USGS - science for a changing world
California Water Science Center

Regional Hydrology

This website has been archived. For more up to date information on Basin Characterization Model (BCM), please visit https://www.usgs.gov/centers/california-water-science-center/science/basin-characterization-model-bcm

Simulated Unimpaired Mean Daily Streamflow in the Russian River and Upper Eel River Basins, California, under Historical (1910-2013) and Projected Future (2001-2099) Climates

The Sonoma County Water Agency (SCWA) has a need for information regarding ongoing and future changes in streamflow, in order to provide input to revisions of the outdated Decision 1610 issued by the State Water Resources Control Board in 1986. Future streamflow research is necessary so that SCWA can optimize the balance of water needs among instream flows, flood protection, and public water supply on the basis of the most scientifically defensible information that reflects potential future hydrologic conditions. As with most local water agencies throughout the country, the SCWA has a need to plan for water-resource changes that may occur as a result of changing climate or climate variability. This dataset was developed as input to their water management model so they can assess water management and delivery options under various management scenarios and future climate scenarios.


Figure 1. Map of Russian River study area illustrating the model boundaries for flow nodes and calibration, rivers and streams, and the locations of the flow nodes and calibration gages.

Figure 1. Map of model domain with flow nodes and calibration basins.

Purpose and Scope

Water demand and allocation in the Russian River basin for human, agricultural, and ecological benefits is becoming increasingly challenging in the 21st century. Water-resource managers and regulators need tools and datasets to plan for changes in water resources that may result from projected climate change and variability. The purpose of this study was to estimate unimpaired daily flows for the Russian River, California, using the Basin Characterization Model (Flint and others, 2013) that has been revised to a daily timestep and calibrated to historical measured flow and using projections of future climate from Global Climate Models. The dataset was developed in cooperation with SCWA.

The SCWA requires information for long-term water management planning. However, this data release is being provided prior to formal publication of the dataset because SCWA has used the data as input to a reliability study that addresses the vulnerability of water storage in Lake Mendocino (fig. 1) that will be released this spring to the California State Water Resources Control Board (SWRCB). Lake Mendocino is located in the East Fork of the Russian River. Flows from Lake Pillsbury in the Eel River are diverted to the East Fork via the Potter Valley Project and are temporarily stored in Lake Mendocino. Releases from Lake Mendocino sustain spring, summer, and fall flows in the upper Russian River. Depletion of water storage in Lake Mendocino could reduce reservoir releases and affect the quantity and quality of upper Russian River flows. Improved understanding of the hydrologic response to climate is a high priority in the Russian River because of likely more-restrictive limits on diversions, increased populations, updated biological requirements, increasing climate variability, and changing land uses. The SCWA is preparing the reliability study, at the request of the SWRCB, with the goal of optimizing the amount of water allocated for instream flows and public water supply by using the most scientifically defensible information that reflects potential changes in mean daily streamflow under future hydrologic conditions. Long-term uses of the dataset are for input to the SCWA water management model (HEC-RESSIM; Klipsch and Hurst, 2007) to assess various scenarios for water management throughout the basin, including changes in land use, population, climate, and policy.

This data release includes two datasets: (1) a daily time-series of simulated unimpaired streamflow from 1/1/1910 to 12/31/2013 based on historical climate input, and, (2) a daily time-series of simulated unimpaired streamflow from 1/1/2001 to 12/31/2099 for four different future climate scenarios. The two datasets include simulated flows for a series of node locations representing all of the Russian River's largest tributary basins and four segments of the mainstem.

The unimpaired flows in this dataset were used to develop an understanding of how streamflow responded to changes in historical climate and water demand and for a range of future climate projections. Mean daily unimpaired streamflow was estimated by using a daily water-balance model that was calibrated to historical daily and monthly streamflow; estimates of stream losses owing to agricultural demand; and, regression analyses. These data will support water-resource planning and numerous ongoing flow-ecology and climate-vulnerability studies.

Methods

Development of the Russian River mean daily unimpaired flows used a modified version of a published monthly water-balance model (Basin Characterization Model, BCM; Flint and others, 2013), and a series of post-processing steps, to estimate unimpaired streamflow from recharge and runoff calculated by the BCM. The BCM is a regional water-balance model that deterministically calculates how precipitation is converted into infiltration into soils, evapotranspiration, runoff, and percolation (below the root zone) that recharges the groundwater system. Climate inputs, bedrock permeability, and soil properties (fig. 2), all components of the daily water budget, and hydrologic response variables are calculated at a 270-m grid cell resolution. Hydrologic response variables include recharge and runoff that are used to calculate basin discharge, and climatic water deficit (CWD), a variable related to crop water demand that is used in calibration.

