UCLA

Understanding 21st century changes in California surface hydrology is critical to ensuring enough freshwater resources for the state’s municipal, ecological and agricultural purposes and assessing future ecosystem health and wildfire risk. To project 21st century surface hydrology over California – a region with highly complex topography that is not well captured by global climate models (GCMs) – downscaling is necessary. This work projects future changes in surface hydrology over the Los Angeles and Sierra Nevada regions through dynamical and statistical downscaling techniques.

Dynamical downscaling is employed over Los Angeles to produce 2-km resolution regional projections for the mid-21st-century under an aggressive warming scenario. These projections reveal annual mean runoff and actual evapotranspiration are nearly insensitive to warming. This insensitivity is an artifact of the region’s Mediterranean-type climate: Because the warm season receives almost no precipitation, the strongest warming-induced potential evapotranspiration enhancement coincides with dry soils, severely constraining actual evapotranspiration increases. This surprising result highlights that this important semi-arid region is less susceptible to long-term changes in runoff and soil moisture due to its Mediterranean climate.

Over the Sierra Nevada Mountains, dynamical downscaling is used to produce high-resolution (3-km) simulations of end-of-21st-century surface hydroclimate. The high resolution and physical realism of these simulations provides unprecedented detail into the elevational dependence of hydroclimate changes and allows us to examine hydroclimate changes at the watershed level. These downscaled simulations reveal future warming leads to a shift toward significantly earlier snowmelt-driven surface runoff timing at each elevation throughout the Sierra Nevada, particularly in mid-elevations (2000-2750m) in the western and northern Sierra. Moreover, these projections show that any precipitation increases are outweighed by warming induced snowpack reductions and evapotranspiration increases, resulting in statistically significant drying of spring and summer soils and a substantial lengthening of the summer dry period. Relationships and patterns that emerge through dynamical downscaling over the Sierra Nevada are exploited to build simple statistical models that mimic dynamical model behavior. Using this hybrid dynamical-statistical downscaling model, high-resolution end-of-21st-century runoff timing and soil moisture changes are projected for all available GCMs from phase 5 of the Coupled Model Intercomparison Project and the four forcing scenarios adopted by the Intergovernmental Panel on Climate Change’s Fifth Assessment Report. These multi-model projections allow us to quantify and characterize ensemble-mean changes and the associated uncertainty due to inter-model GCM spread, as well as the consequences associated with choice of emissions scenario. Averaged across the Sierra, April-September soil moisture is projected to decrease 17.1% in the 35-model ensemble mean under RCP8.5 (with an approximate intermodel range of -12.9% to -21.0%), but only 9.1% with an approximate intermodel range of -5.7% to -12.9%) under RCP4.5, a reasonable mitigation scenario.