1. Introduction
Groundwater is an important part of hydrological processes and plays a vital role in the provision of water for agriculture, industrial use, municipal use, and domestic uses in the arid and semiarid regions (Zektser and Loaiciga 1993; Li and Barry 2000; Cardenas and Wilson 2006). Previous studies on groundwater mainly focused on groundwater flow regime, infiltration condition, heterogeneity of aquifer, and groundwater pumping (Bunn et al. 2010; Cuthbert et al. 2010; Carrera-Hernández et al. 2012; Odling et al. 2015). Because of ongoing warming of the global climate, the relations between precipitation and groundwater levels have become complex (Lee et al. 1999; Jan et al. 2007; Treidel et al. 2012). Many hydrogeologists pay much attention to the interrelationship of groundwater and climate changes (Taylor et al. 2013; Klove et al. 2014). Chen et al. (2004) found that precipitation displays a strong correlation with annual groundwater levels in the upper carbonate aquifer in southern Manitoba in Canada. Hao et al. (2006) revealed that the continuous decline of water level in the karst aquifer of Liulin Springs might be largely attributed to the decline in regional precipitation over the past two decades. Lee et al. (2014) concluded that more abundant rainfall (from 2000 to 2010) in the wet season does not contribute significantly to groundwater recharge, whereas less rainfall that occurs in the dry season can cause a decrease in the groundwater level in South Korea.
Recently, scientists extended the research of groundwater and climate change to effects of teleconnections and large-scale climate phenomena on groundwater. Holman et al. (2011) demonstrated that groundwater levels from three boreholes in different aquifers across the United Kingdom are correlated with North Atlantic ocean–atmosphere teleconnection patterns using wavelet coherence. Gurdak et al. (2007) found that large-scale climate patterns [El Niño–Southern Oscillation (ENSO) and Pacific decadal oscillation (PDO)] dominated groundwater resources of the High Plains aquifer in the United States and controlled groundwater deep infiltration and chemical mobilization there. To better understand groundwater recharge in Canada, Tremblay et al. (2011) used correlation, wavelet analyses, and wavelet transform coherence (WTC) to investigate the relations between climatic indices and groundwater level in three Canadian regions. Results showed that groundwater level variability in the Prince Edward Island region was mostly influenced by the North Atlantic Oscillation (NAO) and the Arctic Oscillation (AO). The groundwater levels in southern Manitoba near the Winnipeg region were affected by the Pacific–North American (PNA) pattern, and the groundwater levels in the Vancouver Island region were impacted by NAO, AO, and ENSO. Furthermore, Perez-Valdivia et al. (2012) used Spearman rank correlation and spectral analyses to assess the effects of climate teleconnection patterns on the groundwater levels in the Canadian Prairies and found that the groundwater level showed 2–10- and 18–22-yr oscillations in response to the influences of ENSO and PDO, respectively. The groundwater levels tended to decline during the warm ENSO or positive PDO periods. Kuss and Gurdak (2014) used singular spectrum analysis (SSA), WTC analysis, and lag correlation to quantify the effects of teleconnection patterns on principal aquifers in the United States and indicated that groundwater levels were partially controlled by interannual to multidecadal climate variability and were not solely a function of temporal patterns in pumping. ENSO and PDO contributed to a larger portion of variability in groundwater levels than NAO and Atlantic multidecadal oscillation (AMO) across the United States, particularly in the western and central regions. Although a few studies have linked large-scale climate patterns to groundwater, relations between large-scale climate pattern variability and subsurface hydrological processes are poorly understood, especially for the karst terrain areas (Green et al. 2011; Treidel et al. 2012). Karst aquifers are extremely heterogeneous and vulnerable to environmental change. It is essential to investigate the impacts of climate change on karst aquifers.
In general, there are two types of proxy for groundwater variability: the groundwater level (or hydraulic head) or groundwater discharge (which usually depends on the hydraulic conductivity and the hydraulic gradient). Many previous studies employed groundwater level as an indicator to reflect aquifer status. However, aquifers are heterogeneous, and different observation wells may have different groundwater levels. Thus, it is often difficult, and sometimes impossible, to find a representative groundwater level as the proxy of an aquifer. In other words, the groundwater level measured at a particular location is a localized phenomenon and does not represent a regional behavior of an aquifer. On the other hand, a spring is a natural discharge point of an aquifer, and its discharge variation reflects the combined information of permeability and groundwater level changes over a regional scale. Thus, in our opinion, spring discharge becomes a better proxy for groundwater variability at a regional scale and should be more closely influenced by climate phenomena.
