What Produced Water Tells You About Aquifer Risk
A Marcellus Shale well in Pennsylvania can surface 150,000 milligrams of dissolved solids per liter. That number sits at three hundred times the EPA secondary drinking water recommendation of 500 milligrams per liter. The produced water that accompanies natural gas is not a dilute waste stream. It is a concentrated brine, and the engineer who treats it as generic industrial effluent makes a serious mistake.
The USGS National Produced Waters Geochemical Database holds records for more than 520,000 samples from oil and gas operations across the United States. The Mid-Atlantic subset covers Pennsylvania, New York, New Jersey, Delaware, and Maryland. These five states contain both deep Marcellus Shale wells and shallow conventional formations, and the chemical difference between the two matters enormously for risk assessment.
Ionic strength is the first number that matters. A water sample’s ionic strength summarizes all dissolved ion concentrations, weighted by the square of each ion’s charge. Fresh groundwater from a shallow aquifer sits near 0.005 molal. Marcellus brine can reach 5 molal, three orders of magnitude higher. At that concentration, dissolved ions do not behave like independent particles. They interact with each other and depart from ideal solution behavior. The Davies activity coefficient translates measured concentrations into activities, the thermodynamically effective values that control mineral formation and dissolution rates.
The calcite saturation index converts those activities into an operational judgment. A fluid with an index above zero tends to deposit calcite wherever it slows down: in wellbore tubing, in surface separators, in injection disposal lines. Scale deposits add tens of thousands of dollars per well per year to maintenance costs. A fluid with an index below zero dissolves carbonate formations and attacks wellbore cement. Cement dissolution creates the migration pathway regulators consider the worst case: a direct connection between a producing formation and a shallow drinking-water source.
Two populations dominate the scatter. Samples in the lower left are acidic and dilute, likely shallow conventional wells with minimal buffering capacity. Samples in the upper right are alkaline and highly concentrated, representing Marcellus and similar deep-formation brines. The two populations present different risks and require different management responses.
Get Kyle Jones’s stories in your inbox
Join Medium for free to get updates from this writer.
Lead and arsenic bring regulators into the room. The EPA maximum contaminant level for lead in drinking water is 10 micrograms per liter. A measurable fraction of samples in the NPWGD Mid-Atlantic dataset exceed that threshold in the produced water itself. This does not mean residential taps carry contaminated water. It means the operator has a documented mass of lead at the surface that requires a tracked disposal pathway. Arsenic exceedances appear at a lower rate but carry the same compliance implication.
The spatial autocorrelation of TDS reinforces what regional geology predicts. The empirical variogram of TDS across Pennsylvania shows correlation structure that reaches practical zero at roughly 50 to 90 kilometers. Chemistry follows stratigraphy, not county lines. A spill at one location does not indicate the ambient water quality 80 kilometers away. That distance matters for emergency response, for monitoring network design, and for background concentration arguments in regulatory proceedings.
The practical output of this analysis is a defensible chemical characterization with explicit connections to regulatory thresholds. An operator who presents ionic strength, saturation index, and metals exceedance rates enters the permit process with quantitative footing. An operator who presents total dissolved solids as the sole descriptor invites follow-up requests that extend timelines and increase legal exposure. The chemistry requires no analysis beyond a standard water quality panel. The discipline is in the systematic application.
All code for this pipeline, including the DuckDB queries, the Davies activity coefficient calculation, and the variogram analysis, is available on GitHub.
Data source: U.S. Geological Survey National Produced Waters Geochemical Database v3.0. Available from USGS ScienceBase.
