Surface-water quality and flow Modeling Interest Group
USGS, Water Resources Division
333 W. Nye Lane
Carson City, NV 89706
Internet: rltaylor@usgs.gov
Phone: (702) 887-7677
FAX: (702) 887-7629
The U.S. Geological Survey (USGS), to support Department of Interior implementation of the Truckee-Carson-Pyramid Lake Water Rights Settlement Act of 1990 (P.L. 101-618), developed two physically-based water-quality models for simulating stream temperature and dissolved solids, for the Truckee River. The foundation of these water-quality models is the physically-based USGS daily flow-routing model of the Truckee River using Hydrological Simulation Program-FORTRAN (HSPF). The flow-routing model routes streamflow, which transports stream temperature (heat), and dissolved solids along 114 miles of the mainstem Truckee River from just downstream of Lake Tahoe, California to Marble Bluff Dam, just upstream from Pyramid Lake, Nevada.
Figure 1. Map of the Truckee River Basin.
(A larger map [56k GIF] is
available.)
Data to calibrate, validate, and evaluate these models included daily streamflow data; hourly stream-temperature and meteorological data; and dissolved-solids and specific-conductance data covering the period from June 1, 1993 to September 30, 1995. Development of these water-quality models is a step toward development of a "modular framework modeling system" that provides a mechanism for integrating many hydrologic and operational analysis models into a single tool. This article focuses on the results of the stream-temperature model.
As with many streams in the arid West, there have been long-standing and intense conflicts among various economic, political, ecological, and institutional interests over water in the Truckee River Basin. Truckee River water is used for: power generation upstream from Reno; municipal/industrial supply for the Lake Tahoe Basin and the Reno-Sparks vicinity; irrigation in both the Truckee and Carson River Basins; maintaining Pyramid Lake levels; and providing flows of sufficient quantity and quality to enhance populations of the endangered cui-ui lakesucker and the threatened Lahontan cutthroat trout in Pyramid Lake. The diversity in interests results in a wide range of alternatives for planning, allocating, and managing the water resources of the Truckee River.
The interbasin transfer of water for irrigation at Derby Dam causes streamflow to become very low at the USGS gaging station Truckee River Near Nixon (USGS Station ID 10351700, fig. 1) in the lower river. Streamflow at this station is less than 25 cfs 10 percent of the time each year on average, which creates the potential for elevated stream temperatures. It was necessary, because of the shallow and wide nature of the Truckee River during these low flows, to modify HSPF during the course of this project to include a streambed heat conductance algorithm. Without the streambed conductance algorithm, simulated fluctuations between daily minimum and maximum temperatures were too large.
Results of the stream-temperature model were evaluated at three USGS gaging stations along the Truckee River for June 1, 1993 to May 31, 1994 (calibration period) and June 1, 1994 to September 30, 1994 (validation period). The validation period included summer streamflows lower than the calibration period summer flows. In fact, the maximum streamflow for July-September 1994 at the Nixon station was less than the minimum streamflow for the same period in 1993 (38 and 45 cfs respectively). The three evaluation stations were Truckee River at Farad (10346000), Truckee River at Vista (10350000), and Truckee River at Marble Bluff Dam (10351750) (fig. 1).
Results are shown graphically for daily maximum stream temperatures at Vista (where generally the best simulations were found) and at Marble Bluff (where generally the worst simulations were found) in figures 2 and 3.
Figure 2a. Graph showing scatterplot of observed and simulated daily
maximum stream temperatures for the calibration period (June 1, 1993 through
May 31, 1994) at Truckee River at Vista, Nevada.
Figure 2b. Graph showing scatterplot of observed and simulated daily
maximum stream temperatures for the validation period (June 1, 1994 through
September 30, 1994) at Truckee River at Vista, Nevada.
Figure 3a. Graph showing scatterplot of observed and simulated daily
maximum stream temperatures for the calibration period (June 1, 1993 through
May 31, 1994) at Truckee River at Marble Bluff Dam, Nevada.
Figure 3b. Graph showing scatterplot of observed and simulated daily
maximum stream temperatures for the validation period (June 1, 1994 through
September 30, 1994) at Truckee River at Marble Bluff Dam, Nevada.
Statistical comparisons at all three stations of simulated and observed values generally showed that during the calibration period mean absolute errors were less than about 1oC and there was generally a small negative bias less than about 0.5oC. For the validation period, the mean absolute errors were still generally less than about 1oC for daily maximum and minimum values. An exception to this was the error in daily maximum stream-temperature simulations at Marble Bluff which increased from 0.8oC during the calibration period to 1.8oC during the validation period. There were also increases in the bias and variability of the errors for the validation period which are shown in figures 2b and 3b.
Statistical comparisons were also done by simulated streamflow class and showed that there was a tendency for daily maximum/minimum and hourly model simulation errors to decrease with increasing streamflow. The best simulations were found when simulated streamflow was > 500 cfs. When simulated streamflows were > 500 cfs, daily and hourly mean absolute errors were 0.4oC to 0.8oC for the calibration and validation periods at all three gaging stations (except daily maximum error at Marble Bluff which was 1.5oC with only 5 values to compare in this flow class).
The addition of a streambed algorithm to the HSPF stream-temperature module makes HSPF more useful in rivers like the Truckee River where flows can be low and stream geometry can be wide and shallow. There is still an increase in error during low flow with an increase in bias and variability in the errors. In general, though, stream-temperature simulations are within about 1.5oC of observed stream temperatures and errors decrease with increasing simulated streamflow.
Back to the SMIG Features Page
Home | Mailing List | Features | Conferences | Classes | Reading | Model Archives | Feedback