Analysing multidimensional meteorological and oceanographic (MetOcean) datasets across space and time remains challenging due to data volume, structural heterogeneity, and the need to coordinate exploratory analysis with statistical characterization of extreme events. In practice, visualization, exploratory data analysis, and extreme value analysis (EVA) are often performed using loosely coupled tools, limiting analytical coherence when spatial, temporal, or variable constraints change. This paper presents MetOceanVis, a modular web-based visual analytics system designed to support a coordinated, constraint-driven analytical workflow for spatio-temporal MetOcean data. The system adopts a four-layer architecture encompassing data ingestion, processing, integrated analytics, and interaction, enabling user-driven querying over temporal windows, spatial domains, and selected variables. Exploratory data analysis (EDA) views including time series plots, boxplots, scatter plots, correlation heatmaps, and histograms are integrated with EVA modules implementing block maxima and peak-over-threshold models, allowing extreme value characteristics to be explored within their spatial and statistical context. The system is demonstrated through a case study using the SEAFINE-2 dataset, focusing on extreme event-oriented analysis of MetOcean conditions in the South China Sea. A performance evaluation shows that the system maintains interactive responsiveness at the data scales considered in this work, supporting exploratory and extreme value analysis under varying query constraints. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2026.