wells nobody wrote down

Problem statement

An "orphaned" oil or gas well has no solvent owner to plug it; an undocumented orphaned well is one for which the state has no record of location or ownership at all — typically wells drilled between the 1860s and the mid-20th century, before permitting. State regulators cannot plug what they cannot find, and the DOE program brief states the challenge plainly: location and ownership records "could be a complete unknown," the steel casing "could have been removed (WWII efforts)," surface equipment may be overgrown or removed with the wellbore still in place, and "modern structures could be concealing location." States report between 310,000 and 800,000 such wells; the practitioners who must find them — state oil-and-gas agencies, tribal environmental offices, and their contractors — are working with a discovery rate DOE describes as "<50 well/year (NM), ~500ish wells/week nationwide."

Why this matters

Undocumented wells leak methane and brine, can convey gas into basements and aquifers, and are invisible to the plugging programs now funded at scale: the Bipartisan Infrastructure Law put $4.7 billion into DOI/BLM/state/tribal plugging, which DOE frames as $2,350–$23,500 per well or 21,000–200,000 wells per year over a ten-year timeline — but a well can only enter that pipeline once it is documented. DOE's own program document concludes that its "Products/Framework need to upscale current decisioning by factor of 40–400× to fully meet expected needs." The mismatch is a professional-context problem: the money and the plugging crews exist; the bottleneck is the throughput of the people and methods that convert an undocumented well into a mapped, characterized one.

What’s been tried and why it hasn’t worked

The state of practice, per NETL, combines review and georeferencing of historic production documents; desktop studies of geology, anthropogenic features and water-well history; ground surveys (electromagnetic induction, ground-penetrating radar, resistivity, seismic refraction, excavation); and aerial magnetometry and methane measurement. Aeromagnetic surveys — flown by crewed helicopter and later by drone since 2014 over Hillman State Park, Pennsylvania (about 190 mostly undocumented wells on 3,600 acres) — work well where a steel casing remains, but they cannot see wells whose casing was pulled for scrap or (reportedly) the oldest wells cased in wood, and they are labor-intensive to design (grid spacing, height above ground, drone speed) and to interpret. Historic-map digitization and machine learning over "disparate datasets" are being scaled by the national labs, but records are incomplete and inconsistently georeferenced. Each method has been demonstrated; none has been integrated into a workflow that a state agency with a small staff can run over whole counties, which is why DOE's stated goal is "to integrate existing technologies to economically identify and characterize undocumented orphan wells and work with the states so they will adopt these technologies."

What would unlock progress

Progress hinges less on a new sensor than on integration and prioritization: fusing digitized historic maps, land-ownership and lease records, LiDAR microtopography (useful for early wood-cased wells), and magnetometry into a probability surface that tells a survey crew where to fly and where to dig, and a best-practice survey design that non-specialists can execute. Adjacent fields have solved the analogous problem — unexploded-ordnance clearance uses magnetometer/EMI surveys with statistically designed transects and prioritized digging, and archaeology fuses historic maps with LiDAR to target excavation. A second unlock is a cheap field-verification kit (smartphone magnetometer plus methane sniffer) so that a landowner or technician can confirm a candidate at low cost.

Entry points for student teams

A data team could build the probability-surface prototype for one Pennsylvania or New York county using state well databases, digitized historical topographic and oil-field maps, and public LiDAR, and evaluate it against wells found at a characterized demonstration site like Hillman State Park. A hardware team could design and field-test a drone or backpack magnetometer survey protocol optimized for detection-per-hour and quantify what fraction of wells with pulled casing it misses. A policy/operations team could model the state-agency workflow and show where the 40–400× throughput gap actually sits (survey, verification, or paperwork). Skills: GIS, geophysics, machine learning, remote sensing, public-records research.