Quantifying urbanization-associated changes in terrestrial hydrologic system memory

A common feature of watershed urbanization is increased hydrograph ‘flashiness,’ whereby river discharge fluctuations grow more erratic. Such changes might be intuitively interpreted as a decrease in watershed-scale hydrologic system memory. Here, I investigate this hypothesis through a paired-catchment experiment. The serial correlation coefficient, a common metric of short-term time series memory, is applied to daily winter streamflow data from urbanizing and rural watersheds in the Puget Sound lowland of Washington State, USA. Statistical comparisons confirm that this metric shows highly significant decreases over time in the catchment undergoing land use change, but not in the control watershed, which remains rural over the hydrometric record. Moreover, the mean serial correlation coefficients are statistically indistinguishable between the two catchments over the early period of record, when both watersheds are largely rural, whereas the system memory is far weaker in the urbanized stream relative to the rural stream over the late period, following land use change in the former. The results appear readily interpretable in terms of the physical hydrologic changes typically associated with urbanization. The serial correlation coefficient thus appears to be an instructive measure of urbanization impacts for small streams in this region.