Laserfiche WebLink
i <br /> WISCONSIN ADMINISTRATIVE CODE 134 <br /> Unofficial Text(See Printed Volume}. Current through date and Register shown on Title Page. ' <br /> Continue for the remaining rows.Stam each row and enter result at the end of the row. Add the sum of each row down <br /> i <br /> to obtain the Mann-Kendall Statistic(S). See Table A as an example. ' <br /> Table A <br /> Mann-Kendall Statistic <br /> Sampling Sampling Sampling Sampling Sampling r <br /> I <br /> Event I Event 2 Event 3 Event 4 Event 5 <br /> Contaminant concentra- Sum Rows <br /> tion-; 1 100 50 85 75 50 <br /> Compare to Evart I -> -1 -1 -1 -1 -4 <br /> Compares to Event 2 —> +1 +1 0 +2 <br /> Compare to Event 3 -1 -I -2 <br /> Compare to Event 4 -I ' <br /> Mann Kendall Statistic(Total) -5 <br /> b. Manual Mann-Kendall Statistic Look up Table. Table B gives the maximum S statistic(S }to accept a declining , <br /> trend alternative at an a level of significance. If the eanputod S is greater than S,.(or S is a smaller negative number <br /> than Ste),then there is either a no-trend or an increasing trend in the data. <br /> Table B ' <br /> Mann-Kendall Statistic Look Up Table <br /> S. <br /> N Range of S a =0.2• <br /> 4 -6to+6 -4 <br /> 5 - 10 to+ 10 -5 <br /> 6 - 15 to+15 -6 <br /> 7 -21 to+21 -7 <br /> 8 -28 to+28 -8 <br /> 9 -36 to+36 -10 <br /> 10 -45 to+45 -11 ' <br /> The prvbsWity that the computed Mann-Kcadall sta istic S 5 S, is at most a. <br /> 5. Test for a declining trend. Evaluate data treads for each contanumant identified in the plume. Evaluate the null hypothe- <br /> sis of no trend against the altawtive of a decreasing trend. The null hypothesis can be rejected in favor of a decreasing , <br /> tread if both of the following conditions are met: <br /> a. S is a large negative number(see Table B for magnitude of S) ' <br /> b. The probability value,given n(number of data)and the absolute value of S.is LESS than the a priori significance <br /> level,oc of the test. An a:5 0.2 is acceptable. <br /> 6. Test for an increasing trend. An increasing trend alternative(i.e.,an advancing plume)is shown if both of the following <br /> conditions are mct: <br /> a. S is positive. <br /> b. S?I S.I at a given a level of significance(see Table 8). If the computed S is equal to or greater than the absolute ' <br /> value of S,..,then it can be concluded the plume is advancing at an a level of significance. An a S 0.2 is acceptable <br /> for this test. ' <br /> 7. Test for Plume Stability.If the Mann-Kendall Test indicates no-trend is present,perforin the coefficient of variation <br /> test. As a non-parametric test,the Mann-Kendall Test docs not take into account the magnitude of scatter in the data. A <br /> data set with a great deal of scatter may return a Mann-Kendall test indicating there is no trend,when,in fact,no conclusion <br /> can be drawn regarding trend because of data variability. In this case,additional data collection may be nc=sary to deter- <br /> mine that the plume is stable,declining or advancing. As a simple test,the coefficient of variation can assess the scatter <br /> in the data: <br /> Reeisier.January.2401.No.541 <br />' I <br /> i <br />