Predictability assessment of marine heatwaves in the Northeast Pacific based on SEAS5

Abstract

Marine Heatwaves (MHWs), extreme ocean warming events, have attracted global attention. This research utilizes forecast data from SEAS5 (Seasonal Forecasting System 5) and OISST (Optimum Interpolation Sea Surface Temperature), applying a range of evaluation metrics from both deterministic and probabilistic forecasting viewpoints. It assesses the forecasting performance of the SEAS5 in the Northeast Pacific (NEP) over the period from 1994 to 2021, examining both spatial and temporal dimensions. The midwest of the NEP exhibits subpar performance in terms of both deterministic and probabilistic predictions when compared to other regions. Furthermore, the SEAS5's forecast skill for MHWs is significantly influenced by the El Niño-Southern Oscillation (ENSO) and seasonal variations. This study establishes probability thresholds for MHWs' occurrences to assess MHWs' predictability using SEAS5, demonstrating that forecasting effectiveness across NEP subregions strongly depends on probability thresholds. To evaluate model performance in terms of reliability and resolution, the research concentrates on the Brier Score decomposition, revealing that the southeastern NEP region exhibits superior reliability and resolution. Additionally, the study focuses on not only the comprehensive efficacy of SEAS5 forecast on the NEP as a whole, but also on the specific performance across different regions. The proposed reliability categorization of MHWs indicates that the majority of regions within the NEP fall into Category 3 and above (at least marginally useful) across all lead times. The SEAS5 has shown high predictability in forecasting the occurrences of MHWs in the NEP, exhibiting diverse forecasting accuracy for MHWs across various maritime regions.