Figure 2. Features of the Sacramento and Feather Rivers

The principal objective of the Sacramento River Modeling Project was to provide an analytical capability to quantitatively assess consequences of specific temperature regulation measures, either structural or operational, designed to enhance salmon productivity in the Sacramento and Feather Rivers. This entailed selecting and adapting appropriate existing mathematical models to the headwater reservoirs and the river systems downstream; assembling basic data on system geometry, hydrology and meteorology; monitoring stream temperatures at selected locations (see Riparian Vegetation Study, below); calibrating and verifying the models; and demonstrating their use on the Sacramento and Feather River systems.

Three mathematical models were utilized in the project to simulate the temperature regimes of the headwater reservoirs, regulating reservoirs, and the downstream river reaches. System components modeled included Shasta and Trinity Reservoirs, Keswick Reservoir, and the Sacramento and Feather Rivers. For Shasta and Trinity Reservoirs the one-dimensional finite difference model, WQRRS (Water Quality for River-Reservoir Systems), was applied. Review of field data indicated that during spring and summer periods both reservoirs become strongly stratified, thus restricting the availability of cold water at the levels of fixed outlets in Shasta and Trinity Dams. A one-dimensional conceptualization of such a water body, where vertical temperature gradients dominate inflow and outflow, as is represented in WQRRS, was considered appropriate. Given the requisite boundary conditions of flow and heat energy, the model creates a history of thermal stratification within the reservoirs and their release flows and temperatures which are then used as input to the stream models.

Implementation of WQRRS for the headwater reservoirs entailed a sequence of steps: description of the geometric configuration of the impoundments and arrangement of outlet facilities; determination of boundary conditions of flows and water temperatures; estimation of meteorological conditions in the reservoir environment; and scheduling of reservoir operation. The reservoir is conceptualized in WQRRS as an assemblage of horizontal slices of uniform thickness, the volumes of which are determined by their elevations in the impoundment. Flows are introduced at levels in the reservoir where temperatures correspond to those of the inflows, and release temperatures correspond to fixed outlet elevations. Hydrologic balance for the impoundment is achieved by mass balance.

Net heat exchange at the air-water interface for both reservoirs and streams was determined by estimation of the sum of five component fluxes:

• short wave solar radiation

• long wave atmospheric radiation

• long wave radiation from the water surface

• evaporative heat flux

• sensible heat flux between air and water.

These were formulated in accord with the methods employed originally by the TVA (1972) and incorporated into both WQRRS and RMA-11. Meteorological data required, e.g., wind speed, air temperature, relative humidity, cloud cover, etc., were obtained from meteorological stations at Redding and Sacramento airports. Temperature changes between volumetric elements in the reservoirs and streams were calculated by summing advective and dispersive transport of heat along the principal axes of the reservoirs (vertically) or streams (longitudinally), heat exchange at the air-water interface, and other sources or sinks of heat energy, e.g., tributary inflows.

Each of the models developed in this project was calibrated first to a data set of actual field observations, adjusting appropriate model parameters until acceptable agreement was achieved between model simulation and measured field observations. The models were then verified against a second data set, retaining the parameter values determined in calibration. Once verified, the models were considered available for application in the investigation of the effects of temperature control strategies for the protection of salmon. The processes of calibration and verification, as they were carried out in the project, are described briefly below.

WQRRS for Trinity Reservoir was calibrated using 1987 water temperature profile data provided by the USBR. Adjustments were made to the coefficients for evaporative heat flux, water column stability, and dispersion to achieve good agreement between model and field data over the period of greatest stratification and in the strata near the withdrawal elevation. Verification of model performance was achieved using a data set from 1993 that was derived from several sources. Due to limited available local meteorological data for 1993, a minor adjustment to an evaporative heat flux coefficient was required to reflect different the data sources used for the verification period.

To simulate temperatures in the main stem of the Sacramento River it was necessary to define the river's upstream boundary conditions, i.e., flows and temperatures at the discharge from Keswick Reservoir. These properties of the system are governed primarily by the meteorological conditions, thermal regime of Shasta Reservoir, operation of the Shasta power plant and low level outlets, temperature changes occurring in Keswick Reservoir, and transfers through Whiskeytown Reservoir and Spring Creek Power Plant.

