Plotting specific conductance versus total dissolved solids (TDS) from measured water quality data from water wells in the Hill Country region of Texas, United States, yields two distinct trends (Figure 1). This presents a problem when estimating TDS in this region, either from measuring specific conductance from water samples, or from calculating formation water resistivity from geophysical well logs. The two trends were visually isolated (Figures 2 and 3) for comparison of chemical, spatial, and formational patterns between the two. Analysis of the two trends reveals the water for each is identical in terms of major and common minor constituents (Figure 4), the two trends occur in all formations of the Trinity aquifer system, and there is no overall regional pattern of one trend vs the other, though the trends locally cluster (Figure 5, example from the Upper Glen Rose Formation). Water of both trends are dominated by Ca-Mg-HCO3 and Ca-Mg-SO4, and as TDS increases, calcium-magnesium-sulfate becomes the dominant water type, with up to 94% of the total milliequivalents per liter comprised of Ca-Mg-SO4.
Initial thoughts were the difference in the trends could be explained by ion complexing in some areas but not others, perhaps catalyzed by subtle geochemical constituents in the subsurface, or Eh-pH conditions (there is no significant difference in pH between the two trends). But current thoughts are that ion complexing cannot explain a nearly 50% reduction in specific conductivity from the lower trend to the upper trend (given the same TDS value), which would require nearly 50% of the ions to form complexes. Other hypothesis is the difference between the two trends could be due to bias in sampling and analysis. Perhaps for the upper trend nearly 50% of the Ca-Mg-SO4 precipitates as solid CaSO4 and MgSO4 when sampled and specific conductance is measured (measuring a low specific conductance relative to the weight of constituents in the water), whereas the sample is then mixed at the laboratory and the precipitate is re-dissolved before analyzing the concentration of constituents. This is only a conjecture. I am not a geochemist and I have rudimentary knowledge of water sampling procedures – my background is primarily geologic mapping and structural geology.
Points of discussion regarding this are:
· What are the potential explanations for two significantly different trends of TDS versus specific conductance for otherwise identical water types? Keep in mind the water types in this case are dominantly Ca-Mg-HCO3 and Ca-Mg-SO4 (the latter dominating at higher TDS)
· Could ion complexing be responsible, or is the magnitude of decreased specific conductance demonstrated above too great?
· Do you know of any other studies where two trends of TDS versus specific conductance has been examined for a region, or other relevant publications?
· Which of the two trends are more likely to occur in the subsurface? Or is it likely they both occur locally in the subsurface? (Our goal is to estimate TDS from resistivity logs; thus, we need a single, reliable TDS vs specific conductance trend)
· Are you familiar with collecting, sampling, and analyzing water samples, and have some insight into whether the specific conductance could measure relatively very low for a given quality or water?
A copy of the data is attached (there are two tabs, one for each trend). Thank you for considering this problem!
Evan Strickland, P.G. | Geoscientist
Texas Water Development Board
Innovative Water Technologies | BRACS team
512-463-6929 | [email protected]
Well,
The most accurate is to measure specific conductance from water samples.
The measuring of EC as 1/ resistivity from the geophysical well- logs means you measure the conductivity of both water and formation media.
Here and in the lower trend relation, the porous media contains clays or shale of high potential conductance(means low resistivity).
The reverse case can be seen in the upper trend relation(low percent of clay content and reflects high resistivity means low EC).
The ADVICE here to use the measurements of water samples as a base to define the relation of EC-TDS for the groundwater, while the other trend shows the relation according to the aquifer media + groundwater.
Regards
Thank you Bayan Hussien . The problem here is, as you advise, we are attempting to establish an EC-TDS relationship with water samples (the data plotted above is measured from water samples). However, two trends emerge without any chemical, formational, or regional pattern that I can discern between the two. I otherwise agree with your remarks about the effects of clay content on resistivity log measurements (which we have yet to interpret). Thank you for your consideration of this problem!
Just a thought, have you tried plotting well locations over soil type?
