|
WRIR 02-4030: Streamflow and Water-Quality Data for Selected Watersheds in the Lake Tahoe Basin, California and Nevada, through September 1998 |
|
|
|
|
Streamflow was measured and continuous streamflow gaging stations were operated according to USGS methods (Buchanan and Somers, 1969; Kennedy, 1983). Periods of missing data for streamflow gaging stations were estimated by comparing records with other gaging stations on the same or nearby watersheds and with temperature and precipitation records.
Streamflow rates at the time of sampling were determined either by making a current-meter measurement or by using a rating-curve estimate; the latter was used if the stage-streamflow relation was available and stable for the particular primary and secondary stations. When samples were collected during periods of changing stage, an average stage was calculated using the gage heights observed or recorded at the beginning and end of the sampling period. The average streamflow then was obtained by looking up the corresponding value on the rating curve.
Annual unit values of runoff were derived from the average annual runoff (acre-feet) divided by the station drainage area (square miles) and reported as acre-feet per square mile. Flood-frequency values were calculated from annual peak streamflow for periods of record for each gaging station using methods described in Thomas and others (1997) and U.S. Interagency Advisory Committee on Water Data (1982).
For this report, drainage areas for sampling stations (table 1) and watershed areas (table 3) were obtained from either the Tahoe Environmental Geographical Information System (TEGIS) project as reported in Cartier and others (1995) or LaRue Smith (U.S. Geological Survey, written commun., 2001). Because of improved geographic technology, these drainage areas supersede previously published values in Water Resources Data Reports (U.S. Geological Survey, 1989-99, published annually). Drainage-basin perimeters and station elevations are in Cartier and others (1995); and channel elevation ranges, maximum watershed elevations, and channel lengths are in Jorgensen and others (1978). Distance from mouth and channel length above station (table 1) were determined from USGS topographic maps.
Water-quality measurements of water temperature, specific conductance, pH, dissolved oxygen, and dissolved-oxygen saturation were made in the field at the time of sampling. All measurements were made in accordance with USGS methods (U.S. Geological Survey, 1998). Air temperatures were measured at the same time as water-quality parameters.
Samples of nutrient and suspended sediment were collected using USGS methods (Edwards and Glysson, 1988; U.S. Geological Survey, 1998). Discharge-weighted samples were integrated by depth and used the equal-width increment (EWI) method (Edwards and Glysson, 1988) for most of the study. Nutrient samples, collected first, were composited and mixed in a churn splitter. The nutrient samples were then preserved and shipped overnight to TRG laboratories in Tahoe City and Davis, Calif., for analysis using LTIMP procedures described in Hunter and others (1993). Suspended-sediment samples were collected separately, after the nutrient samples, and shipped to the USGS Sediment Laboratory in Salinas, Calif., for analysis using USGS standard guidelines (Guy, 1969).
Two load calculation computer programs, ESTIMATOR (Cohn and others, 1989) and FLUX (Walker, 1996), were investigated for estimating nutrient and suspended-sediment loads for this study. The use of a third computer program, LOADEST2 (Crawford, 1986), was contemplated, but problems encountered in running this program with the large LTIMP data set precluded its use.
Loads for the 12 miscellaneous stations were not calculated because the samples were collected mainly during high flow (spring snowmelt and storm) events. Daily streamflow measurements also do not exist for these ungaged stations. For load calculations, total nitrogen was obtained by combining total ammonia and organic nitrogen and dissolved nitrite plus nitrate nitrogen. Soluble reactive phosphorus is equivalent dissolved orthophosphate.
ESTIMATOR, a log-linear, multiple regression model, relates constituent concentrations to as many as 16 environmental variables or parameters. The program was developed in 1988 to assist USGS personnel in estimating stream-nutrient loads. ESTIMATOR implements the Minimum Variance Unbiased Estimator (MVUE) for the use of estimating fluvial transport of nutrients and sediment (Cohn and others, 1989) and the Adjusted Maximum Likelihood Estimator (AMLE) for the use of data sets containing censored (less than) values (Cohn, 1988 and Cohn and others, 1992). The LTIMP data set contains censored values, particularly those associated with dissolved ammonia.
ESTIMATOR represents concentrations as a function of three factors; flow, time, and a seasonal factor. Outputs include daily and monthly load rates and annual loads for each calendar and water year. To determine a constituent's total load for a given month, the estimated daily mean load rate was multiplied by the number of days in that month. The 95-percent confidence interval was calculated in the program by multiplying the standard error of prediction (SE PRED) by 1.96, which is the value for a 95 percent confidence interval (G. Baier and others, USGS, Reston, Va., written commun., 1993). Load estimates with standard errors (SE) less than 30 percent were accepted, between 30 and 50 percent were marked as questionable and were reviewed and included in the data set if found acceptable after data verification, and greater than 50 percent were not accepted. Monthly load-estimate results are reported in kilograms per month from the ESTIMATOR program. During extreme streamflow events, the program may overestimate load values (Doug Glysson, U.S. Geological Survey, oral commun., 2000). Estimated nutrient daily and monthly loads used in this report are listed in appendix 2, with estimated SE, estimated SE PRED, and 95-percent confidence intervals.
During lower streamflow events FLUX estimates of nutrient loads for small watersheds exhibited some unusual fluctuations that were not present in the ESTIMATOR results. Thus, estimated values for nutrient loads in this report were taken from ESTIMATOR.
FLUX, an interactive computer program used to estimate the loadings of nutrients or other water-quality constituents such as suspended sediment, is described in Walker (1996). Data requirements for FLUX include: constituent concentrations, collected on a weekly to monthly frequency for at least a year; date collected; corresponding flow measurements (instantaneous or daily mean values); and a complete flow record (daily mean streamflow) for the period of interest.
Six estimation algorithms are available within FLUX. For calculations in this report the following were used: (1) flow-weighted concentrations (ratio estimate), (2) modified ratio estimate, (3) first order regression, (4) second order regression, and (5) regression applied to individual daily streamflow. FLUX maps the flow versus concentration relation developed from the sample record onto the entire flow record to calculate total mass, streamflow, and associated error statistics. An option included to stratify the data into groups based upon flow was used to improve the fit of the individual models.
FLUX was used for monthly suspended-sediment load estimations, which are listed in appendix 2. Estimates of suspended sediment from FLUX and ESTIMATOR were similar except for overestimates by the latter program for months that included extreme peak events, like the January 1997 flood. Because those ESTIMATOR flood-load results were well above expected reasonable limits for three stations (Blackwood, Ward, and Logan House Creeks), estimated values for suspended-sediment loads in this report were taken from the FLUX program (app. 2).
Calculations for estimating trends in nutrient and suspended-sediment concentrations were made using the Seasonal Kendall test (Helsel and Hirsch, 1992). This is a non-parametric test for a monotonic linear trend that is resistant to outliers and is not dependent on the normality of the data. This test reduces seasonal effects on concentrations when testing trends by comparing the data by season. A Locally Weighted Scatter Plot Smoothing (LOWESS; Helsel and Hirsch, 1992) was used to flow adjust the data by removing the effect of streamflow variations on the concentrations. The Seasonal Kendall test was performed on the flow-adjusted data. Trends detected by the test were considered significant if they had a p-value less than or equal to 0.05; slightly significant with a p-value greater than 0.05 but less than 0.10. Trends were not considered significant with a p-value greater than or equal to 0.10.
|
|
|
Table of Contents
Report Home Page