In this project Keswick Reservoir was treated as a vertically mixed stream, and as such was simulated with RMA-2 and RMA-4Q. In a previous study the reservoir was simulated as a two-dimensional stratified system, but because vertical stratification was weak it was determined that the most appropriate treatment of Keswick Reservoir was as a deep slow river – the same approach adopted for the main stems of the Sacramento and Feather Rivers. The resulting Keswick Reservoir model was comprised of 32 one-dimensional elements and 65 nodes along its principal axis.

The Sacramento River was represented in model form as a system of 718 elements and 1437 nodes. Elements were uniformly 500 m (1640 ft) in length and of trapezoidal cross section with variable widths and side slopes. Upstream boundary conditions were defined by Keswick Reservoir outflow conditions (flows and temperatures), either directly observed or simulated by the Keswick Reservoir model. Downstream boundary conditions were prescribed explicitly by a stage-discharge relationship at Verona near the river's confluence with the Feather River.

Calibration of the Sacramento River model for hydrodynamics was complicated by the requirements for water balance along the axis of the river. Accretions and depletions, e.g., irrigation diversions and return flows, had to be independently specified for each simulation period. Also, changes in the value of Manning's roughness coefficient (n) affected simulated flows at key gage locations. Sensitivity analysis indicated that a Manning’s n of 0.030 produced the best agreement between model flows and observed flows for August 1993. Verification was achieved for data of August 1994. Since flow data for both periods were of comparable magnitude it may be expected that the models would be best applied with a Manning’s n equal to 0.030 during the summer period when flows are regulated by project operation. Some adjustment in the Manning coefficient may be needed for much greater flows. At this point in the process the hydrodynamic model was considered ready to provide the input necessary for temperature simulation. Calibration results for the Sacramento River hydrodynamic model are shown in Figure 5.

Calibration for temperature was initially attempted using a limited set of observations for six stations situated along the river where data were reported only as daily means, maxima, and minima. Headwater temperatures were available from results of Keswick Reservoir simulated outflows to define the upstream boundary condition. Heat fluxes along the river were generated from the heat budget equations incorporated in the water temperature model, RMA-11. Temperature data for tributary flows were not available, so it was necessary to use the equilibrium temperature concept to estimate the temperatures of accretions from these sources. Temperature profiles along the river generated by the model for both August 1993 and August 1994 conditions agreed closely with the available mean, maxima, and minima water temperature data at six observation points. Typical results are shown in Figure 6.

The process of adapting the RMA models to the Feather River followed a sequence similar to that for the Sacramento River, although the system was somewhat less complicated, with fewer tributary flows. A finite element grid comprised of 272 elements and 549 nodes was generated, including a special junction to accommodate flows from Thermalito Afterbay and the headwater conditions below Oroville Dam. Elements were approximately 400 m (1312 ft) in length and of trapezoidal cross section with variable widths and side slopes.

Figure 6. Calibration of temperature simulation along the Sacramento River for August,1994

Upstream boundary conditions for the hydrodynamic model were prescribed by time series of flows at the Fish Barrier Dam, Thermalito Afterbay, and the two main tributaries, the Yuba and Bear Rivers. A downstream hydrodynamic boundary condition was defined by a time series of river stage at Verona on the Sacramento River near the confluence of the two rivers. Satisfactory calibration and verification were achieved for conditions of August 1993 and August 1994, respectively, using a Manning’s n equal to 0.050.

Time series of temperatures at the afterbay outlet served to define the upstream temperature boundary condition for the Feather River model. Temperature observations at Gridley and near White Oak Ranch were utilized in calibration and verification. Meteorological data from the Sacramento airport were used in RMA-4Q to predict heat fluxes. Results from a typical temperature calibration for the month of August 1993 are shown in Figure 7. Results indicate close agreement between model and field data, both with respect to diurnal and day-to-day variations.

