An Assessment of the Stationarity of Climate and Streamflow in Watersheds of the Colorado River Basin

Several studies drawing upon general circulation models (GCMs) have investigated the potential impacts of future climate change on precipitation runoff to streamflow in the Southwest United States and suggest a trend to 10-20% less runoff by mid-century in response to changes in precipitation patterns and increasing temperatures. The implications to surface water resources of a more arid future are potentially significant, posing challenges for water management and for sustaining population and economic growth within the semi-arid Colorado River Basin (CRB). With the hydroclimatic changes considered to be underway, climate scientists have counseled water management professionals to abandon historical assumptions of stationarity in the natural systems governing surface water replenishments. Stationarity is predicated upon the assumption that the process generating outcomes is in equilibrium around an underlying mean and that variance about the mean remains essentially constant over time. The historical streamflow instrument record can be used to estimate the mean, variance and probabilistic nature of seasonal runoff outcomes. The consequent risks to resource operations can be incorporated within management decision-making using stationarity-based or other approaches to modeled outcomes. Eight sub-basins across the CRB were identified for this study having unregulated runoff to streamflow gages, providing a 22% spatial sampling of the CRB. Their 115-year climate history of air temperature and precipitation were obtained along with corresponding streamflow gage records. A methodology was developed to identify runoff seasonality so that the analysis is specific to each sub-basin's cumulative winter runoff - the key focus of water management for resource replenishments. Time series were evaluated with a suite of statistical methods against a set of hypothesis testing criteria established per statistical literature definitions of stationarity. The CRB watersheds have experienced increasing winter air temperature trends since the late 1970s. Increases in the Upper Basin have been approximately 1oC while those in Lower Basin have been approximately 1.5oC. These increases have taken place over ~40% of the timeline towards the mid-century GCM milestone at which declines in streamflow were projected. This study therefore sought to assess any evolving transition of precipitation and runoff to non-stationary behavior. Despite the statistically significant increases in air temperature, hypothesis tests of the winter precipitation and runoff time series conclude that they remain stationary processes. Transitions through periods of drought and excess have been characterized, with current streamflows found to be close to the long-term means. Seasonal precipitation is found to be independent of adjacent years and seasons, while runoff is influenced by antecedent season climate, without year-to-year autocorrelation. These findings suggest that water management practices should not hastily abandon assumptions of runoff stationarity, as outcomes remain consistent with characteristic historical behavior and probability distributions.