Water Quality Variations and Trends in Monument and Fountain Creeks

Home Browse Search

not problems in a nonparametric approach because they do not affect the rank of the data. Trends are evaluated separately for each specified season (months were used in this study), and results are combined into a single test statistic (tau). The significance level (p- value) of tau then is adjusted to account for serial correlation. In addition to testing the original data, the seasonal Kendall test can be used to identify trends in the time series of flow - adjusted concentrations. Flow adjustment would be used to eliminate the effect of correlation between streamflow and values of water - quality properties and constituent concen- trations. This adjustment would decrease the possibility of erroneously concluding that the concentrations of a particular constituent have increased or decreased as a result of some change in land or water use in the basin when the change in concentration is a result of a change in flow conditions during the period of record. For this study, flow adjustments were not made to the concentration data because the changes in streamflow and the water - quality properties and constituents have probably resulted, at least in part, from human activity in the basin (for example, increases in discharge of waste- water effluent and urban runoff). Flow adjustment should not be used where human activity has altered the probability distribution of streamflow (Hirsch and others, 1991). Flow adjustments could erroneously remove the changes in values of water- quality properties and concentrations of constituents. Flow adjustment is not appropriate when the purpose of the study is to assess the effect of trends on the suitability of water for use by humans or aquatic organisms rather than to investigate the cause of the trend (Hirsch and others, 1991). The seasonal Kendall test was applied to identify trends in the time series of each water - quality property and constituent at each station. For this study, the seasonal division was defined to be monthly because the water - quality sampling usually was done monthly. Results of the trend analyses are summarized in tables 13 through 21 in the "Hydrologic Data" section at the back of the report. The properties or constituents were included in these tables only if the data had the following characteristic §: 1. At least 3 years of record. 2. At least 20 values were larger than or equal to the detection limit at stations 07104000 Monument Creek at Pikeview, 07104905 Monument Creek at Bijou, 07105500 Fountain Creek at Colorado Springs, and 07106500 Fountain Creek at Pueblo; or at least 30 values were larger than or equal to the detection limit at stations 07103747 Monument Creek at Palmer Lake, 07103780 Monument Creek at USAFA, 07103700 Fountain Creek near Colorado Springs, 07105530 Fountain Creek below CSWWTP, and 07105905 Fountain Creek below Fountain. 3. No more than 10 percent of the values were less than the detection limit (censored values). Several significance levels are reported in the tables. The significance level of the trend test is the maximum probability of erroneously identifying a trend that does not actually exist. The significance level unadjusted for serial correlation is reported for all trends. If the period of record is 10 years or longer, the significance level adjusted for serial correlation also is reported. Generally, the adjusted significance level will be larger; 14