In this study, we use spring discharge as a proxy indicator to diagnose the relationship of the large-scale climate patterns [ENSO, PDO, Indian summer monsoon (ISM), and west North Pacific monsoon (WNPM)] with variability of a karst aquifer in the Niangziguan Springs (NS) basin, northern China.
2. Study area and data
a. Study area
The Niangziguan Springs complex lies in the Mianhe River valley, Taihang Mountains, eastern Shanxi Province, and spreads out across about 7 km of the Mianhe River bank (Fig. 1). The main aquifers of the basin are comprised of Cambrian and Ordovician karstic limestone, Quaternary sandstone, and unconsolidated sediments. The limestone and Quaternary sediment aquifers are hydraulically connected (Han et al. 1993). Karst groundwater flows from the north and the south toward NS in the east. At the Mianhe River valley, the springs arise from the occurrence of a geologic unconformity, where groundwater perches on low-permeable strata of dolomite and eventually intersects the land surface, thus creating NS (Hu et al. 2008). Small basins and gentle sloping river valleys are the primary geographic features of the NS basin, and extensive areas of the basin consist of rough hilly terrain where the elevation ranges from 1200 to 1600 m MSL. The western part of the basin is higher than the eastern part, with the general topography of the basin inclining to the east. The Mianhe River valley, where Niangziguan Springs discharges, has the lowest elevation in the NS basin, ranging from 360 to 392 m MSL (Fig. 1). The main outcropping strata in the NS basin are Ordovician carbonate rocks, Carboniferous coal seams, Permian and Triassic detrital formations, and Quaternary deposits.
The hydrogeological map of the NS basin.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0085.1
The NS basin is located in the warm temperate zone with a semiarid continental monsoon climate. The greatest total annual precipitation recorded was 844 mm in 1963 and the least was 292 mm in 1972. The average annual precipitation is 530 mm based on records from 1958 to 2010. As much as 60%–70% of annual precipitation usually occurs in the months from July to September. The average annual temperature in the basin is 10.9°C. The average monthly barometric pressure is 932 hPa (1954–2008).
The NS basin is the largest karst spring in northern China. According to records from 1959 to 2011, the NS complex had an annual average discharge of 9.81 m3 s−1, a maximum monthly flow of 18.10 m3 s−1 (in September 1985), and a minimum monthly flow of 4.69 m3 s−1 (in March 1995; Fig. 2). NS receives water from a catchment with an area of 7394 km2, covering the city of Yangquan and the counties of Pingding, Heshun, Zuoquan, Xiyang, Yuxian, and Shouyang (Fig. 1). Precipitation is the primary source of recharge to the aquifers in the basin (Han et al. 1993). The recharge rate is 0.27 in the exposed karst region and 0.10 in the buried karst region in the NS basin (Yuan 1982).
The NS discharge and its trend.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0085.1
b. Data
Monthly spring discharge data of NS from June 1958 to December 2011 were collected from the Niangziguan gauge station in the Mianhe River (Han et al. 1993; Liang et al. 2008). The observed monthly precipitation time series at six meteorological stations (Yangquan City, Yuxian County, Shouyang County, Xiyang County, Heshun County, and Zuoquan County) in the NS basin (Fig. 1) were obtained, spanning a time period from January 1959 to December 2010 (Hao et al. 2012). To obtain a precipitation sequence that reflects the main characteristics of precipitation over the NS basin, a principal component analysis was conducted on the six monthly precipitation sequences, and the first principal component explained 90.3% variance (Takio 2014). The observed monthly temperature data from December 1954 to December 2010 and barometric pressure data from December 1954 to December 2008 observed at Yangquan meteorological station (Fig. 1) were also used in this study, which can be downloaded from China Meteorological Administration (2014, unpublished data).
Monthly climate indices of ENSO (Niño-3.4 from 1950 to 2013; NOAA 2014) and PDO dataset (from 1950 to 2011) produced by Mantua (2000) were used in this paper. Monthly mean data of ISM and WNPM indices (both from 1948 to 2013) were provided by Dr. Yoshiyuki Kajikawa (2014, unpublished data).
