Abstract Understanding the spatial structure of extreme floods is critical both for reliable design flood estimation and for coordinated development of regional response and flood mitigation strategies. Yet, analysis of rare, high‐magnitude floods is challenged by the limited sample size. This study investigates the spatial covariability of extreme floods across the coterminous United States (CONUS) for large return periods (2–100 years) by proposing three distinct co‐dependency measures: (a) annual co‐occurrence probability (), (b) 7‐day co‐occurrence probability (), and (c) Measures of Co‐occurrence within a 500 km radius. The proposed measures are developed to associate flood spatial dependence with the underlying physical drivers and are evaluated against null distributions of spatially independent floods that preserve their seasonality. Results show that floods co‐occur far more often than expected under independence, with stronger dependence for larger return periods (e.g., for 100‐year floods ≈19%, vs. 1% under independence). The analysis indicates that snowmelt‐driven basins exhibit high dependence for smaller floods (2–25 years), but rainfall‐driven regions (particularly coasts) dominate for extreme events (50–100 years). MOC hotspots confirm that summer tropical storms (East Coast) and winter atmospheric rivers (West Coast) are the primary drivers of widespread extremes. Given the proposed co‐dependency measures' effectiveness flood processes at various spatial and temporal scales, we suggest they could be leveraged for regionally tailored, season‐specific flood mitigation and emergency‐response strategies.
Bae et al. (Thu,) studied this question.