Search for: All records

Abstract We investigate coupled climate model initialized predictions of the Pacific Decadal Oscillation (PDO) prediction skill in the Community Earth System Model (CESM) Seasonal to Multi Year Large Ensemble (SMYLE). The PDO is predictable up to a year in advance in SMYLE; however, the predictability depends on verification month, with skill degrading most rapidly in boreal spring for all initializations. To examine the role of teleconnections from El Niño–Southern Oscillation (ENSO) in the prediction skill of the PDO, we use a multi‐linear regression model. The linear model shows that initial value persistence explains most of the PDO prediction skill in SMYLE. In addition, the PDO prediction skill's seasonal dependence is fully reproduced only when ENSO is included as a predictor. These results suggest that ENSO has a strong influence on the seasonality of PDO predictions. 

Abstract Accurate and interpretable marine heatwave (MHW) forecasts allow decision makers and industries to plan for and respond to extreme ocean temperature events. Recent work demonstrates skillful pointwise prediction of MHWs. Here, we evaluate a method of detecting and predicting spatially connected MHW objects. We apply object‐based forecast verification to the Community Earth Systems Model Seasonal‐to‐Multiyear Large Ensemble (SMYLE) experiment, a set of initialized hindcasts with 20‐member ensembles of 24‐month simulations initialized quarterly from 1970 to 2019. We demonstrate that SMYLE predicts MHWs that occur near observed MHWs with high skill at long lead times, but with errors in location, area, and intensity that grow with lead time. SMYLE exhibits improved skill in predicting the intensity of MHWs in December and January, and worse skill from August to October. This work illustrates the capacity to forecast connected MHW objects and to quantify the uncertainty in those forecasts with potential applications for future community use.