3. Methods
a. Preprocessing
Considering the strong impact of human activities on hydrological processes including the groundwater cycle (Milly et al. 2008), a systematic method to investigate the influence of large-scale patterns on spring discharge must provide a means for detrending that removes anthropogenic effects (Hanson et al. 2004). In the NS basin, the rates of decline for spring discharge have slowed down over time because of the recent implementation of sustainable development and groundwater resource conservation policies (China Preparatory Committee 2012). The long-term trend (exponential function) is subtracted from the NS discharge, and the residual of the spring discharge (i.e., detrended spring discharge, hereafter referred to as spring discharge) is acquired.
b. SSA-MTM
The multitaper method (MTM) of spectral analysis provides a way for spectral estimation (Thomson 1982; Percival and Walden 1993) and signal reconstruction (e.g., Park 1992) of a time series, which uses a series of tapers that reduce the variance of spectral estimates. This method has been widely applied to problems in geophysical signal analysis, including analyses of atmospheric and oceanic data, paleoclimate data, geochemical tracer data, and seismological data (Vautard et al. 1992; Ghanbari and Bravo 2009). In this paper, the SSA-MTM toolkit is applied to identify dominant oscillation modes in spring discharge, climate indices, and meteorological parameters, and to reconstruct significant quasi-periodic signals (Dettinger et al. 1995; Ghil et al. 2002). A nearly optimal reconstruction can always be obtained through seeking the weighted linear combination of three lowest-order boundary constraints that minimize the mean-square misfit of the reconstructed signal with respect to the raw data series (Mann and Park 1993).
c. CWT
Continuous wavelet transform (CWT), naturally dedicated to nonstationary signals, provides a time–frequency domain decomposition of the signals. In this paper, as a complement to SSA-MTM, CWT is used to identify dominant oscillations in the NS discharge and climate indices, with added features in the time–frequency domain (Torrence and Compo 1998; Holman et al. 2011; Grinsted et al. 2004; Labat et al. 2000; Labat 2005, 2008; Hao et al. 2012). The Morlet wavelet with frequency parameter 
d. WTC
e. Calculation procedure
First, lagged correlations between monthly spring discharge and climate indices (ISM, WNPM, ENSO, and PDO) and meteorological parameters (temperature, barometric pressure, and precipitation) are calculated and discussed. Next, the dominant oscillations in spring discharge are identified by both MTM and CWT. Then, these periodic components of spring discharge are reconstructed in the time domain, using information from the multitaper decomposition via the SSA-MTM toolkit (Ghil et al. 2002). Later, the reconstructed oscillation modes with different periods of spring discharge are correlated with climate indices successively. The climate index corresponding to the maximum correlation is selected and regarded as the primary driver of hydroclimatic variability affecting the spring discharge at the period. Finally, WTC between the selected climate index and spring discharge is calculated and analyzed.
4. Results and discussion
a. Data preprocessing for the NS discharge
The detrended spring discharge.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0085.1
b. Identifying periodic components of spring discharge by spectral analyses
1) MTM
The SSA-MTM toolkit is applied to identify dominant oscillation modes in time series of monthly spring discharge, climate indices, and meteorological parameters. The results are illustrated in Table 1 and Figs. 4 and 5. It is noteworthy that for the oscillations larger than or close to 1 year, the significant level was set to 0.05, whereas the oscillations less than 1 year and larger than 4 months were tested on a more strict significant level of 0.01. This selection is to avoid adopting high frequencies (which are noises) as much as possible.
Significant periodicities detected by MTM.
(left) Multitaper and (right) wavelet analysis for (a) the detrended NS discharge, (b) ISM, (c) WNPM, and (d) ENSO. The colored shading represents the magnitude of CWT coefficients as shown in the color bar. The regions surrounded by the thick black contours are identified as significant regions, and the thick black contour represents the 5% significance level against red noise. COI where the edge effects are not negligible is shown as a lighter shade.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0085.1
As in Fig. 4, but for (a) PDO, (b) temperature, (c) barometric pressure, and (d) precipitation.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0085.1
Significant periodicities of 0.5, 1, 3.4, and 26.8 years were observed in the spring discharge (Figs. 4, 5; Table 1). The periodicities of 0.5 and 1 year were also found in the meteorological parameters and ISM and WNPM indices (Figs. 4, 5; Table 1). Meanwhile, significant periodicities of 1 and 3 years were observed in both ENSO and PDO, and significant periodicities of 9.48 and 31 years were observed in PDO (Figs. 4, 5; Table 1). This suggests that the periodicities of the spring discharge at the intra-annual and annual time scales were related to the monsoon, and the periodicities at the interannual and interdecadal scales were related to the climate teleconnection patterns. In other words, the spring discharge may be dominated by monsoons at the intra-annual and annual scales and by climate teleconnections at the interannual and interdecadal scales.
