“Atmospheric rivers (ARs) are intense synoptic-scale plumes of tropospheric water vapor that can lead to extreme precipitation and flooding when they make landfall. These features cause extreme flooding events not only along the west coast of the contiguous United States (CONUS), but also in Canada and Alaska. The ability to forecast ARs would provide society with advanced knowledge of their extreme impacts. Recent work by the our team demonstrates an inverse relationship between winter-time ARs hitting Alaska and CONUS, driven by the presence of a blocking anticyclone over the east Pacific that acts to divert the ARs away from CONUS and into the Gulf of Alaska. The potential exists to forecast the probabilities of North Pacific blocking and AR occurrence through knowledge of the Madden-Julian oscillation (MJO).
Specifically, additional recent work by our team demonstrates that the pattern of blocking leading to an increase in Alaskan ARs (and subsequent decrease in CONUS ARs) is driven, at least in part, by phase 8 of the MJO – a phenomenon that is potentially predictable on timescales of 4 weeks or longer. Thus, while ARs are synoptic features unlikely to be forecast explicitly for lead times beyond 10 days, great promise exists for forecasting their probabilities based on knowledge of the MJO. However, many climate models and numerical weather prediction models cannot simulate the MJO with fidelity. The overarching goal of the proposed work is to quantify the extent to which east Pacific blocking and AR probabilities can be skillfully
forecast at lead times of multiple weeks through their dynamical link with the MJO, including an explicit investigation of how AR prediction skill varies with a model’s ability to forecast the MJO.
To achieve this goal, the tasks outlined in this proposal address two main objectives:
Objective I: Quantification of the predictability and prediction skill of North Pacific blocking and atmospheric river probabilities through knowledge of the MJO.
Objective II: Assessment of the sensitivity of forecast skill to MJO skill and model setup.
To address Objective I, the proposed work will develop statistical forecast models to predict blocking and AR probabilities using knowledge of the MJO and other predictors. Then, we will quantify the ability of operational models to forecast AR and blocking probabilities by analyzing data from two hindcast data sets: the S2S and ISVHE databases. The second objective of the proposed work will address the extent to which AR forecast skill is sensitive to the prediction system setup by further analyzing the S2S and ISVHE databases and performing a series of simulations with two GCMs. Specifically, we will address the extent to which the forecast skill of blocking and AR probabilities are sensitive to: the model’s MJO forecast skill, physics (i.e. cloud parameterization), resolution and forecast lead time.
Relevance to NOAA and Task Force Contributions: This proposal directly addresses the
FFO “MAPP – Research to Advance Prediction of Subseasonal to Seasonal Phenomena” by improving the understanding of how formulation of the prediction system, including physics and resolution, impacts the forecast skill of extreme events (i.e. atmospheric rivers and blocking) and what physical processes lead to their predictability. We will aid NOAA’s NGSP by improving process-level understanding and prediction skill of persistent flow regimes associated with blocking, and extreme rainfall associated with atmospheric rivers, to provide more accurate “assessments of current and future states of the climate system that identify potential impacts and inform science, service, and stewardship decisions.” The proposed efforts will contribute to the new MAPP S2S Task Force by (1) providing scientific leadership of the Task Force (2) linking international efforts on advancing S2S prediction, (3) contributing new understanding of the influence of prediction system setup on forecasts of extreme events (4) delivering a database of extreme events, blocking and atmospheric river occurrence and statistical forecast models.”