Figure 2. Spatial properties used for developing unimpaired flows for the Russian River Basin using the Basin Characterization Model.
Figure 2. Spatial properties used for developing unimpaired flows for the Russian River Basin using the Basin Characterization Model.

Figure 2. Spatial properties used for developing unimpaired flows for the Russian River Basin using the Basin Characterization Model, (A) geologic bedrock and bedrock permeability, K, and (B) total soilstorage.

Development of Daily Climate Inputs

Downscaled historical climate data were developed by using daily station data and monthly data from the Precipitation-Elevation Regression on Independent Slope Method (PRISM; Daly et al 2008). These data were used with a method that is modified from that described in Flint and Flint (2012) in order to incorporate daily station data (Flint and others, in preparation). Future-climate data were developed by using methods developed and described by Flint and Flint (2012).

Estimating Daily Basin Discharge

In order to develop daily unimpaired flows using a series of calibrations, daily basin discharge was estimated using the BCM for 10 Russian River reaches and 2 Eel River reaches defined by the contributing area upstream from 12 USGS gaging stations ( fig. 1; table 1). Daily results for all grid cells upstream of the stream gage were summed to create time series for runoff and recharge. To transform these results into a form that can be compared to the pattern and amount of gaged streamflow, the water balance is conceptualized as consisting of three groundwater reservoirs that are hydraulically connected (Flint and others, 2013). The surface reservoir is responsive to daily storm events and snowmelt. The shallow groundwater reservoir consists of the shallow saturated zone that varies seasonally and provides much of the dry-season baseflow. Following large storm events, the shallow groundwater reservoir can account for much of the recession flow that occurs. The deep groundwater reservoir represents the regional aquifer in most locations and may contribute some flow to the shallow groundwater reservoir over long time frames.

A series of equations describing the various parts of the hydrograph were derived (Flint and others, 2013) to optimize the match between the simulated and measured daily hydrograph for the various seasons when surface flows, shallow flows, and deep flows have proportionally different effects on hydrograph responses. These empirical discharge equations use exponential notation and scaling factors to define recession flows and the system losses to groundwater. For additional details see Flint and others (2013).

Calibration of Daily Streamflow Model

Figure 3(A). daily time series from water year 1975 to 1999
Figure 3(B). exceedance probability, annual discharge, and cumulative annual discharge for winter flows (December to February) from 1958 to 2013
Figure 3(C). daily discharge before and after conditioning.

Figure 3. Goodness-of-fit examples that compare modeled basin discharge estimated using the Basin Characterization Model to measured streamflow at Healdsburg, California, (A) daily time series from water year 1975 to 1999; (B) exceedance probability, annual discharge, and cumulative annual discharge for winter flows (December to February) from 1958 to 2013; and (C) daily discharge before and after conditioning.

Converting the BCM estimated basin discharge into unimpaired flows was performed in three steps by using measured tributary and mainstem flow data.

Step 1: Calibrate Monthly Flows
In the first calibration step, the BCM-calculated monthly flows were calibrated to measured unimpaired flows from the upland tributaries ( fig. 1);
  • 11460940 Russian R at Redwood Valley
  • 11462700 Feliz Creek
  • 11464860 Warm Springs Cr nr Asti CA
  • 11463900 Maacama Cr nr Kellogg CA
  • 11465000 Dry Creek at Warm Springs Dam

During this initial calibration, bedrock permeability (fig. 3) was adjusted iteratively until acceptable relative proportions of recharge and runoff were achieved to provide matches to the monthly hydrographs for these upstream gages (Flint and others, 2013). The resulting bedrock permeability map developed from the first calibration step using the monthly model was then used in the application of the daily BCM measured and estimated flows (fig. 3A).

Step 2: Estimate Streamflow Losses

The second calibration step was to develop regression equations to estimate monthly gage-to-gage losses in streamflow related to agricultural demand for each of the 10 Russian River tributary and mainstem reaches. This was done to account for seasonal losses-to-demand in the calibration and to ensure more accurate representation of unimpaired flows during the irrigation season. Regression coefficients were selected iteratively to optimize the match between total simulated and measured streamflow volume at flow nodes. This was done using measured streamflow data and aggregated seasonal volumes to accurately represent flow seasonality (fig.3B). Streamflow loss due to agricultural demand was correlated to CWD, a hydrologic response variable that represents the seasonal soil moisture deficit calculated as potential minus actual evapotranspiration. The correlation between gage-to-gage loss and CWD was used to estimate agricultural demand during the primary irrigation season from May to October and to develop an equation that can be used to estimate streamflow losses due to agricultural demand under future climates. The estimated seasonal loss of streamflow related to agricultural demand was subtracted from the BCM discharge estimates to optimize the match to the measured hydrograph at each stream gage. Total volumes of measured and simulated flows for 1958-2013 matched to within 0.5 percent for each flow node.