2) CWT
CWT is also applied to analyze the time–frequency characteristics of the monthly spring discharge, climate indices, and meteorological parameters (Figs. 4, 5).
One of the most striking features for CWT and MTM is that monsoon indices and meteorological parameters exhibit a strong 1-yr periodicity (Figs. 4b,c and 5b–d), which is consistent with the results of a monsoon significantly affecting the climate of northern China (Guo et al. 2003; Tan et al. 2011). Simultaneously, 1-yr periodicity of spring discharge was observed, but the magnitude of coefficients was relatively small. This may suggest that when the climatic signals are transmitted into spring discharge via precipitation infiltration and groundwater propagation, the signals are likely to be attenuated by the karst aquifer process. When precipitation infiltration reaches the saturated zone (i.e., groundwater), groundwater level rises and the groundwater pressure wave propagates through conduits, fractures, and pores. In different porous media (i.e., conduit, fracture, and pore), the groundwater levels are different. Then the groundwater pressure wave propagates and reaches a valley or ground depression where groundwater levels are higher than ground surfaces and, consequently, springs occur. During the above processes, the precipitation signals will be attenuated, changed, and delayed by the aquifer.
c. Reconstructed oscillation modes of spring discharge
To further explore the linkage between spring discharge and climate patterns, the significant periodic components are reconstructed from spring discharge time series and correlated with climate indices (Table 2). The significant periodicities of spring discharge are 1, 3.4, and 26.8 years based on the MTM and CWT analyses (Table 1).
Max correlation coefficients for reconstructed residual spring discharge modes and climate and meteorological indices. Boldface coefficients are significant at the level 0.01, and the subscript number denotes the lag in months at which max negative or positive correlation is reached.
At the 1-yr periodicity, the spring discharge is significantly correlated with monsoon indices at the significant level of 0.01 (Table 2). At the 3.4-yr periodicity, the spring discharge is significantly and negatively correlated with ENSO at the level of 0.01, while at the 26.8-yr periodicity, the spring discharge is significantly and negatively correlated with PDO (Table 2). These findings reveal that the spring discharge oscillations at different time scales were influenced by different climate indices. At the intra-annual and annual scales, the spring discharge was mainly affected by monsoon. At the interannual scale, the spring discharge was dominated by ENSO. At the interdecadal scale, the spring discharge was controlled by PDO (Table 2).
d. WTC between spring discharge and dominated climate indices
We selected ISM, ENSO, and PDO to perform WTC with the residual spring discharge because they showed the best correlations with reconstructed modes of the spring discharge on 1-, 3.4-, and 26.8-yr periodicities, respectively (Table 2). Figure 6 shows the correlations between the reconstructed spring discharge and the climate indices (Fig. 6, left) and WTC between the spring discharge with lags and the climate indices (Fig. 6, right). In the WTC plot, the colored shading represents the magnitude of the coherence values as shown in the color bar, which varies from 0 to 1 and represents the correlation between selected climate indices and the lagged spring discharge in the time–frequency domain.
(left) The correlations between the reconstructed spring discharge and the climate indices and (right) WTC between the spring discharge with lags and the climate indices. (a) Correlation between the reconstructed spring discharge at the 1-yr periodicity and the ISM index, and WTC between the spring discharge with the lag of 3 months and ISM. (b) Correlation between the reconstructed spring discharge at the 3.4-yr periodicity and the ENSO index, and WTC between the spring discharge with the lag of 4 months and ENSO. (c) Correlation between the reconstructed spring discharge at the 26.8-yr periodicity and the PDO index, and WTC between the spring discharge with the lag of 23 months and PDO. The relative phase relationships are shown as arrows, with in-phase pointing right (positive correlation) and antiphase pointing left (negative correlation). As in the CWT plot, the regions surrounded by the thick black contours, representing the 5% significance level against red noise, are identified as significant regions, and COI is shown as a lighter shade.
