A hydrologic feature detection algorithm to quantify seasonal components of flow regimes

摘要

Seasonal flow transitions between wet and dry conditions are a primary control on river conditions, including biogeochemical processes and aquatic life-history strategies. In regions like California with highly seasonal flow patterns and immense interannual variability, a rigorous approach is needed to accurately identify and quantify seasonal flow transitions from the annual flow regime. Drawing on signal processing theory, this study develops a transferable approach to detect the timing of seasonal flow transitions from daily streamflow time series using an iterative smoothing, feature detection, and windowing methodology. The approach is shown to accurately identify and characterize seasonal flows across highly variable natural flow regimes in California. A quantitative error assessment validated the accuracy of the approach, finding that inaccuracies in seasonal timing identification did not exceed 10%, with infrequent exceptions. Results for seasonal timing were also used to highlight the statistically distinct timing found across streams with varying climatic drivers in California. The proposed approach improves understanding of spatial and temporal trends in hydrologic processes and climate conditions across complex landscapes and informs environmental water management efforts by delineating timing of seasonal flows.