Restoration of cold water fisheries requires assessment of management alternatives that can modify the temperature regime of the riverine environment. Among these alternatives are measures to restore riparian forests that could provide important thermal refugia for fish. Shade and cool air temperatures from the riparian forest may extend across part of the water surface, decreasing net heat flux at the air-water interface. The presence of vegetation and associated large woody debris along the river margins can alter local velocities, sustaining natural cycles of erosion and deposition, thereby increasing the capacity of the riverine system to maintain a favorable thermal environment for salmonids and their supporting ecosystems. A field study designed to evaluate the interaction of riparian vegetated zones on the thermal regime of the Sacramento River was included as part of the Sacramento River Temperature Modeling Project. Preliminary findings from the initial phase of the Riparian Vegetation Study and their contributions to the overall goals of the project are described briefly below.

A representative study reach of the Sacramento River was selected for study to quantitatively define the influences of riparian vegetation on water temperature conditions in habitats preferred by cold water fish. The reach extended over about 58 km (36 mi) of the river between Woodson Bridge State Park (River Mile (RM) 220) and Ord Ferry Bridge (RM 184) and included areas of both dense and sparse vegetation along the river banks. The field study was comprised of two major components: monitoring of water temperatures at selected locations along the main stem of the Sacramento River and monitoring of atmospheric conditions in vegetated and non-vegetated zones along the river.

Water temperatures were monitored using temperature loggers that recorded continuously at 15 minute intervals with a precision of ± 0.2° C (0.36° F). Loggers were deployed in near-bank locations, in midstream, and in backwater locations as needed to identify the temperature characteristics of the river. Atmospheric conditions were monitored at a weather station equipped for continuous recording of air temperature, wind speed and direction, solar radiation, and relative humidity.

Instrumentation was deployed during the summers of 1994 and 1995 for periods of several months, sufficient to characterize the significant dynamics of flow, temperature, and atmospheric conditions within the study reach. Data were down-loaded from temperature loggers periodically for analysis. Some preliminary results illustrate the nature of the survey.

Data were processed in weekly sets of 672 data point to produce average diurnal patterns of water temperature variation. A typical sequence of averaged diurnal observations of temperatures mid-stream and along an unshaded bank at RM 192 are shown in Figure 8, for a period in September, 1994. In this example the water near the unshaded bank exhibits significantly greater temperatures than at midstream, by as much as 0.5° C (0.9° F) or more during daylight hours.

A unique discovery of the field study was the occurrence of zones of minimal diurnal temperature variation, spaced at intervals along the river corresponding to one day’s travel time at the mean flow velocity. Figure 9 illustrates this phenomenon in the pattern of diurnal water temperature variation at two locations, RM 192 and RM 185, during August, 1994. At RM 192 the pattern is the expected diurnal cycle with an amplitude of about 1.5° C (2.7° F), while at RM 185, 11 km (7 mi) downstream, diurnal variation in water temperature is almost zero. It was discovered also that these "nodes" of minimal diurnal temperature variation (not related to finite element model nodes) are present at multiple locations along the longitudinal axis of the Sacramento River, separated by distances roughly equivalent to one day’s travel time at average velocities. In addition to nodes of minimum diurnal temperature variation, "anti-nodes" of maximum diurnal temperature variation are also present, offset from the nodes by one half day’s travel distance along the river. Moreover, the positions of nodes and anti-nodes are governed implicitly by flow rate, a characteristic that proved to be of special value in calibrating RMA-2, the hydrodynamics model.

Results of the preliminary field study revealed that the effects of riparian vegetation on water temperatures are difficult to discern on a river system as large and complex as the Sacramento River and that there is a need for more detailed characterization of the processes governing heat exchange between the atmosphere and the water column. This requires not only more precise instrumentation, but greater spatial resolution of the study domain. Also, it requires more information on the hydrodynamics of the riverine environment that governs transport of heat by advective and dispersive processes. These requirements are being addressed in a new project sponsored by the United States Fish and Wildlife Service and also by the California Department of Fish and Game. The project will be extended through 1997, incorporating field data from campaigns in the summers of 1995, 1996 and 1997.

