Understanding the inter-event variability of recession flow characteristics and its drivers

Abstract

The characteristics of recession flows are known to show significant inter-event variability which reflects dynamic changes in baseflow generation mechanisms. The objective of this study is to quantify the variability of recession characteristics in watersheds and to understand its underlying physical drivers. We use a parallel decay model to estimate the number and residence times of watershed storage components contributing to baseflow in each recession event using inverse Laplace transform of recession flows. The proposed method allows for interpretation of both spatial heterogeneity and temporal variability of baseflow generating processes in watersheds. We apply the methodology to streamflow data from 579 catchments in the contiguous United States (CONUS) and analyse the spatial patterns of inter-event variability of recession characteristics. The physical drivers of inter-event variability of recession characteristics are then analysed using random forest models. We find significant variability of recession flow characteristics in all watersheds of CONUS, with higher variability in the watersheds of eastern US and west coast of US, in which residence times have a standard deviation of 35–40 days. On the other hand, comparatively lower variability in recession characteristics is found in central and southwest US, with residence times having standard deviation of 24–30 days. We find that watershed area has the strongest control over inter-event variability of recession flow characteristics with higher variability observed in larger watersheds. Watersheds with winter dominated precipitation regime are found to show higher variability in the number of contributing storages whereas forested watersheds show higher variability in residence times.