WSP11764 - Laserfiche WebLink

Home Browse Search

r trends are identical to two decimal places. Separate ..... estimates of b, for each season would result in a o betterfitto the data and inmore precise estimates F~~ of long-term trends by equations 3.9 and 31, C> where the difference between b,ilnd Cm is evalu- ated. In several cases in table 7, Cm is larger than b, (see equations 30and 31), causing a reversal in the sign of the estimated long-term trends. In these cases, only the sign of b3 in equation 22 is listed. Tables 9, 10, and 11 contain the estimated long- term trends for mass fraction where significant values of b3 are indicated in equations 17, 19, and 21. The three empirical equations estimated almost identical long-term trends in mass fractions. Table 12 contains all trends that are indicated to be significant by any of the three mass fraction regressions. Where no trend is indicated for an entry in tables 4 through 12, an actual trend may exist, but the data do not justify rejection of the null hypothesis of zero long-term trend. Only a few (4 to 15) months of data are available for the Cameo station, which waS included in the trend analysis because of its importance as a major gaging station on the Colorado River. Generally, for each row in tables 4 through 8, the sum of the significant trends in concentration for the six individual ions (columns 3 through 8) is approximately equal to the signifi- cant trend estimated for the sum of the six ions (column 9). For each row of mass fractions in tables 9 through 12, the sum of the significant trends of the individual ions is approximately zero. Long term declines in the TDS (sum of six ions), corrected for the variation in streamflow by multiple regression, are evident at 8 of the 12 gaging stations (see table 8). The Dolores River near Cisco and the Yampa River near Maybell are the only stations indicated to be experiencing an increase in salinity. At eight gaging stations the regression results indicate a negative trend in sulfate ion concen- tration and mass fraction (see tables 8 and 12). Declines in the concentrations of magnesium and sodium ions are evident at seven stations (not including Cameo). For the base flow season of December to March, in the Colorado River at Lees Ferry and near the Grand Canyon, the concen- trations of the sulfate, magnesium, and sodium ions have declined in stoichiometric proportions, causing a decline in TDS of approximately 2mg/L per year. Figures 6 through 9 and 10 through 13 illustrate the declines in magnesium and sulfate concentrations during December, January, Feb- ruary, and March in the Colorado River near the Grand Canyon. The v-variable is transformed to accommodate a two-dimensional figure. The v- variable in figures 6 through 8 and 10 through 12 is the ion concentration with a linear or power correction for the natural variability of streamflow. The regression lines are the results ofthe empirical multiple linear regressions (equations 15, 17, and 19). The v-variable in figures 9 and 13 is the estimated value of the product of the solid-liquid mass transfer coefficient (k) and the total solid- liquid contact area (A) upstream of the gaging station (see equations 8 and 9). The regression lines are the result of nonlinear regression with the residence time equation (equation 22). IMPACT OF FLOW REGULATION When an impoundment is created, upstream waters are mixed in the reservoir and quality fluctuations diminish. Water released downstream during periods of natural low flow decreases the normally high salt concentrations. In addition, chemical changes caused by the dissolution and precipitation of mineral salts within the reservoir may change the character of the release water [8]. At locations such as Lees Ferry, where flow is almost completely regulated, concentration is no longer related to flow. At other sites, where flow has become partially regulated, the concentration- flow relationship may be weakened or changed altogether. The impact of regulation must then be evaluated independent of streamflow. Therefore, two statistical analyses were used to test for changes in water quality as a function of time alone. In both analyses, the monthly data were divided into two periods: (1) before 1963, and (2) after 1965. These represent the times before and after the filling of the Colorado River Storage Project reservoirs. Solute loads, mass fractions, and flow- weighted concentrations were computed for the three seasons: December-March, May-June, and August-November. Student's ttest was then used to determine significant differences between mean seasonal values. Trends following regulation were tested by linear regression, with time as the sole independent variable. All stations were treated alike, even if the flow there did not become regulated during the 1963-65 period. The results could then be compared between stations and regulation impact could be separated from other basinwide influences. Complete results of these tests are displayed in tables 10 and 11 of appendixes B through P. 17