The objective of the Sacramento River Temperature Modeling Project was to produce workable tools - mathematical models - of the principal features of the Sacramento and Feather River systems that could be used to quantitatively evaluate the effects of structural or operational alternatives on improving the habitat of salmon species at risk. These tools were developed for the systems of concern, reservoirs and river reaches, and were calibrated and verified against field observations. Their capabilities for selected scenarios are described below.

Choice of alternative scenarios was governed by the Central Valley Project Operations Criteria and Plan (CVP-OCAP), defining a set of twelve month long operation studies based on different hydrologic conditions, carry-over storage quantities, and operating criteria. Many combinations of hydrology, storage, and operation criteria were possible from which to demonstrate the capabilities of the models to simulate temperatures in the Sacramento River below Keswick Dam. A small set of scenarios was selected to be representative of a range of hydrologic conditions; Wet, Above Normal, Dry without transfer through Spring Creek Tunnel, and a special case for Dry year conditions with inter-basin transfers from the Trinity Basin via Spring Creek Tunnel. Based upon the Sacramento River Index, Wet, Above Normal, and Dry years have 75, 50, and 30 percent probabilities of non-exceedence, respectively. Two different levels of carry-over storage in Shasta Reservoir were initially considered: High-Medium and Low-Medium; however, only Low-Medium storage was used to demonstrate the application of the models to Keswick Reservoir and the Sacramento River. Scenarios and system components are outlined in Table 1.

Hydrologic boundary conditions, i.e., inflows to the system, were prescribed by the selection of hydrologic cases. Temperature boundary conditions for tributary inflows to Shasta Reservoir as well as major tributaries in the Sacramento River system were defined by assuming that the inflows were at equilibrium temperature. Temperatures of inflows to Keswick Reservoir from Spring Creek Tunnel were taken as measured values from August
1994. Meteorological data for calculation of heat fluxes at the air-water interface were
derived from 1994 records at the Redding airport for Shasta and Keswick Reservoirs and the river reach downstream to Vina. Data from Sacramento airport were applied for the reach from Vina to the confluence of the Sacramento and Feather Rivers.

Table 1. System components modeled under selected scenarios

Starting at Shasta Reservoir, WQRRS was run to determine release temperatures based on the simulated temperature profiles during the month of August, as illustrated in Figure 10, for a Dry year at Low-Medium carry-over storage. It is noted that at the level of the power release, temperatures varied over the month from about 11°C (51.8°F) to 15°C (59.0°F). However, lower temperatures in the vicinity of the low level outlets, about 8.5°C (47.3°F), indicate that a significant volume of cold water is available to the temperature control device which may provide release temperatures more favorable to salmon spawning downstream.

Shasta Reservoir simulation provided the upstream boundary conditions for Keswick Reservoir and the hydrodynamic and temperature models were then run to simulate temperatures at the Keswick Dam outlet, thus defining the upstream boundary condition for the Sacramento River model. Average daily simulated temperatures at the outlets of Shasta and Keswick Reservoirs obtained for four selected operation scenarios are summarized in Table 2 for August 10 and August 20.

Figure 10. August temperature profiles in Shasta Reservoir as simulated by WQRRS: Dry year at Low-Medium carryover storage

Table 2. Average daily simulated release temperatures from Shasta and Keswick Reservoirs for Wet, Above Normal, and Dry years: August 10 and August 20

Model results suggest possible difficulties in delivering sufficient cold water from the upstream projects to assure compliance with the 13.3°C (56.0°F) temperature target in the upper reach of the river below Keswick Dam, especially in Dry years. The potential difficulty in this case seems to be associated with Shasta Reservoir itself. As illustrated in Table 2, Shasta Dam release temperatures change significantly at the power penstock elevation between August 10 and August 20. A likely solution may depend on the effectiveness of the newly installed temperature control device at Shasta Dam.

