In this section
2014 Water quality status and trends report
- Summary table
- Summary table
- Surface water
- Summary tables
Method used to generate the
2014 status and trends report
Water quality data for 31 springs monitoring stations located throughout the St. Johns River Water Management District were analyzed for status and trends. Water quality status is an indication of the condition of a water body and for a given analyte is represented by the median of annual median values from the most recent five-year period of record. Median values were chosen to represent water quality status as they are not greatly skewed by outliers, and thus serve as a robust indicator of central tendency. Water quality trends indicate whether a series of analyte values (for example, color or total phosphorus) is increasing or decreasing (changing) over time. Trends were calculated using the nonparametric Mann-Kendall test. A non-seasonal version of the test was used if there were no statistically significant differences between seasonal values (Kruskal-Wallis test, p<0.05). A seasonal version of the Mann-Kendall test was used if there were significant differences. This report does not attempt to analyze the cause or impacts of any trends.
The period of record for this 2014 trend assessment was the 15 years from Jan 1, 2000, to Dec 31, 2014. No data obtained prior to Jan 1, 2000, nor any data obtained subsequent to Dec 31, 2014, were used. At least 10 years of data were required from the 15-year period of record to calculate a trend. The 10-year period of record could be any 10 years within the 15-year period, including periods for which there was a gap in the data, although the ending year had to be 2014. Years in the period of record that had more samples than other years were adjusted such that each year was represented by the same number of samples, when possible.
The status assessment period was the most recent five years of the period of record, or Jan. 1, 2010, to Dec. 31, 2014. At least three years of data during the five-year status assessment period were required in order to complete the status assessment, and the last year had to be 2014.
Water quality data were derived from samples collected primarily by District field staff. The majority of these samples were “grab” samples, which means that they were collected by using a technique of either inverting the sample bottle in the water column or pouring sample water from a separate collection device, such as a VanDorn sampler, into the sample bottle. The samples were analyzed using U.S. Environmental Protection Agency (EPA) methods at the District’s lab or at various contract labs.
All data were analyzed using SAS Statistical Software and were filtered in a multi-step process. Since most water bodies in the District are shallow and not stratified, data from all depths were used for the assessment, including vertically integrated samples.
Collection, analysis and processing of water quality samples and data can be an error-prone process, and a number of problems can occur. On those rare occasions, the lab will associate qualifier codes with the data. In this assessment, such qualifier codes were evaluated and any suspect data filtered out.
Hardness and total nitrogen (TN) were calculated for each day of data, as these are not usually measured in the lab. TN was calculated from the sum of total Kjeldahl nitrogen (TKN) and total nitrate+nitrite (NOx), even if NOx was missing. If TKN was missing, then TN was not calculated. If total NOx was missing, then dissolved NOx was used instead, if it was available. Finally, calcium and magnesium were used to calculate a hardness value, which was then added to the data set.
All analytes were evaluated for outliers by using a range test, and any potential outliers were identified. However, since the assessment primarily uses nonparametric statistics, the magnitude of any outliers should not greatly affect the results. Thus, all values were retained in the data set, with the exception of any pH values less than 0 or greater than 14. Individual station and analyte datasets that contained non-detect values were evaluated using “survival statistics” methods.
After calculating the status values, trends were determined using the Mann-Kendall test. The test is a nonparametric test that works on a simple concept. It compares subsequent time-ordered values against the preceding values to determine the differences among them. Note that it compares all remaining values in a series against each time-ordered value (for example, x1–x2, x1–x3, x1–x4…x1–xn and then x2–x3, x2–x4…x2–xn, etc.). Obviously, many differences must be evaluated. After all differences have been calculated, they are summed and the variance is calculated. A sum of 0 indicates that there is no trend. A positive sum indicates that values are increasing over time, while a negative sum indicates that values are decreasing over time. The sum must then be evaluated for statistical significance. Although statistically significant trends are reported, this report does not attempt to analyze the causes nor impacts of any trends.
The sums are assumed to be normally distributed, so a normal probability table was used to determine critical values (p < 0.05). Finally, a Sen’s slope (with a 95 percent confidence interval) was calculated to indicate the magnitude of the trend. Analytical results were then summarized and tabulated for reporting purposes.
The status value is reported in the report tables, followed by the trend result. Trends can be increasing, decreasing, stable or insufficient data. Stations may have insufficient data for a variety of reasons. A newly established station that has not been sampled for at least 10 years will be considered to have insufficient data. An older station may have been sampled for many years, but sampling may have been discontinued for the last few years. In some cases, a station will have results for some analytes, but not others. The last three columns in the table indicate the lower 95 percent confidence interval, the estimated slope of the trend, and the upper 95 percent confidence interval. The estimated slope is calculated using the Sen slope method. The 95 percent confidence interval contains the upper and lower bounds on the slope estimate, and means that there is 95 percent confidence that the slope is in the interval.
In addition, status values and trend results were plotted on maps using ArcMap (ESRI). The status results are shown using a series of colored circles to indicate the magnitude of the values. The results are grouped using the Jenks natural breaks method in ArcMap. If a particular analyte was not sampled at a station, that result is indicated with “no data.” Trend results are shown using both colors and shapes. An up triangle indicates an increasing trend, while a down triangle indicates a decreasing trend. Squares indicate a stable result or statistically non-significant trend, crosses indicate insufficient data and circles indicate no data. The trend analysis demonstrates change during the report time period. This report does not attempt to analyze the cause of any change, nor the impact of any change. Finally, the pie charts show the percentage of sites throughout the entire District that exhibit the trends.
Posted on 00-00-2015