenomena of importance to the energy sector.”