Step 3: Optimization Method

In the third calibration step, a conditioning method was developed to optimize the match between simulated and measured streamflow that addressed the lack of fit for both high and low flows (fig. 3C). Unimpaired flows for periods with relatively little agricultural demand, November through April, were empirically estimated and extrapolated to the impaired season, May through October, by using a regression method available in Matlab (Mathworks; Bootstrap AGGregatING ; Aslam and others, 2007). This regression tool correlates the unimpaired-flow error (empirically estimated unimpaired flow minus the BCM unimpaired flow) for each flow location to the estimated BCM unimpaired flows, to reduce the error, bias, and variance between measured and simulated estimates.

Tabulated calibration results are shown for 12 reaches, along with USGS stream gages and overall goodness-of-fit results, in table 1. Goodness-of-fit results are not included for Dry Creek, owing to only having seasonal measurements, or for Mark West, owing to the short period of record. Estimates for these two reaches were developed iteratively in the calibration for the Guerneville reach. Goodness-of-fit for daily flows is generally poor for reaches with dam operations–daily calibration statistics are not applicable at gages below dams because the simulated unimpaired flows are not intended to represent managed flows. However, monthly and annual calibration statistics are appropriate at these gages. Daily peak flows are consistently underestimated by this daily model because of the inability to characterize the localized peak rainfall volumes with few raingages, and because of the inability of the model to represent runoff in response to high intensity, hourly rainfall by using a daily time step. In contrast to daily flows, the monthly volumes are well represented (table 1). Monthly flow volumes represent time periods during which flows are capable of filling surface water reservoirs managed by SCWA, and low flows are reasonably well represented in the mainstem channel reaches.

Unimpaired Streamflow under Historic Climate

Historical Unimpaired Flows (Provisional)

Excel (13.7 MB)

Tab Delineated (3.2 KB)

Unimpaired flows are represented as local flows (between flow nodes) in the spreadsheet. These are represented in figures 4, 5, 6, 7 as average daily for 1981-2010 in comparison to future mid-century and end-of-century 30-year periods. Data are available as downloads of Excel spreadsheets or space-delimited text files.

Unimpaired Streamflow under Projected Future Climates

Future Unimpaired Flows (Provisional)

Excel (51.4 MB)

Tab Delineated (10.7 KB)

The calibrated BCM was used to estimate daily unimpaired streamflow by using four future daily climate projections from the 4th IPCC Assessment (IPCC, 2007). We selected four climate projections capable of simulating the recent historical climate, particularly the distribution of monthly temperatures and the strong seasonal cycle of precipitation that exists in the region (Knowles and Cayan, 2002; Cayan and others, 2008, 2009). The four scenarios include two global climate models (GCMs) and two greenhouse gas emissions scenarios. We selected the Parallel Climate Model (PCM) developed by National Center for Atmospheric Research (NCAR) and Department of Energy (DOE) (Washington and others 2000; Meehl and others, 2003), and the National Oceanic and Atmospheric Administration (NOAA) Geophysical Fluid Dynamics Laboratory CM2.1 model (GFDL) (Delworth and others, 2006; Stouffer and others, 2006). The choice of emissions scenarios included A2 (this scenario represents a continuation of current emission practices, referred to as a "business as usual" scenario) and B1 (this scenario represents mitigated emissions relative to current practices). These climate projections were downscaled and applied to the calibrated BCM. Mean annual hydrographs and cumulative daily streamflow are shown for the historical 30-year period (1981-2010) and mid- and late- century 30-year periods (2040-2069 and 2070-2099) for each projection for the two basins upstream of reservoirs (East Fork and Warm Springs Dam) and for two mainstem gage locations representing upper and lower Russian River basin reaches (Healdsburg, and Guerneville) (figs. 4, 5, 6, 7).


Figure 4. Mean daily streamflow (left column), and mean cumulative streamflow (right column), is shown as 30-year mean annual hydrographs for GFDL A2 climate projection for East Fork (EF), Warm Springs Dam (WSD), Healdsburg (HLDS), and Guerneville (GR) reaches.

