Treating abandoned mine drainage can protect streams cost effectively and benefit vulnerable communities

The study addresses who benefits from large new US investments in cleaning up legacy pollution from abandoned coal mines and what those investments can accomplish, focusing on abandoned mine drainage (AMD) in Pennsylvania. AMD is acidic, metal-laden water that impairs aquatic ecosystems and poses human health risks. Despite decades of mining prior to modern regulations, thousands of stream kilometers in Appalachia remain impaired. The 2021 Infrastructure Investment and Jobs Act (IIJA) appropriates substantial funds to address legacy hazards, but the targeting and effectiveness are uncertain. The authors ask: (1) Are communities most burdened by AMD also more vulnerable to the ongoing energy transition away from coal? (2) What can be expected from new investments based on evidence from past AMD treatment systems? Pennsylvania, with the nation’s largest inventory of abandoned mine liabilities and extensive AMD treatment history and data, provides the case study context and an opportunity to assess both equity and cost-effectiveness dimensions of AMD remediation.

The paper situates AMD within a broader literature on the social and environmental legacies of fossil fuel extraction and the justice implications of energy transitions. Prior work documents the vast extent of AMD-impaired waterways in Appalachia and the historical inadequacy of pre-1977 policies. Studies highlight AMD’s ecological and health impacts (e.g., bioaccumulation of metals, human health risks) and evaluate passive and active treatment approaches. The authors connect their analysis to research on compensatory policies for fossil fuel communities and to ex-post evaluations of water-quality investments, notably Clean Water Act grants, which provide benchmarks for cost-effectiveness and benefits (housing value-based valuations). They also draw on environmental justice metrics (EPA EJSCREEN and CDC EJI) and county-level energy transition vulnerability indices to place AMD burdens within broader environmental and socioeconomic contexts.

Study area and exposure measurement: The analysis focuses on Pennsylvania. AMD exposure is measured at the Census tract level as the percentage of stream length designated as impaired by AMD in the Pennsylvania DEP’s 2022 Integrated Water Quality Report. Tracts are grouped into five categories: no AMD exposure and quartiles of exposure among affected tracts (bottom quartile <7% impaired; top quartile ≥51%). Demographic and economic characteristics (2018–2022 ACS), EPA EJSCREEN composite indices (average of 13 indicators using the EPA demographic index), CDC Environmental Justice Index (burden sub-index), and a county-level energy transition vulnerability index (weighted by carbon intensity of fossil energy production) are merged to characterize communities.

Estimating kilometers of stream protected by treatment systems: The authors assemble data for more than 300 passive AMD treatment systems in Pennsylvania from Datashed, with 265 systems used for inflow/outflow water quality analysis. They link system locations to the National Hydrography Dataset (NHD) stream network and consider downstream segments up to 80 km from each system to capture potential propagation of water quality effects.

Two complementary methods estimate kilometers protected:

• Method 1 (modeled concentrations): For each downstream segment i, the authors compute pollutant loads from upstream system discharges using observed average inflow (no-treatment scenario) and outflow (treatment scenario) concentrations and flows for pH, Fe, Mn, and Al. They combine these loads with each segment’s baseline flow and concentration to derive resulting concentrations under treatment and no-treatment scenarios via mass-balance mixing. Impairment thresholds used are: pH < 6; Fe > 1.5 mg/L; Mn > 2.0 mg/L (EPA effluent limits where state threshold absent); Al > 0.5 mg/L. A segment is impaired if any pollutant violates its threshold. Kilometers protected equal the stream length impaired without treatment minus that impaired with treatment.
• Method 2 (assessment reconciliation): The authors identify segments that would be impaired under Method 1’s no-treatment scenario but are not listed as AMD-impaired in the DEP’s 2022 assessments. Those lengths are attributed as protected by treatment systems, leveraging observed water quality and biological assessments.

Uncertainty: They compute 90% confidence intervals via an adjusted Efron percentile bootstrap, repeatedly resampling stream segments (1,000 samples), recalculating protected kilometers, and deriving percentile-based bounds relative to the original estimate.

Cost estimation and cost-effectiveness: Project documents provide initial capital costs for most systems; for 13 systems, initial cost is imputed using average initial cost per liter times treated flow. Long-term costs assume a 25-year system life with annual operation, maintenance, and reconstruction equal to 4% of initial cost. The present value is calculated using a 3% real discount rate for these costs. Cost per kilometer-year protected is the ratio of the present value of lifetime costs across systems to the product of total protected kilometers and the assumed 25-year lifetime. Confidence intervals for costs mirror the kilometers procedure by allocating each system’s cost proportionally to downstream segments within 80 km and bootstrapping.

Funding needs and availability: Total kilometers requiring protection equals currently impaired kilometers plus those already protected by existing systems (since many are aging and require continued investment). Total funding needed is cost per kilometer-year times total kilometers times 25 years. Available funding combines Pennsylvania’s AMD set-aside balance, projected Abandoned Mine Land (AML) fee disbursements through 2034 (declining with projected coal production), and the IIJA Treasury appropriation over 15 years, discounted at a 5% nominal rate (since disbursements are nominal) with the 2022 IIJA disbursement assumed constant over the appropriation period.

