Insights ·
Water quality investigations that hold up when the data are challenged
A defensible water quality investigation starts with the decision the data have to support—delineation, trend, a receptor, or a source question—and then shows that the wells, the sampling method, and the laboratory package actually speak to that decision. A table of exceedances is not an investigation.
Write the question before the analyte list
Delineating a plume, testing whether a remedy is working, and asking whether two water-bearing zones are connected are three different sampling programs. Analytes, well locations, and vertical placement should follow the question. Adding every constituent the lab can report feels thorough and usually dilutes the record.
That design step is the core of our water quality investigations. The program should also say what result would change the next round of work, so adaptive sampling is a plan and not an afterthought.
The screen has to match the unit you think you sampled
A concentration belongs to an interval, a purge method, and a date. Long screens that straddle two units mix waters and invent a “representative” sample of neither. Wells with silt, a damaged seal, or a screen in fill can dominate a contour map if nobody reads the construction log.
When those results are later used for hydrogeology interpretations or a transport model, the bias travels with them. A usability review asks, for each critical well: what interval, what method, what turbidity or stabilization record, and is there a reason this point should be down-weighted.
QA/QC and legacy data
Blanks, duplicates, holding times, and detection limits are part of the opinion, not an appendix nobody opens. A non-detect from 1998 and a non-detect from last year are not the same fact if the reporting limit dropped by an order of magnitude. Trend claims that ignore that change are easy to take apart.
Historic databases can still be usable. The work is to separate results that remain fit for the question from results that only show that someone sampled. We do that review often when an old dataset becomes the fight.
Trends need a hydrogeologic reason
A line through a handful of events is not a trend. Enough independent samples, a stable well, and a hydrogeologic reason the direction makes sense—capture, a new source, a change in recharge, a shift in redox—should all be on the page. Seasonal water-level swings can mimic a concentration trend if the evaluation ignores head.
Chemistry that might distinguish sources belongs with the physical program, not in a separate stack of plots. The article on hydrogeochemistry and source questions picks up there.
Questions counsel ask
- What is a defensible water quality investigation?
- One where the decision drives the sampling plan, field methods and well construction are documented, laboratory QA/QC is reviewed, and the interpretation states limitations and alternative explanations.
- Can historic groundwater data still be used?
- Often yes, after a usability review. Changing detection limits, long well screens, and missing purge records are typical reasons to down-weight or set aside individual results rather than the entire database.
- How do groundwater and surface-water programs differ?
- Connectivity, mixing, seasonality, and the receptor pathway differ. The media and the question should drive the design. A stream sample and a monitoring well are not interchangeable evidence.
Daniel S. Fisher, P.G.
Owner and principal hydrogeologist, Wetstone Solutions LLC. Pennsylvania professional geologist since 1994. This note is for education. It is not legal advice and not an opinion on any matter.
Related reading
Request a conflict check →