Figure 4. Figure showing 30-year mean annual hydrographs for GFDL A2 climate projection.

Figure 5. Mean daily streamflow (left column), and mean cumulative streamflow (right column), is shown as 30-year mean annual hydrographs for GFDL B1 climate projection for East Fork (EF), Warm Springs Dam (WSD), Healdsburg (HLDS), and Guerneville (GR) reaches.

Figure 4. Figure showing 30-year mean annual hydrographs for GFDL B1 climate projection.

Figure 6. Mean daily streamflow (left column), and mean cumulative streamflow (right column), is shown as 30-year mean annual hydrographs for PCM A2 climate projection for East Fork (EF), Warm Springs Dam (WSD), Healdsburg (HLDS), and Guerneville (GR) reaches.

Figure 6.  Figure showing 30-year mean annual hydrographs for PCM A2 climate projection.

Figure 7. Mean daily streamflow (left column), and mean cumulative streamflow (right column), is shown as 30-year mean annual hydrographs for PCM B1 climate projection for East Fork (EF), Warm Springs Dam (WSD), Healdsburg (HLDS), and Guerneville (GR) reaches.

Figure 7. Figure showing 30-year mean annual hydrographs for PCM B1 climate projection.

References

Aslam, J.A. Popa, R.A., and Rivest, R.L., 2007, On estimating the size and confidence of a statistical audit: Proceedings of the Electronic Voting Technology Workshop (EVT '07), Boston, MA.

Cayan, D.R., Maurer, E.P., Dettinger, M.D., Tyree, M., Hayhoe, K., 2008, Climate change scenarios for the California region: Climatic Change 87(Iss 1 Sppl):21- 42. doi 10.1007/s10584-007-9377-6.

Cayan, D.R., Tyree, M., Dettinger, M.D., Hidalgo, H., Das, T., 2009, California climate change scenarios and sea level rise estimates for the California 2008 Climate Change Scenarios Assessment: California Energy Commission Report No. CEC-500-2009- 014-F.

Delworth, T.L. and others, 2006, GFDL's CM2 global coupled climate models. Part 1. Formulation and simulation characteristics: Journal of Climate 19:643-674.

Flint, L.E. and Flint, A.L., 2012, Downscaling climate change scenarios for ecologic applications: Ecological Processes, 1:1.

Flint, L.E., Flint, A.L., Thorne, J.H., and Boynton, R., 2013, Fine-scale hydrologic modeling for regional landscape applications: the California Basin Characterization Model development and performance: Ecol. Processes 2:25.

IPCC, 2007, IPCC Fourth Assessment Report: Climate change 2007; synthesis report. Core writing team, Pachauri, R. K. and Reisinger A. (Eds). International Panel on Climate Change, Geneva Switzerland.

Klipsch, J.D., and Hurst, M.B., 2007, HEC-ResSim Reservoir System Simulation User's Manual Version 3.0: USACE, Davis, CA, 512 p.

Knowles, N. and Cayan, D., 2002, Potential effects of global warming on the Sacramento-San Joaquin watershed and the San Francisco Estuary:, Geophysical Research Letters 29(18):1891. doi:10.1029/2001GL014339.

Meehl, G.A., Washington, W.M., Wigley, T.M.L., Arblaster, J.M., and Dai, A., 2003, Solar and greenhouse gas forcing and climate response in the twentieth century: Journal of Climate 16(3):426-444.

Stouffer, R.J. and others, 2006, GFDL's CM2 global coupled climate models. Part 4. Idealized climate response: Journal of Climate 19:723-740.

Washington, W.M., Weatherly, J.W., Meehl, G.A., Semtner, A.J., Bettge, T.W., Craig, A.P., Strand, W.G., Arblaster, J., Wayland, V.B., James, R., and Zhang, Y., 2000, Parallel climate model (PCM) control and transient simulations: Climate Dynamics 16:755-774.

Data

Historical Unimpaired Flows (Provisional)

Excel (13.7 MB) | Tab Delineated (3.2 KB)

Future Unimpaired Flows (Provisional)

Excel (51.4 MB) | Tab Delineated (10.7 KB)

Table 1: Model calibration parameters and goodness-of-fit statistics

Excel (43 KB)


Cooperating Agencies

Sonoma County Water Agency


Citation

Flint, L.E., Flint, A.L., Curtis, J.A., Delaney, C., and Mendoza, J., 2015, Provisional simulated unimpaired mean daily streamflow in the Russian River and Upper Eel River Basins, California, under historical and projected future climates: U.S. Geological Survey Data Release, doi.org/10.5066/F71C1TX4