• Exposure and vulnerability: About 2.4 million Pennsylvanians live in tracts with AMD-impaired streams. Tracts in the top quartile of exposure (≥51% of streams impaired) house over 0.5 million people. High-exposure tracts have average median household incomes roughly 30% below unaffected tracts and median housing values about 50% lower. Communities most exposed to AMD are approximately twice as vulnerable to the energy transition as unaffected communities. Within AMD-affected tracts, higher exposure correlates with greater urban population shares, higher percentages of renters and non-white residents, and higher environmental justice scores.
• System performance: Across 265 passive treatment systems, outflows exhibit substantially improved water quality relative to inflows. Average inflow pH is 4.33 (approximately tomato juice acidity) versus outflow pH 5.92; metals decline markedly: Fe from 23.73 to 5.18 mg/L (−18.6 mg/L), Mn from 8.28 to 4.61 mg/L (−3.7 mg/L), and Al from 11.70 to 2.11 mg/L (~−9.6 mg/L). Total suspended solids changes are not statistically significant. Treatment effectiveness varies (pH increase at 25th percentile: +0.5; 75th percentile: +2.7).
• Kilometers protected: Method 1 estimates 1,601 km protected; Method 2 estimates 1,486 km protected. The preferred average is 1,543 km protected from AMD impairment (90% CI: 1,456–1,627 km).
• Cost-effectiveness: The present value lifetime cost implies an annualized cost of about $5,720 per kilometer protected (90% CI: $5,368–$6,168). Total streams needing protection are estimated at 10,381 km (1,543 currently protected by existing systems plus 8,838 still impaired). Protecting all streams for 25 years would cost about $1.5 billion (10,381 km × $5,720/km/year × 25 years).
• Funding gap: Pennsylvania’s non-AMD abandoned mine liabilities are about $3.9 billion, implying total need near $5.4 billion. The present value of available federal funds to Pennsylvania (AML plus IIJA plus set-aside) is about $2.5 billion. Thus, federal funds are sufficient to protect all AMD-impaired streams for 25 years but insufficient to cover other abandoned mine hazards; available funds amount to less than half of the total need.
• Comparative cost-effectiveness and benefits: Compared with Clean Water Act grants to municipal wastewater plants (about $1,000,000 per kilometer per year to make a river fishable), AMD systems appear highly cost-effective; equivalent cost-effectiveness would require treated rivers to be 175 times larger. Estimated benefits per kilometer of river made fishable ($5 million) far exceed the 25-year protection cost ($143,000 per km). For the Conemaugh River, households’ stated willingness to pay ($700 within 70 km) implies benefits of ~$(39) million versus ~$(16) million in 25-year costs for 112 km.

The findings directly address the study’s questions. First, AMD cleanup funding is well targeted from an equity perspective: the most AMD-burdened communities are economically disadvantaged and more vulnerable to the energy transition. Remediation therefore functions as a compensatory policy that can help level disparities and potentially support local economies, complementing broader just transition goals. Second, ex-post evidence from Pennsylvania’s passive treatment systems indicates substantial downstream protection at comparatively low cost, suggesting that expanded investments under IIJA can deliver large water-quality gains cost-effectively. The analysis also frames remediation as a potential economic development strategy, with prospects to employ local workers (including former miners), improve environmental amenities, and expand land availability for alternative uses. Relative to other large-scale water-quality programs, AMD treatment compares favorably on cost-effectiveness, and benefit estimates imply benefits exceed costs by wide margins. The paper discusses heterogeneity and generalizability: passive systems’ performance varies due to design, construction, and maintenance; active systems may be more expensive but also more impactful, affecting cost-effectiveness ambiguously. Space constraints may raise costs in more mountainous regions. However, IIJA’s scale may enable economies of scale, learning-by-doing, and innovation. Policy changes (e.g., the Stream Act allowing up to 30% set-asides for long-term AMD treatment) could improve long-run performance and funding flexibility.

This study shows that AMD remediation in Pennsylvania both advances environmental justice—by targeting economically and transition-vulnerable communities—and delivers water-quality improvements at relatively low cost. Passive treatment systems measurably improve water chemistry and protect substantial downstream lengths, with ex-post cost-effectiveness suggesting that current federal appropriations can sustain protection of all AMD-impaired streams for 25 years, though not other mine hazards. Key contributions include (1) an equity-focused characterization of AMD exposure across communities, (2) an empirical, two-method estimate of downstream kilometers protected based on observed inflow/outflow data and hydrologic mixing, and (3) a transparent, ex-post cost per kilometer-year metric used to size statewide needs and compare with available funding. Future research should quantify remediation’s employment and local growth effects, systematically link design expectations to realized system performance, assess active versus passive system trade-offs, and examine how evolving inflow water quality and technological innovation affect long-run costs and benefits.

• The kilometers-protected estimates rely on average system inflow/outflow performance over time; effectiveness can vary with hydrologic conditions (wet/dry periods) and system aging.
• Hydrologic calculations assume conservative solute transport and do not model storage or geochemical attenuation; this may overstate downstream concentrations in some alluvial settings.
• Impairment determinations depend on thresholds (pH < 6; Fe > 1.5 mg/L; Mn > 2.0 mg/L; Al > 0.5 mg/L) and, in Method 2, on the completeness and timeliness of DEP assessments; outdated or missing assessments can bias estimates in either direction.
• The 80 km upstream window may omit rarer, longer-range effects, though expanding the radius would not reduce estimated cost-effectiveness.
• Cost estimates assume a 25-year system life and annual O&M/reconstruction at 4% of initial costs, discounted at 3% real; longer actual lifetimes or lower reconstruction costs would lower cost estimates. Initial costs for 13 systems were imputed, introducing additional uncertainty.
• The analysis centers on passive systems in Pennsylvania; generalizability to other regions or to active systems may differ due to terrain, system type, or design/maintenance practices.
• Funding availability estimates assume constant nominal IIJA disbursements over 15 years and AML disbursements declining with coal production, discounted at 5% nominal; deviations from these assumptions would alter the funding gap.