Citation: Journal of Hydrometeorology 17, 2; 10.1175/JHM-D-15-0085.1
The monsoon indices are significantly correlated with the reconstructed 1-yr spring discharge at the level of 0.01, and ISM shows the strongest correlation (Table 2). The correlation between the reconstructed 1-yr spring discharge with a 3-month lag and ISM can be determined visually in Fig. 6a, which illustrates that the two time series covary in the same direction, suggesting that the two time series are positively correlated. WTC of spring discharge with the lag of 3 months and ISM illustrates an obvious periodicity of 1 year (Fig. 6a). The arrows approximately point to the right from 1960 to 1980, representing the positive correlation between the spring discharge and ISM during this period. These results suggest that in the 1-yr periodicity, ISM is a key and positive driver for the variability in the spring discharge, and it may affect the spring discharge through its modulating local climate conditions (temperature and precipitation). ENSO, with a lead of 4 months, is significantly and negatively correlated with the reconstructed spring discharge in the 3.4-yr periodicity (Table 2). The calculated WTC between ENSO and the spring discharge with a 4-month lag shows that they are strongly and negatively correlated at the 3.4-yr scale, with arrows approximately pointing to the left from 1985 through 1997 (Fig. 6b). This further confirms that the spring discharge oscillation at the interannual scale is mainly dominated by ENSO. PDO shows the strongest correlation with the reconstructed spring discharge in the 26.8-yr periodicity at the level of 0.01 (Table 2). A negative correlation is observed between the reconstructed spring discharge (in the 26.8-yr periodicity) and PDO at a 23-month lag (Fig. 6c). The arrows on the 26.8-yr periodicity in the WTC plot approximately point to the right during the entire observed period, representing the negative correlation between the spring discharge and PDO. PDO is a decadal or interdecadal variability in SST over the North Pacific sector. The antiphase oscillation at the same periodicity between the reconstructed spring discharge and the PDO index may suggest that the variability in the spring discharge at the interdecadal scale is mainly dominated by PDO.
5. Summary and conclusions
Karst aquifers are highly heterogeneous in nature. They are dominated by secondary or tertiary porosity (i.e., fractures or conduits, respectively) and may exhibit hierarchical permeability structures or flow paths. Different observation wells may have totally different groundwater levels in a karst aquifer. It is difficult, and sometimes impossible, to find a representative groundwater level as the proxy of a karst aquifer. In other words, the groundwater level measured at a particular location is a localized phenomenon and does not represent the regional behavior of a karst aquifer. On the other hand, a karst spring is a natural discharge point of an aquifer, and its discharge variation reflects the combined information of permeability and groundwater level changes of the aquifer over a regional scale. Thus, in our opinion, spring discharge becomes a better proxy for groundwater variability at a regional scale and should be more closely influenced by climate phenomena. The effects of large-scale climate patterns on karst hydrological processes in NS, northern China, were investigated in this paper. Three significant oscillations with 1-, 3.4-, and 26.8-yr periods were identified in the variability of the spring discharge. The spring discharge was significantly and positively correlated with ISM in the 1-yr periodicity and negatively correlated with ENSO and PDO in the 3.4-yr periodicity and the 26.8-yr periodicity, respectively. Moreover, the strongest correlations between the spring discharge and ISM, ENSO, and PDO were observed at the time lag of 3, 4, and 23 months, respectively. In other words, ISM, ENSO, and PDO controlled the NS discharge at the annual, interannual, and interdecadal time scales with time lags, respectively.
Correlation analyses indicate that local climate conditions significantly impact variations in the NS discharge. Precipitation and barometric pressure are positively correlated with the spring discharge, and temperature is negatively correlated with the spring discharge. Simultaneously, monsoons display a stronger correlation with the spring discharge than other meteorological parameters except the temperature. Strong annual periodicity is observed in meteorological parameters and monsoon. Thus, monsoons affect the spring discharge by exerting impacts on the local climate condition. The positive ENSO or positive PDO results in warmer temperatures, less precipitation, and drier air humidity over much of northern China, weakening the inputs to the hydrological cycle and groundwater recharge and, finally, decreasing the spring discharge. ENSO and PDO influence the variability of the spring discharge at the interannual and interdecadal time scales by affecting the regional climate conditions in northern China.