In addition to providing mean daily conditions depicted in the examples of Figure 11, the Sacramento River Temperature Model provides a dynamic description of daily water temperature variations along principal axes of flow. It is possible to represent the hourly temperature variation at a fixed location or to show the range of values that would be experienced over the daily cycle at all locations along the river. The latter is illustrated in Figure 12 for a Wet year condition. Here, the recurrence of nodes of minimum diurnal temperature variation, as confirmed by the Riparian Vegetation Study, are seen spaced along the river at intervals of one day's travel at the prevailing flow rate. In between nodes are reaches where diurnal variations increase, producing anti-nodes of maximum diurnal temperature variation. This pattern of variations in minima and maxima are characteristic of the river’s hydrodynamics and thermal dynamics, and temperature conditions of Keswick Reservoir release. It is notable that the number and position of the nodes and anti-nodes are a function of travel time. As flows increase and daily travel distance increases, the distance between nodes increases, and the number of nodes within the 220-mile reach of the Sacramento River between Keswick Reservoir and Verona decreases. This interaction between the river's hydrodynamics and the diurnal cycle of heating and cooling provides a unique opportunity for improving calibration and verification of the model. The model’s prediction of the node position can be adjusted by changing Manning's n so that the predicted diurnal pattern agrees with actual observations in the field.

The Sacramento River Temperature Modeling Project has produced a set of mathematical models, calibrated and verified, for use in further investigation of alternative measures to regulate and enhance chinook salmon habitat in reaches of the river system where temperature is a controlling environmental factor. These models have been preliminarily applied in a study of selected scenarios of project operations to reveal certain special characteristics of the Sacramento River and its sensitivity to upstream boundary and meteorological conditions. They represent the present state-of-the-art of mathematical modeling of hydrodynamics and temperature dynamics of riverine systems. The models provide a foundation for future investigation of measures to preserve and enhance the habitat for chinook salmon and other species dependent upon the water resources of northern California.

Additionally, the project has provided a new understanding of the role of riparian vegetation in moderating the thermal regime of the river. Field instrumentation revealed unique characteristics of diurnal temperature cycles in the river, with daily ranges from as little as a fraction of a degree to several degrees depending on location, flow rate, mean velocity, and meteorological conditions.

Based on experience in developing, calibrating, verifying, and applying the models, it is recommended:

1. that models developed in the project be applied in future studies to evaluate options for preservation and enhancement of salmon fisheries in the Sacramento and Feather Rivers, including operation of temperature control devices, regulation of carry-over storage in reservoirs, and restoration of riparian vegetation.

2. that data gathering activities be improved in support of flow and temperature evaluation, analysis, and regulation; including monitoring tributaries, project releases, irrigation diversions and return flows, and meteorological conditions.

3. that three-dimensional modeling of hydrodynamics and temperature in Whiskeytown Reservoir be undertaken to complement the present modeling capability of the Shasta-Trinity Division of the CVP.

4. that models of Shasta Reservoir and the Sacramento River be extended to represent other water quality variables of concern in maintenance of healthy salmon habitat.

5. that the Sacramento River water temperature model be extended to represent the life cycles and stages of the salmon ecosystem.

To comply with 205j program objectives, an implementation checklist was formulated to describe reasonable future Sacramento River temperature scenarios. The checklist aims to provide a framework for the Technical Advisory Committee (TAC), formed as a requirement of the 205j funding, to review, address, and prioritize recommendations and findings. Additional detailed recommendations for model enhancements and a checklist for Sacramento River Temperature Project Implementation are presented in the project report.

Work on the project was completed in early 1996. The project report was reviewed in draft form by the Technical Advisory Committee during the latter months of 1996 and submitted to the State Water Resources Control Board in February 1997. Since completion of the project, the USBR has contracted with the University of California, Davis to develop a three-dimensional hydrodynamic-temperature model of Whiskeytown Reservoir. The United States Fish and Wildlife Service has contracted with the University to extend investigation of the effects of riparian vegetation on riverine temperature and its influence on salmon habitat. Also, the California Department of Water Resources has contracted with the University to extend modeling work on the Feather River. All projects are scheduled for completion in 1998.