Diagnosing Non-Meteorically Induced Variations in the Chemistry of Saline Springs of the Vrancea Seismic Area (Romania)

Abstract

A monitoring operation conducted over more than seven years has been addressing chemical fluctuations displayed by several cool, low-discharge springs located close to the highly seismic Vrancea area (Romania). Those outflows compositions proved to be strongly controlled by binary, essentially isochemical mixing between a deep-origin brine and meteoric freshwater. By taking advantage of this particular setting, there have been constructed diagrams aimed at investigating how the groundwater-discharges Na–K–Mg geothermometric parameters fluctuated as a function of the chloride content (taken to be an estimator of the freshwater-induced dilution). We made use of the reciprocal algebraic relationships existing between the equations describing the Na–K–Mg geothermometric parameters “Na–K temperature” and “K–Mg temperature” on the one hand, and the so-called “maturity index” (MI) on the other. In an accordingly derived plot of MI against the logarithm of the chloride concentration, a series of data-points being quite uniformly off-set from the MI dilution-curve constructed for an apparently “regular” period, suggested that, episodically, all concerned springwaters were simultaneously subject to some similar changes in the controlling geochemical processes. One such modification intervened 3–4 months before the occurrence of the strongest earthquake ($M_{\text{w}} = 5.8$) of the hydrochemical monitoring period. The consequently derived interpretation was that then, the numerical values of certain geothermometric coefficients were likely altered: such a process could be consistent with changes in the alkali feldspars solubility relationships, possibly in response to episodic Al–Si complexing which might develop within a hypothesised, still active, exhumation-channel.

Appendices

Appendix 1: Selecting the Na–K Geothermometric Equation Coefficient-Values that Would be Suitable for the Considered Springwaters

A couple of simultaneous plots of T _Na–K vs. $c_{\text{Cl}}$ and T _Na–K vs. $c_{\text{Cl}}$ have been constructed, by taking into account only the samples collected from the considered springs during the “postseismic regime” time-period (as defined in the main text, Sect. 5.1). For one such plot (Fig. 9a), in the computations there were utilized the α _Na–K and β _Na–K coefficient-values (Table 1) proposed by Giggenbach (1988); in the other plot (Fig. 9b), the computations used the corresponding values proposed by Fournier (1979). Each of the two diagrams includes:

• experimentally obtained $T^\star _{{\text{Na}}{-}{\text{K}}},$ and $T^\star _{{\text{K}}{-}{\text{Mg}}}$ values plotted versus the corresponding $c^\star _{\text{Cl}}$ (they are represented as individual data points, illustrated by various symbols): for each analyzed sample, those $T^\star _{{\text{Na}}{-}{\text{K}}},$ and $T^\star _{{\text{K}}{-}{\text{Mg}}}$ values have been computed by substituting—into Eqs. (1) and (2), respectively–the actually determined concentrations $c^\star _{\text{Na}}$, $c^\star _{\text{K}}$ and $c^\star _{\text{Mg}}$;

• the dilution functions $T^{\prime}_{{\text{Na}}{-}{\text{K}}}$ vs. $c^{\prime}_{\text{Cl}}$ (solid line), and $T^{\prime}_{{\text{K}}{-}{\text{Mg}}}$ vs. $c^{\prime}_{\text{Cl}}$ (dashed line): they are computed by means of Eqs. (6) and (7), respectively, with the values of the involved regression coefficients s and i being retrieved from Table 3.

Fig. 9

Springs No. 10, No. 1bis and No. 3 at Slănic Moldova: plots of T _Na–K vs. Cl, and T _K–Mg vs. Cl, constructed only for the samples collected during the “postseismic regime” time-period. The α _Na–K and β _Na–K values utilized in the computations were: in (a), those proposed by Giggenbach (1988); in (b), those proposed by Fournier (1979). Symbols and curves as specified in Fig. 6 of the main text. The intersection of the two curves should assumedly provide “true” values for the saline parent-water chloride content, as well as for the deep-reservoir temperature. The maximum possible chloride content derived from the SiO₂ versus Cl linear mixing trend (Fig. 8 in the main text) is marked by the vertical solid bar

It is assumed (Chiodini et al. 1996) that the coordinates of the intersection-point between the $T^{\prime}_{{\text{Na}}{-}{\text{K}}}$ vs. $c^{\prime}_{\text{Cl}}$ curve and the $T^{\prime}_{{\text{K}}{-}{\text{Mg}}}$ vs. $c^{\prime}_{\text{Cl}}$ one, should provide a reasonably accurate estimate both for the actual temperature of the deep-reservoir, and for the corresponding chloride content of the saline parent-water ($c^{{\prime}{\text{max}}}_{\text{Cl}}$ ). It appears, however, that such a derivation is not quite unequivocal: when the Na–K geothermometer values proposed by Giggenbach (1988) are used (Fig. 9a), the estimates strikingly differ from those obtained when using the Fournier (1979) α _Na–K and β _Na–K coefficient-values (Fig. 9b).

