Changes in streamflow statistical structure across the United States due to recent climate change
Abhinav Gupta, Rosemary Carroll, Sean Andrew McKenna
Abstract
Changes in statistical structure (spectral properties) of streamflow across the USA due to climate change were studied for water years 1980–2013. The Fractionally differenced Autoregressive Integrated Moving Average (FARIMA) model was fit to the deseasonalized streamflow time series to model its statistical structure. FARIMA allows the separation of streamflow into low- (slowly varying) and high-frequency (fast varying) components. In this study, the changes in the contribution of various components to the total streamflow variance were analyzed. Further, the variables affecting these changes were identified using random forest and linear regression. Results show that in the snow-dominated watersheds, the contribution of low-frequency components to total streamflow variance decreased over the study period, and the contribution of high-frequency components increased. The changes in the snow-dominated watersheds were primarily driven by changes in precipitation statistics and changes in snowpack but also by changes in seasonal temperature statistics. Among the rain-driven watersheds, the contribution of high-frequency components generally increased in arid regions but decreased in humid regions. In both humid and arid rain-driven watersheds, increasing winter temperature appears to be responsible for the change in streamflow statistical structure. These results have consequences for the predictability of streamflow in the presence of climate change. Further, the analysis carried out in this study allowed us to understand the plausible changes in watershed hydrologic processes that affect streamflow without using process-based or conceptual models.