In the Laurentian Great Lakes Basin, water quality issues persist despite ongoing management and monitoring efforts. The need for enhanced understanding of nutrient transport from agricultural fields to surface water is imperative to reduce consequences on human and aquatic life, such as contaminated drinking water and continued algal blooms. Watershed nutrient export to surface waters is governed by seasonally varying and complex hydrologic processes including subsurface flow pathways. Recent advancements in data analytics, particularly machine learning-based approaches, have enabled accessible multivariate analysis of large environmental datasets, yet their application remains under-explored in nutrient studies. Here, data-driven analysis was used to identify seasonal hydrologic controls of nutrient (nitrate and total phosphorus) export, including groundwater-surface water interactions, in an agricultural watershed in southwestern Ontario. A two-year dataset consisting of field measurements, remote sensing products and engineered hydroclimate data was analyzed using correlation analysis, principal component analysis (PCA), K-means clustering, and hierarchical clustering. Seasonal correlation analysis and PCA captured distinct hydrologic modes between the growing and non-growing season, reflecting shifts in dominant surface and subsurface controls. Clustering analysis independently grouped observations into growing and non-growing season clusters without seasonal labels, highlighting the importance and difference between these seasons. Both PCA analysis and clustering revealed that non-growing season events were associated with the greatest watershed nutrient export. Results from this study show that advanced statistical analysis can reveal important seasonal hydrologic controls (surface and subsurface) and hot moments of nutrient loss, which can aid in improved water quality monitoring and management for healthier ecosystems.