I don’t have much experience in groundwater but I know water temperature can have a large effect on conductivity measurements. It’s possible the two distinct trends arise from different sampling handling procedures, which is an important issue to resolve early on. For instance, was specific conductance reported by a meter that automatically corrected to 25 C or were water samples brought to 25 C before measuring? There could be other discrepancies in methods...that’s just one idea.
Thank you James P. Guinnip . I have not tried comparing locations to soil type. Most wells are screened to depths greater than a hundred feet, so rather I've divided the samples by geologic formation to find the two trends occur in all of the formations to some extent. I had thought about temperature and sampling procedures, but from my reading, measuring specific conductance in the field and converting to 25 C seems to have been standard in the procedures and equipment for a long time.
Good point about the well depth making geology more relevant there. Were all these measurements collected recently over a short period of time?
From personal experience, I wouldn't be so quick to ignore the aspect of sampling procedures. I spent a lot of time trying to model a nonlinear, decadal trend in water quality for my dissertation only to find out it was caused by sample contamination in an analytical lab that apparently never changed their procedures. Turns out people are lazy/stubborn and not always transparent about changes in their data collection. If these samples were collected over many years, I'd recommend consulting with more senior colleagues in your organization to get additional perspective on specific methods.
Did you take into consideration how each ion imparts its specific conductance? Note from the article "The Use of Specific Conductance..." how dissolved organic matter (DOM) played a large roll in the relationship you are developing. Unfortunately, you are relying on data performed years ago which did not go into the depth of analysis to better provide what you need.
To articles are attached for your review.
James P. Guinnip , thank you for sharing your personal experience! I will do as you suggest and reach out to one of our water sampling experts. As for your other question, the samples are from the 1950's to the recent. However, both trends occur from then to now (based on my separation of the points in the graphs above, and plotted against date).
Thank you Gary Engelgau for your input and the useful reference on dissolved organic matter. Yes, I had considered the effects of each ion by calculating specific conductance from the concentrations of dissolved ions from my water quality analyses, using equations from the USGS report you attached (WSP 2311). The calculation yielded a linear relationship of TDS to SC similar to the lower, linear trend from my graphs above. Applying one of their median values of an "exponential correction factor" ('f' in their paper) to account for ion complexing aligned the calculated SC linear trend with that of the measured SC linear trend. Interestingly, since posting this discussion I found an unpublished report of the same study area I am looking at (prior to me joining the agency). They ran into the same confusion with the two TDS vs SC trends I describe. In their investigation, they modeled the specific conductance with the USGS program WATEQ4F. Interestingly, with the available water analyses, that program modeled specific conductance that matches the upper, curved trend very well. I have attached an example of the two graphs, one with a linear TDS-SC trend calculated from the USGS WSP 2311 paper, and the other curved TDS-SC trend modeled by WATEQ4F in the unpublished report (each calculated trend matches one of the two measured TDS-SC trends). Unfortunately, the unpublished report does not describe their inputs into WATEQ4F. My ultimate suspicion is that both trends are real, where the lower, linear trend the water has undergone limited ion complexing, and the upper, curved trend, the water has undergone significant ion complexing, however, the cause for which I have little knowledge or imagination to guess.
Assume two sets of water types: one dominated by monovalent ions, another dominated by bivalent ions. Check this against a Piper plot.
James P. Guinnip you were spot on when you questioned analytical procedures! The lower linear trend mostly corresponds to samples analyzed from a particular laboratory that measured "diluted conductance", rather than specific conductance. The methodology in measuring diluted conductance would certainly negate any effects of ion pairing, which is substantial for the Ca-Mg-SO4 system I am dealing with, and which effect is represented by the upper curved trend. Thanks again for your input, and stressing the importance of scrutinizing procedures, especially for historic data! Cheers.
Credit is due to F. Paul Bertetti, P.G., who identified this lab as a source of erroneous conductance measurements in an unpublished report.
- Badges
- Science topic