Connections Between Low Frequency Streamflow Extremes And Nonlinear Dynamics In The Upper Colorado Basin
Presenter: Patricia Puente1
Co-Author(s): Balaji Rajagopalan
Advisor(s): Dr. Laura E Condon
1Department of Mathematics, University of Arizona
Currently many western states are experiencing long-lasting droughts without precedent in the measured historical record (ie. roughly the last hundred years). However, looking deeper into the past we can learn more about patterns in extreme events that go beyond our observations. Extended streamflow time series created through paleo-reconstruction provide records going back thousands of years which demonstrate the existence of decadal drought periods in the Upper Colorado River Basin (USRB). Previous work has shown that nonlinear embedding can provide insights into epochal variability and predictability in paleo-reconstructed streamflow time series. Here we use a nonlinear dynamical approach to extract high and low predictability regimes through the calculation of Lyapunov exponents, and explore the connection between predictability of regimes and extreme hydrologic events at Lees Ferry, Colorado (the lower boundary of the UCRB). We find a positive linear relationship between streamflow variability and predictability. However there is still great variability in streamflow magnitude across predictability regimes. We also find that patterns of predictability are difficult to extract during periods of extreme events. In order to understand the spatial variability of predictability we extend this study using the same methodology to other gauges on the Colorado River network.