Fortunately, in the particular case of the Slănic Moldova springs, an independent estimate of the saline endmember chloride concentration ($c^{{\prime}{\text{max}}}_{\text{Cl}}$) could be derived by means of the SiO₂ vs. Cl⁻ crossplot constructed for the “postseismic regime” samples (Fig. 8). In this respect, advantage has been taken of the fact that silica in the deep-origin brine proved to be significantly less abundant than in the freshwater component (low-mineralized shallow groundwater displaying SiO₂ concentrations larger than those of nearby mineral or thermal springwatersthat have been previously mentioned also by Gökgöz and Tarcan 2006, and by Weaver et al. 2006).

There was consequently possible to extrapolate the reverse linear mixing-trend displayed by the SiO₂ vs. Cl⁻ plot (Fig. 8, with regression parameters indicated in Table 3), down to a zero-silica concentration. It was obtained, as a result, a maximum-possible Cl⁻ content of about 19.6 g kg⁻¹. Accordingly, it was inferred that Na–K geothermometric equations coefficients, which provided larger chloride values for the Slănic Moldova parent-water were not applicable: in particular the Giggenbach (1988) Na–K geothermometer function, which led to a maximum Cl⁻ content of 40.7 g kg⁻¹ (Fig. 9a). Alternatively, a much more plausible description of the concerned springwater behaviour appears to be provided by the Na–K geothermometer coefficients proposed by Fournier (1979): the indicated parent-water Cl⁻ content is then 14.6 g kg⁻¹ (Fig. 9b), thus, authorizing one to assume that the saline component still possessed a small amount of dissolved SiO₂ when it mixed with the shallow freshwater. Correspondingly, the deep reservoir temperature resulted in being 146$^{\circ}.$

We, consequently, concluded that in the case of the considered springs, the α _Na–K and β _Na–K coefficient-values provided by Fournier (1979) were the most appropriate for being further used—in conjunction with the α _K–Mg and β _K–Mg geothermometric coefficients provided by Giggenbach (1988)—in the MI values computations.

Appendix 2: Extending to the Currently Addressed Springwaters Previous Isotopic Evidence About the Involvement of a Non-Meteoric Component

On two distinct occasions (14 July 2003 and 25 June 2004), simultaneously with the springs No. 1bis, No. 3 and No. 10, we had also sampled two saline water discharges located nearby. They had been previously discussed also by Vaselli et al. (2002), who designated them as Slanic 14 and Slanic 15.

It is important to stipulate that for the Slănic 15 water sample, Vaselli et al. (2002) have provided both chemical, and isotopic ($\delta$D and $\delta ^{18}$O % SMOW) analytical results. The only other groundwater discharge at Slănic Moldova for which the indicated authors had provided a similar kind of information was that designated by them as Slănic W2 (which we did not sample). Both the sample Slănic 15, and the sample Slănic W2 display in the $\delta ^{18}$O - $\delta$D diagram discussed in Vaselli et al. (2002) significant $^{18}$O enrichments with respect to the meteoric water line, thus, strongly suggesting a non-meteoric fluid contribution to the corresponding discharges.

Fig. 10

Plot of Na⁺ versus Cl⁻, aimed at comparing Slănic Moldova samples included in the study of Vaselli et al. (2002) (dots), with samples collected (at 14 Jul. 2003 and 25 Jun. 2004) in the framework of the present study (triangles). Most data-points are tightly aligned, which suggests that except for the sample “Slanic W2”, all spingwaters originate in a common saline endmember (presumably, of deep-origin), that is subject to various amounts of freshwater dilution. When comparison is made with a fluid derived from dilution of seawater (dotted line), the computed mixing trend indicates a significant Na⁺ enrichment; there is also an obvious Na⁺ excess with respect to the stoichiometric dissolution of halite (dashed line)

While for all the samples we collected on 14 July 2003 and 25 June 2004, complete chemical analyses were performed, we did not have the possibility of conducting isotopic determinations as well. Therefore, we attempted to check (by means of a Na⁺ vs. Cl⁻ cross-plot—Fig. 10) if the same deep-origin endmember was involved both in the supply of the springs chiefly addressed by the present study (No. 1bis, No. 3 and No. 10), and in that of the discharges Slănic 15 and Slănic W2, which had already been proven, by means of isotopic data Vaselli et al. (2002), to include a non-meteoric component.

In Fig. 10 there can be noticed a very good alignment of most data-points—the Slănic 15 samples included—on a well-defined Na vs. Cl mixing line. It, hence, seems quite probable that the water discharged by the springs No. 1bis, No. 3 and No. 10 includes the same non-meteoric origin endmember as that involved in the supply of the Slănic 15 outflow. In contrast, the sample Slănic W2 plots rather far away from the indicated mixing line, suggesting that certain additional components—or processes—probably contribute to shaping its composition.