Laserfiche WebLink
<br />Oaten et al. (1986). In the absence of an appropriate control site, process-based hydrologic <br /> <br />models that simulate climatic and runoff processes or other climate analysis techniques can be <br /> <br />used to create an "unimpacted" data set (see Maheshwari et al. 1995 for a similar application). <br /> <br />Case Study Applications <br /> <br />- ~....+->~ <br />The Roanoke River in North Carolina and the Edwards Aquifer system in Texas have 'been <br />'.-~.; <br /> <br />selected, to illustrate the application~,of:the-IHkmethod for assessing hydrologic' 'aJterarlo~n .'. <br />The Roanoke River' has been impacted by dam operations and the Edwards Aquifer '~1~ <br /> <br />.,-- , " ,"d:lo! <br />experienced progressively greater volumes of ground water pumping over the past 50 yeai-s. <br />In choosing appropriate estimators of the central tendency (e.g.; mean, medi~r~" 'o' .' <br />. .'..,:~~~~:~;,""-"- <br /> <br />dispersion (e.g., variance, coefficient of variation) of the hydrOlogic 'attribti~;:' Qgffil'''~~1~~::'' , <br /> <br />consideration needs to be given to the efficiency of the estimator, as well as the efficieniYf..ti <br />:- '. -. '" '. -.; "i~~J' <br />assumptions of statistical tests used to evaluate the difference between time periods: The' mean <br /> <br />is the most efficient estimator of central tendency when the underlying distribution is NO~~ <br /> <br />. - .-.." i:t..-..:<i1~1 <br />and various Hike tests~basedon'the--mean:are-applicable even when assumptions of the <br /> <br />standard t test (e.g., Normal distribution, equal variances) are violated (Stewart~Oa~~etM::';n, <br /> <br />. -. >.. ."~:""'1 ""~g;ri>' '_ . <br />1992). In the two case studies presented below, we use the mean as an estimate ofeentl:lill'c. ,. <br /> <br />tendency, and the coefficient of variation as an' eStimate ofl:lispersion. <br /> <br />'!:/. . . <br /> <br /> <br />. ',' ,... _' .:.- ';'o,}! ,.""';;~~~. <br />For each of the 32 hydrologic attribiItes, the differences between the pre- and post-" c' <br /> <br />~ "'-"~'~"'~~~ <br />impac! time periods in both the mean and coeffiCient of variation are presented, expressed as ,oo <br /> <br />both a magnitude of difference and a deviation percentage (Tables 3;4): ' The cOmpafisoH~ <br /> <br />-".- <br /> <br />.:. ~ .i ~",n'-f <br />-. '. ""-~'-~." <br /> <br />16 <br />