Spectral analyses including MTM, CWT, and WTC were used in this paper. These methods complemented each other and confirmed the reliability of results. The significant periods of spring discharge, climate indices, and meteorological parameters were determined by the consistent results of MTM and CWT. The correlations between the reconstructed spring discharge and the climate patterns were confirmed by the nearly identical results of WTC.
Acknowledgments
This work is partially supported by National Natural Science Foundation of China Grants 41272245, 41402210, 40972165, and 40572150 and the Open Research Fund of State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, China Institute of Water Resources and Hydropower Research, Grant NO IWHR-SKL-201403. Our thanks also extend to Dr. Yoshiyuki Kajikawa for providing the monthly mean data of ISM and WNPM indices. The authors sincerely thank two anonymous reviewers for their detailed and constructive comments on both the content and language to help us improve the quality and presentation of this manuscript.
REFERENCES
Appenzeller, C., Stocker T. F. , and Anklin M. , 1998: North Atlantic Oscillation dynamics recorded in Greenland ice cores. Science, 282, 446–449, doi:10.1126/science.282.5388.446.
Bunn, M. I., Jones J. P. , Endres J. P. , and Rudolph D. L. , 2010: Effects of hydraulic conductivity heterogeneity on vadose zone response to pumping in an unconfined aquifer. J. Hydrol., 387, 90–104, doi:10.1016/j.jhydrol.2010.03.036.
Cardenas, M. B., and Wilson J. L. , 2006: The influence of ambient groundwater discharge on exchange zones induced by current–bedform interactions. J. Hydrol., 331, 103–109, doi:10.1016/j.jhydrol.2006.05.012.
Carrera-Hernández, J. J., Smerdon B. D. , and Mendoza C. A. , 2012: Estimating groundwater recharge through unsaturated flow modeling: Sensitivity to boundary conditions and vertical discretization. J. Hydrol., 452–453, 90–101, doi:10.1016/j.jhydrol.2012.05.039.
Chen, Z., Grasby S. E. , and Osadetz K. G. , 2004: Relation between climate variability and groundwater levels in the upper carbonate aquifer, southern Manitoba, Canada. J. Hydrol., 290, 43–62, doi:10.1016/j.jhydrol.2003.11.029.
China Preparatory Committee, 2012: The People’s Republic of China national report on sustainable development. Rep. to the UN, People’s Publishing House, 100 pp.
Cuthbert, M. O., Mackay R. , Tellam J. H. , and Thatcher K. E. , 2010: Combining unsaturated and saturated hydraulic observations to understand and estimate groundwater recharge through glacial till. J. Hydrol., 391, 263–276, doi:10.1016/j.jhydrol.2010.07.025.
Dettinger, M. D., Ghil M. , Strong C. M. , Weibel W. , and Yiou P. , 1995: Software expedites singular-spectrum analysis of noisy time series. Eos, Trans. Amer. Geophys. Union, 76, 12–21, doi:10.1029/EO076i002p00012.
Gedalof, Z., and Smith D. J. , 2001: Interdecadal climate variability and regime-scale shifts in Pacific North America. Geophys. Res. Lett., 28, 1515–1518, doi:10.1029/2000GL011779.
Ghanbari, R. N., and Bravo H. R. , 2009: Trend and oscillations in the ice-cover duration of Lake Mendota, Wisconsin, USA. Hydrol. Sci. J., 54, 497–512, doi:10.1623/hysj.54.3.497.
Ghil, M., and Coauthors, 2002: Advanced spectral methods for climatic time series. Rev. Geophys., 40, 1003, doi:10.1029/2000RG000092.
Green, T. R., Taniguchi M. , Kooi H. , Gurdak J. J. , and Allen D. M. , 2011: Beneath the surface of global change: Impacts of climate change on groundwater. J. Hydrol., 405, 532–560, doi:10.1016/j.jhydrol.2011.05.002.
Grinsted, A., Moore J. C. , and Jevrejeva S. , 2004: Application of the cross wavelet transform and wavelet transform coherence to geophysical time series. Nonlinear Processes Geophys., 11, 561–566, doi:10.5194/npg-11-561-2004.
Guo, Q., Cai J. , Shao X. , and Sha W. , 2003: Interdecadal variability of East-Asian summer monsoon and its impact on the climate of China (in Chinese). Acta Geogr. Sin., 58, 569–576.
