Diminished Flows in Southwestern Snow-Fed Rivers: Assessing and Normalizing Climate Change Projections for Use in Heavily Managed Hydrologic Systems

Climate model output has been coupled to surface water models to generate projections of future streamflow. Not surprisingly, projections of warmer temperatures in the 21st Century generally lead to diminished flows in major snow-fed river systems across southwestern North America. However the spread of model-projected flows is very large in important basins such as the Colorado and Rio Grande. Adapting to the impacts of climate change on surface water resources in the semiarid Southwest would be greatly facilitated both by improvements in our understanding of the physical processes that control runoff generation (to reduce model uncertainty), and by incorporating existing water management impacts on downstream flows into climate change projections.

Focusing on the Rio Grande, we have developed metrics with the aim of assessing CMIP5-based simulations based on their demonstrated ability to reproduce historical changes in snowpack and climate/streamflow covariances in the relatively unimpaired headwaters basin, during recent decades which have exhibited pronounced increases in temperature and decreases in snowpack such as are expected to continue in future decades. Our analysis is used to diagnose the effects of changing snowpack, temperature and precipitation on snowmelt runoff in both observations and model simulations. We find that the models that best reproduce observed changes in the historical period tend to be those that simulate pronounced decreases in future flows.

Farther downstream, all model-projected flows are utterly unrealistic at face value, because the natural flow regime simulated by climate models does not include anthropogenic impairments (mostly diversions for irrigated agriculture) that reduce downstream flows dramatically. We have developed a statistical method to parameterize upstream water management that reduces downstream flows to realistic levels. Our management-normalized flows are then suitable for studies of the impacts of climate change on reservoir levels and surface water availability in downstream reaches where flows are greatly reduced from their natural level. Furthermore, the normalization constants based on observed flows can be maintained into the future, or can be modified to describe potential changes to upstream management and diversions. Explicitly incorporating water management in this simplified way provides a means to jointly assess the effects of climate change and water policy changes on future flows in managed river systems.