8195 - Laserfiche WebLink

Home Browse Search

1168 :I 111 '1,11 oio;ic.111el-aliun High pulse: Low pulses: 1600 r count=16 count-8 average dunlion6 davs average durvi-6.13 da.wz 1400 1200 1000 i c° 800 E C7 600 400 [ 200 r ?I!.\ A II.IJ'I ,II -IY ???? lll iiki'ik..`.Id v ---__ ---- --------- ".r pi.Iu.br m J F M A M J J A S O N D Month Figure 2. The computations for IHA group 4 parame- ters are of frequency and duration of high and low pulses. High pulses are identified as those periods dur- ing which water levels rise above the 75th percentile of all pre-impact daily flows, which is equivalentto 258 tins for the Roanoke River. Low pulses represent drops in water levels below the 25th percentile, equivalent to 97 cros. tra-annual cycles of environmental variation and can pro- vide a measure of the rate and frequency of intra-annual environmental change. Assessing Hydrologic Alteration In assessing the impact of a perturbation on the hydro- logic regime, we want to determine whether the state of the perturbed system differs significantly from what it would have been in the absence of the perturbation. In particular, we want to test whether the central tendency or degree of inter-annual variation of an attribute of in- terest has been altered by the perturbation (Stewart- Oaten et al. 1986). The assessment of impacts to natural systems often poses difficult statistical problems, how- ever, because the perturbation of interest cannot be replicated or randomly assigned to experimental units (Hurlburt 1984; Stewart-Oaten et al. 1986; Carpenter 1989; Carpenter et al. 1989). The lack of replication does not hinder estimation of the magnitude of an effect but rather limits inferences regarding its causes. This is- sue has received considerable attention recently, and more sophisticated experimental designs that incorpo- rate replication over time and sampling at "control" and "impact" sites have been suggested (Stewart-Oaten et al. 1986, 1992). We are fully cognizant of the replication issue, and we have based the IHA method on the simple design of comparing hydrologic attributes of a single site before and after a putative perturbation. This method allows es- Riciller el rd_ 850 750 Firs. - I Fin. fall 4-d rive 16aA m'/» 650 FaN rasrHdB m /csds, 1.'?A m'ISl ' 550 E / 450 Rhe. -u .3 W..", E \ v 350 250 12J3A aih, loss nle-529 n isles, R3ss -1431 miJds,' 150 pol. /» 50 161b V. i» Days Figure 3. The computations for IHA group 5 parame- ters are of rate and frequency of change in conditions, illustrated by means of a h}pothetical hydrograph. Two hydrograph rises and one hydrogzgpb fall are identified in the time period shown, along with their corresponding rates of rise and fall. A single rise or fall may last for multiple days and is terminated once the hydrograph begins to fall. Also, rise and fall rates are computed for each day within each year of record analyzed. timation of the magnitude of impacts but does not en- able strong inferences regarding the cause. We take this simpler approach for two reasons. First, in many loca- tions, no control site may be available. Second, causal inference, although desirable, may not always be a nec- essary prerequisite for prescribing management or res- toration actions to mitigate for observed effects. But the IRA method is robust and can be easily adapted to more- sophisticated experimental designs. To ensure consistency in the application of the IHA method, users should clearly identify the presumed cause of the impact(s) being evaluated-for example, the impact of an upstream reservoir or irrigation diver- sion on streamflow, or the effects of ground water pumping on wetland pond levels. The time period for which hydrologic records exist prior to the presumed perturbation can be defined as pre-impact, and the pe- riod of record since initiation of the presumed perturba- tion can be defined as post-impact. Once pre- and post- impact time periods have been defined, the hydrologic regimes from the two periods can be characterized and compared. A standard statistical comparison of the 32 IHA param- eters between two data series includes tests of the null hypothesis that the central tendency or dispersion of each has not changed. But this null hypothesis is gener- ally far less interesting in impact assessments than ques- Conservation Biology VOILIMe 10. No. 4. Allgllst 1996