*2.1. Data Used for Analysis*

Historical yearly streamflow data (1950–2015) for eleven California rivers (Sacramento River, Feather River, Yuba River, Bear River, American River, Mokelumne River, Calaveras River, Stanislaus River, Tuolumne River, Merced River, and San Joaquin River) were collected from Cal-Adapt which is an online system to access data and information on climate impacts in California [8]. These eleven river stations were selected by Pierce et al. [6] who conducted climate impact assessment and identified these stations as locations in the variable infiltration capacity (VIC) hydrologic model as bias corrected to unimpaired flows that are common to the eleven stations used by California Department of Water Resources. These stations represent flows in the absence of human activities.

Future scenarios of climate (e.g., temperature and precipitation) have been downscaled from 100 km to 6 km resolution using the LOCA statistical downscaling method for California and Nevada to be used in climate assessments [9]. Pierce et al. [6] evaluated the possible effects of the changing climate on the eleven unimpaired streamflow under eight climate simulations representing four different climate models and two different climate scenarios. The eight projections include four generalized circulation model (GCM) simulations/models at two emission levels as representative concentration pathways (RCP) 4.5 and 8.5. The emission level of RCP 4.5 is an intermediate climate change scenario, which will occur if policy makers enact mitigative policies that will minimize greenhouse gas emissions [6]. The RCP 8.5 pathway incorporates the highest level of greenhouse gas emissions resulting from a large population with high energy needs devoid of climate change policies [10]. There is no greater likelihood of occurrence for RCP 4.5 or 8.5 [11]. The four simulations are from models representing different future conditions-HadGEM2-ES (Warm Dry), CNRM-CM5 (Cool Wet), CanESM2 (Average), MIROC5 (Other). The MIROC5 (Other) simulation is a model most unlike the first three to provide inclusion of as many different possibilities. These four models were identified in the Cal-Adapt study to best represent the climate in California. [6]. Eight future projected water flows (2020–2099) were collected from Cal-Adapt for each of the eleven streamflow stations (see Figure 1). Yearly streamflow data were transformed from ft3/sec to million-acre feet of water per year (MAF) (Note: 1 MAF = 1233 million cubic meters MCM) to be used in the drought analysis.

**Figure 1.** Map of California streamflow station locations used in this study. Map generated from ArcGIS Hub and station coordinate obtained from National Water Information System [12].
