Proceedings of the 2022 Mississippi Water Resources Conference

Year: 2022 Authors: Raczynski K., Dyer J.

Understanding the patterns of low streamflow (a.k.a., low flow) frequency and intensity is critical in defining potential environmental and societal impacts on processes associated with surface water resources; therefore, the objective of this study is to quantify the multiannual variability of low flow river conditions over the Southeastern US. The study was performed using National Water Model retrospective simulations (v2.1), aggregated to daily mean flow values at 73,891 stream segments (of Strahler order 3 and higher) for the period 1979 to 2020. The data were used to calculate annual sums of outflow deficit volumes, from which the autocovariance function (ACF) and the Hurst exponent (H) were calculated to quantify low flow patterns. The ACF approach is commonly used for examining the seasonal and multiannual variation of extreme events, while the Hurst exponent in turn allows for classification of "process memory", distinguishing multi-seasonal processes from white noise processes. The results showed diverse spatial and temporal patterns across the Southeast US study area, with some locations indicating a strong seasonal dependence. These locations are characterized by a longer temporal cycle, whereby low flows were arranged in series of several to dozens of years, after which they didn't occur for a period of similar length. In these rivers, the values of the Hurst exponent were in the range 0.8 +/- 0.15, which indicates a stronger relation with groundwater during dry periods. In other river segments within the study region the probability of low flows appeared random, determined by the Hurst exponent oscillating around the values for white noise 0.5 +/- 0.15. The initial assessment of the Hurst exponent distribution, as well as results of the ACFs, suggests no strict spatial relationships. Also, the correlation coefficients between low flow patterns and river order do not indicate the occurrence of a statistically significant relationship (r ≈ 0.08). The results of the research provide useful information about the spatial and temporal patterns of low flow occurrence across the Southeast US, and also indicate that the NWM retrospective data are able to differentiate the time processes for the occurrence of low flows. The next stage of work will be to estimate the accuracy of the NWM retrospective data in terms of low flow analysis through comparison with observed data from available USGS gauges.