Gurdak, J. J., Hanson R. T. , McMahon P. B. , Bruce B. W. , McCray J. E. , Thyne G. D. , and Reedy R. C. , 2007: Climate variability controls on unsaturated water and chemical movement, High Plains aquifer, USA. Vadose Zone J., 6, 533–547, doi:10.2136/vzj2006.0087.
Han, X., Lu R. , and Li Q. , 1993: Karst Water System: A Study on Big Karst Spring in Shanxi (in Chinese).Geological Publishing House, 340 pp.
Hanson, R. T., Newhouse M. W. , and Dettinger M. D. , 2004: A methodology to assess relations between climatic variability and variations in hydrologic time series in the southwestern United States. J. Hydrol., 287, 252–269, doi:10.1016/j.jhydrol.2003.10.006.
Hao, Y., Yeh T.-C. J. , Gao Z. , Wang Y. , and Zhao Y. , 2006: A gray system model for studying the response to climatic change: The Liulin karst springs, China. J. Hydrol., 328, 668–676, doi:10.1016/j.jhydrol.2006.01.022.
Hao, Y., Liu G. , Li H. , Zhao J. , Wang Y. , and Yeh T.-C. J. , 2012: Investigation of karstic hydrological processes of Niangziguan Springs (north China) using wavelet analyses. Hydrol. Processes, 26, 3062–3069, doi:10.1002/hyp.8265.
Holman, I. P., Rivas-Casado M. , Bloomfield J. P. , and Gurdak J. J. , 2011: Identifying non-stationary groundwater level response to North Atlantic ocean–atmosphere teleconnection patterns using wavelet coherence. Hydrogeol. J., 19, 1269–1278, doi:10.1007/s10040-011-0755-9.
Hu, C., Hao Y. , Yeh T.-C. J. , Pang B. , and Wu Z. , 2008: Simulation of spring flows from a karst aquifer with an artificial neural network. Hydrol. Processes, 22, 596–604, doi:10.1002/hyp.6625.
Jan, C. D., Chen T. H. , and Lo W. C. , 2007: Effect of rainfall intensity and distribution on groundwater level fluctuations. J. Hydrol., 332, 348–360, doi:10.1016/j.jhydrol.2006.07.010.
Klove, B., and Coauthors, 2014: Climate change impacts on groundwater and dependent ecosystems. J. Hydrol., 518, 250–266, doi:10.1016/j.jhydrol.2013.06.037.
Kuss, A. J. M., and Gurdak J. J. , 2014: Groundwater level response in U.S. principal aquifers to ENSO, NAO, PDO, and AMO. J. Hydrol., 519, 1939–1952, doi:10.1016/j.jhydrol.2014.09.069.
Labat, D., 2005: Recent advances in wavelet analyses: Part 1. A review of concepts. J. Hydrol., 314, 275–288, doi:10.1016/j.jhydrol.2005.04.003.
Labat, D., 2008: Wavelet analysis of the annual discharge records of the world’s largest rivers. Adv. Water Resour., 31, 109–117, doi:10.1016/j.advwatres.2007.07.004.
Labat, D., Ababou R. , and Mangin A. , 2000: Rainfall–runoff relations for karstic springs. Part II: Continuous wavelet and discrete orthogonal multiresolution analyses. J. Hydrol., 238, 149–178, doi:10.1016/S0022-1694(00)00322-X.
Lee, B., Hamm B. , Jang S. , Cheong J. , and Kim G. , 2014: Relationship between groundwater and climate change in South Korea. Geosci. J., 18, 209–218, doi:10.1007/s12303-013-0062-7.
Lee, K. S., Wenner D. B. , and Lee I. , 1999: Using H- and O-isotopic data for estimating the relative contributions of rainy and dry season precipitation to groundwater: Example from Cheju Island, Korea. J. Hydrol., 222, 65–74, doi:10.1016/S0022-1694(99)00099-2.
Li, L., and Barry D. , 2000: Wave-induced beach groundwater flow. Adv. Water Resour., 23, 325–337, doi:10.1016/S0309-1708(99)00032-9.
Liang, Y., Han X. , and Xue F. , 2008. Water Resources Conservation for Karst Spring Basin of Shanxi Province (in Chinese). China Water and Power Press, 256 pp.
