The harmonic analysis of the time series of waterhead variations on the network of piezometers detects the presence of four distinctive periodicities: a decadal cycle, a cycle of 3.2 years, an annual cycle and a semi-annual cycle. The decadal cycle is related to the climatic 11-year cycle, in turn related to sunspot activity in the context of the North Atlantic Oscillation (NAO). The 3.2-year cycles could be related to the climatic cycle of the NAO (it has been recognized that the NAO has an influence on climate on a cyclic basis known as the quasibiennial oscillation, which ranges between two and four years). The annual cycle is related to the hydrological annual cycle, and similarly the half-year cycle is related to two precipitation seasons in a single year. Table 1 gives the percentages presented by the cycles according to the established confidence intervals.

The 11-year cycle is apparent in most of the piezometric series (85%), marking one of the

Table 1. Percentage of cycles according to confidence intervals


Level of significance (%)



































ND, Not distinguishable; D, distinguishable.

ND, Not distinguishable; D, distinguishable.

main features of the temporal behaviour of the aquifer piezometric level, together with the annual cycle. The 3.2-year cycle, present in one-third of the series, also has a sparse distribution throughout the aquifer. The annual cycle appears in all the piezometric series, as expected. Finally, the semiannual cycle is weakly represented, in just 21% of the series. This might be related to a precipitation cycle (in this area there are two precipitation seasons in a single year), which would have a minor influence on the recharging of the aquifer.

Figure 5 shows the typical power spectra found in the harmonic analysis. Not all cycles are equal in their presence and intensity in the power spectra of the different piezometers. As shown in Table 1, while decadal and annual cycles are detected in most of the piezometers, the 3.2-year and half-year cycles are detected only in a set of them. According to our probabilistic estimation, there is also a difference in the statistical significance of every cycle from one piezometer to the next and, consequently, there is a spatial variability in their importance in different areas of the aquifer.

The procedure used to assess this spatial variability was to give a code to every cycle at each piezometer according to its statistical significance in the estimated power spectrum of each piezometer. The categorical code used was:

the cycle is not distinguishable in the power spectrum;

the cycle is distinguishable but is not statistically significant at the 90% level; the cycle is statistically significant at the 90% level but not at the 95% level; the cycle is statistically significant at the 95% level but not at the 99% level; the cycle is statistically significant at the 99% level.

The previous categorization, although arbitrary, allows us to highlight differences in the behaviour of cycles when the categories are represented on maps. Figure 6a-d present the spatial distribution of the statistical significance of peaks for the decadal, 3.2-year, annual and semi-annual cycles, respectively.

Figure 6a presents the spatial distribution of the statistical significance of the decadal cycle. The most significant presence of this cycle (category 4) can be seen in the borders of the aquifer and in relation to rivers and other watercourses that represent input of water from an important drainage network. We believe this is because climatic variations of around 11 years, related to the sunspot cycle, may only be seen in the variability of rainfall when integrated in an area such as the drainage basin. This implies a variability in the amount of water that enters into the aquifer from surface drainage networks; it would be detectable near the mouth of those tributary channels, but its effect is damped in the interior of the aquifer or farther away from the surface watercourse. The lesser significance of the 11-year cycle in the central area can be attributed to the small variation of the water level in this zone, where the signs of this cycle would be reduced. In the eastern sector, in contrast, the fluctuations in the piezometric level are pronounced, and decreases or increases accumulate, forming waves over an approximately 11-year period, well reflected in the spectral analysis. Moreover, for this cycle we found that the piezometric series were highly parallel to graphs of the NAO index (Hurrell 1995) in that the more positive the NAO index, the lower the piezometric levels. In this sense, it is generally accepted that the NAO marks climatic behaviour at these latitudes (Hurrel 1995; Qian et al. 2000; Rodrigo et al. 2000), though some authors relate this cycle to sunspot activity (Eddy 1976; Reid 1993).

Most researchers point to the NAO as the cause of the three-year cycle (Pozo-Vazquez et al. 2000), although some studies conclude that the behaviour of the rainfall in this region could be influenced by the El Nino South Oscillation (ENSO) (Rodo et al. 1997). Thus, the statistically significant presence of this cycle would be produced in a set of piezometers located in sectors where the vertical permeability in the proximities of surface currents

Fig. 5. Power spectra piezometric levels represented in Figure 3(a-d) from Blackman-Tukey.

is high. There is also a preferential location at points near the aquifer borders, or by torrential watersheds and more permeable detritic stretches. The NAO cycle is a subtle one that may be difficult to detect because its influence may be small in comparison with the decadal and annual cycles. We believe this cycle is exhibited in the form of minor pulses on the piezometric levels that will essentially coincide with stormy episodes. The lack of statistical significance or the absence of this cycle from other piezometers of the aquifer could be attributed mainly to a lower permeability of the sector where the piezometer is located, and to the recharge being more localized.

Figure 6c reflects how the annual cycle is highly significant (category 4) in most of the piezometers. This ubiquitous presence is logical because it is related to the yearly hydrological cycle, which affects the whole aquifer. From Figure 6d it may be seen how the semi-annual cycle is detected only at some piezometers. While there is no simple explanation for its spatial distribution, it might be related to some other climatic cycle (in this area there are two precipitation seasons in a single year).

Was this article helpful?

0 0

Post a comment