Laserfiche WebLink
<br /> <br /> <br />1 <br /> <br /> <br /> <br />i <br /> <br /> <br /> <br /> <br /> <br /> <br />flow measurement station on the day field measurements are taken should be recorded in <br />the data log. USGS flow measurements are available on the Internet. <br />In this particular case, measurements taken will constitute a baseline or background set of <br />measurements and a "compliance" set of measurements taken once dewatering operations <br />begin. It is likely that Lafarge and neighbors will be interested in determining whether a <br />change has occurred as soon as possible after dewatering begins. For this reason, a <br />Tolerance Limit estimating procedure will be most useful. By measuring baseline data, <br />an upper and lower limit can be set. When comparison measurements are taken, these <br />can be added to the analysis one at a time. The resulting test will determine what percent <br />"coverage" or number of measurements is included within the tolerance limits. If the <br />percent coverage drops below the confidence interval, this constitutes a significant <br />change. So, fora 90% confidence interval, the coverage must remain above 90% to show <br />no change has occurred. The drawback to this approach is that it is highly dependent on <br />the number of measurements taken to determine when compliance will be met. If only 10 <br />measurements are taken, only 1 outside the tolerance limits will constitute a change, but <br />if 100 are taken, it would take ] 0 measurements outside the limits to reflect a significant <br />change. This is the nature and limitation of statistics. On the positive side, this approach <br />allows the data to be re-analyzed on a monthly basis, eliminating the need to wait for <br />long periods of time once dewatering begins to determine if mitigation needs to occur. <br />Atwo-tailed, non-parametric tolerance limit procedure will be used. This approach <br />creates tolerance limits that correspond to the highest and lowest measured values for <br />each well during the baseline measurement period. This approach then accounts for <br />seasonality without a sepazate test or data manipulation. For this particular study, <br />measurements are limited to values between zero (empty well) and the depth of the well <br />(full well). This method is resistant to missing data. <br />Outliers will be removed from the analysis by eliminating the single highest and lowest <br />measurements from each data set. This will have the effect of creating a more <br />constrained tolerance limit (potentially overestimating the probability of a change). It is <br />assumed that if a measurement is truly an anomaly it will be radically different from the <br />rest of the measured data. If the highest and lowest measurements are close to the second <br />highest or lowest data point, these measurements may be accurate, but then the effect of <br />eliminating them will not be severe. <br />In summary, the results of this analysis will provide a conclusion about whether <br />' measurements taken before mining begins and measurements taken after mining begins <br />have changed for a range a confidence intervals (90% is recommended). These <br />conclusions will be drawn once the measurements have been adjusted for river flow. <br />' 3.2 Analysis Procedure <br />A commercially available software package, WQSTAT Plus, can be used to produce all <br />the tests recommended in the previous section. A spreadsheet input file is necessary for <br />WQSTAT to read. A formatted EXCEL spreadsheet has been included in this protocol in <br />' 10 <br />