Mann, M. E., and Park J. , 1993: Spatial correlations of interdecadal variation in global surface temperatures. Geophys. Res. Lett., 20, 1055–1058, doi:10.1029/93GL00752.
Mantua, N., 2000: PDO INDEX. Joint Institute for the Study of the Atmosphere and Ocean, accessed 19 September 2014. [Available online at http://jisao.washington.edu/pdo/PDO.latest.]
Milly, P. C. D., Betancourt J. , Falkenmark M. , Hirsch R. M. , Kundzewicz Z. W. , Lettenmaier D. P. , and Stouffer R. J. , 2008: Stationarity is dead: Whither water management? Science, 319, 573–574, doi:10.1126/science.1151915.
NOAA, 2014: Monthly atmospheric and SST indices. Subset used: monthly sea surface temperature (SST), accessed 19 September 2014. [Available online at http://www.cpc.ncep.noaa.gov/data/indices/ersst3b.nino.mth.81-10.ascii.]
Odling, N. E., Serrano R. P. , Hussein M. E. A. , Riva M. , and Guadagnini A. , 2015: Detecting the vulnerability of groundwater in semi-confined aquifers using barometric response functions. J. Hydrol., 520, 143–156, doi:10.1016/j.jhydrol.2014.11.016.
Park, J., 1992: Envelope estimation for quasi-periodic geophysical signals in noise: A multitaper approach. Statistics in the Environmental and Earth Sciences, A. T. Walden and P. Guttorp, Eds., Edward Arnold, 189–219.
Percival, D. B., and Walden A. T. , 1993: Spectral Analysis for Physical Applications: Multitaper and Conventional Univariate Techniques. Cambridge University Press, 583 pp., doi:10.1017/CBO9780511622762.
Perez-Valdivia, C., Sauchyn D. , and Vanstone J. , 2012: Groundwater levels and teleconnection patterns in the Canadian Prairies. Water Resour. Res., 48, W07516, doi:10.1029/2011WR010930.
Takio, K., 2014: Principal component analysis (PCA). Computer Vision, I. Katsushi, Springer, 636–639.
Tan, L., Cai Y. , An Z. , Yi L. , Zhang H. , and Qin S. , 2011: Climate patterns in north central China during the last 1800 yr and their possible driving force. Climate Past, 7, 685–692, doi:10.5194/cp-7-685-2011.
Taylor, R., and Coauthors, 2013: Groundwater and climate change. Nat. Climate Change, 3, 322–329, doi:10.1038/nclimate1744.
Thomson, D. J., 1982: Spectrum estimation and harmonic analysis. Proc. IEEE, 70, 1055–1096, doi:10.1109/PROC.1982.12433.
Torrence, C., and Compo G. P. , 1998: A practical guide to wavelet analysis. Bull. Amer. Meteor. Soc., 79, 61–78, doi:10.1175/1520-0477(1998)079<0061:APGTWA>2.0.CO;2.
Torrence, C., and Webster P. J. , 1999: Interdecadal changes in the ENSO–monsoon system. J. Climate, 12, 2679–2690, doi:10.1175/1520-0442(1999)012<2679:ICITEM>2.0.CO;2.
Treidel, H., Martin-Bordes J. J. , and Gurdak J. J. , 2012: Climate Change Effects on Groundwater Resources: A Global Synthesis of Findings and Recommendations.CRC Press, 414 pp.
Tremblay, L., Larocque M. , Anctil F. , and Rivard C. , 2011: Teleconnections and interannual variability in Canadian groundwater levels. J. Hydrol., 410, 178–188, doi:10.1016/j.jhydrol.2011.09.013.
Vautard, R., Yiou P. , and Ghie M. , 1992: Singular-spectrum analysis: A toolkit for short, noisy chaotic signals. Physica D, 58, 95–126, doi:10.1016/0167-2789(92)90103-T.
Yuan, C., 1982: Calculation of recharge rates of Niangziguan Springs basin by using Turc equation (in Chinese). Karst and Karst Water in Northern China, Geological Publishing House, 122–126.
Zektser, I. S., and Loaiciga A. , 1993: Groundwater fluxes in the global hydrologic cycle: past, present and future. J. Hydrol., 144, 405–427, doi:10.1016/0022-1694(93)90182-9.
