外文翻譯--主成分分析確定河邊的地下水開采管理指標(biāo)的時間序列變量（英文）

1、Principal component analysis of time series for identifying indicator variables for riverine groundwater extraction managementRebecca M. Page a,?, Gunnar Lischeid b, Jannis Epting a, Peter Huggenberger aa Applied and Env

2、ironmental Geology, Institute of Geology and Paleontology, Department of Environmental Sciences, University of Basel, Bernoullistrasse 32, CH-4056Basel, Switzerland b Institute of Landscape Hydrology, Leibniz Centre for

3、Agricultural Landscape Research, Eberswalder Strasse 84, D-15374 Müncheberg, Germanya r t i c l e i n f oArticle history:Received 28 March 2011Received in revised form 11 December 2011Accepted 13 February 2012Availa

4、ble online 21 February 2012This manuscript was handled by PhilippeBaveye, Editor-in-Chief, with the assistanceof K.P. Sudheer, Associate EditorKeywords:Groundwater qualityDrinking waterRiver–groundwater interactionPrinci

5、pal component analysiss u m m a r yAlthough alluvial aquifers connected to rivers can be a rich source of drinking water, they are susceptibleto contamination by infiltrating river water. The processes governing river–gr

6、oundwater interaction arevariable in time and space. Natural filtration mechanisms are often not sufficient during high dischargeevents in the river. To capture the dynamics of river–groundwater interaction, indicator pa

7、rameters thatcan serve as proxies for river water infiltration need to be derived. Principal component analysis of con-tinuously measured time series was used to identify indicator wells and derive indicator parameters f

8、or astudy area in NW Switzerland. The results showed different sources of variation in the parameters,including river stage fluctuations. The multivariate approach highlighted differences between observa-tion wells based

9、 on the response of the measured parameters to effects of damping and delay of the inputsignals. Three observation wells were shown to be representative of river–groundwater interactiondynamics in the study area. Of the

10、three parameters analysed, groundwater head and electrical conduc-tivity are recommended as a combined proxy for river water infiltration in the study area. In contrast,temperature proved not to be a reliable indicator.?

11、 2012 Elsevier B.V. All rights reserved.1. IntroductionAlluvial deposits are often rich sources of drinking water owing to the high permeabilities in the proximity of rivers (Su et al., 2007). In the last few centuries,

12、many groundwater extraction wells were constructed in the proximity of rivers and water suppli- ers rely on natural groundwater recharge from the river to main- tain the availability of extractable water (Sheets et al.,

13、2002). River–groundwater interaction is heterogeneous both in time and space and depends on multiple factors, including river discharge, the morphologic and sedimentologic character of the river bed and the heterogeneity

14、 of the hydraulic properties of the boundary layer (Sophocleous, 2002). For example, the conductance of the riv- er bed is a major factor in determining river–groundwater ex- change and is a function of river discharge.

15、The heterogeneity of hydrogeological zonation is scale dependent and non-stationary. The focus of river–groundwater interaction has frequently been on quantity, however, the exchange of water between the river and the aq

16、uifer is one of the most important factors for groundwa- ter quality (Wroblicky et al., 1998).Microbial contamination of riverine origin is one of the major concerns for drinking water suppliers extracting riverine groun

17、d- water (Sheets et al., 2002; Regli et al., 2003). Under average hydro- logical conditions, the natural filtration capacity of the subsurface matrix is sufficient to remove many physical and biological compo- nents from

18、 the infiltrating river water (McDowell-Boyer et al., 1986; Taylor et al., 2004; Gupta et al., 2009). However, high dis- charge events often result in a degradation of water quality (Wil- kinson et al., 2006). McKergow a

19、nd Davies-Colley (2010) have shown that during storm events, the concentration of Escherichia coli in river water can be up to 30 times higher than during base flow conditions. As a consequence, the concentration of micr

20、oor- ganisms in the adjacent groundwater can also increase by several orders of magnitude (Regli et al., 2003). Taking this into account, and that the first barrier to colloids and contaminants entering the aquifer – the

21、 clogging of the river bed – is weakened during high discharge events (Sophocleous, 2002), the potential for riv- er-borne microbial contamination of extraction wells near the river can be significantly elevated (Regli e

22、t al., 2003). Continuous monitoring of the microbial load in groundwater is not suitable for applications in the field, as the cost and time investments are considerable. There have been recent advance- ments in real-tim

23、e microbial monitoring methods, these are how- ever not yet readily available for many applications (Martinez et al., 2010). Thus, the question arises whether proxies can be used0022-1694/$ - see front matter ? 2012 Else

24、vier B.V. All rights reserved.doi:10.1016/j.jhydrol.2012.02.025? Corresponding author. Tel.: +41 61 2673447; fax: +41 61 2672998.E-mail addresses: rebecca.page@unibas.ch (R.M. Page), lischeid@zalf.de(G. Lischeid), jannis

25、.epting@unibas.ch (J. Epting), peter.huggenberger@unibas.ch(P. Huggenberger).Journal of Hydrology 432–433 (2012) 137–144Contents lists available at SciVerse ScienceDirectJournal of Hydrologyjournal homepage: www.elsevier

26、.com/locate/jhydrolis no obvious break between reliable and interpretable compo- nents and components due to random noise (Jackson, 1993). The third method, also called communality rule, considers the amount of variance

27、described by each component and the cumulative sum. The following PCA results considered the Kaiser-Guttman criterion, the scree test and the communality rule.3. Results and discussionThe time series used for analysis we

28、re recorded in January 2009. Selected time series are shown in Fig. 2. During the measurement period, two high discharge events occurred (max. discharge: 80 m3 s?1, 19/20 and 24/25 January 2009). Decreases in T and EC we

29、re observed during the high river discharge events, as were in- creases in GH. Groundwater was continually extracted throughout the two high discharge events. Especially the GH data showed a strong relation to groundwate

30、r extraction. This effect was observa- ble in all observation wells north of the extraction well field (downstream), but not in observation well J (or G, both upstream of well field). Observation wells close to an extrac

31、tion well, e.g.F3 showed a stronger response than wells further away, e.g. J. Other potential sources of variation included air and river water temperature fluctuations. The PCA was carried out separately for the water t

32、able (GH and river stage, 16 time series), T (13 time series) and EC (7 time series). The resulting patterns differed considerably and are discussed sep- arately. Fig. 3 shows the representation of the GH, T and EC load-

33、 ings for PC1 and PC2. The loadings were calculated as the correlation between the PC scores (projections of the input time series into eigenvalue space) and the input time series. The results of the PCA are displayed in

34、 relation to a unit circle and in depen- dency of PC1 and PC2 (Fig. 3). The unit circle gives the maximally possible loadings for time series where the variance is completely explained by the first two components. Each p

35、oint represented the loadings of the data set from each observation well for PC1 and PC2. The smaller the distance from the unit circle, the more vari- ance was explained by the first two PCs, and thus the less signifi-

36、cant other sources of variation were in determining the observed fluctuations. For all three parameters, only the eigenvalues of the first two PCs exceeded one. Both the scree test and the communality ruleFig. 1. (a) Loc

37、ation of study area, observation and extraction wells. The zones with different hydraulic conductivities are also shown, as are the groundwater head contoursand average groundwater flow direction during an average hydrol