A commercial-grade desktop application for geoscientists evaluating CO sequestration sites. Score heterogeneous evidence — well logs, seismic horizons, fault maps, caprock mineralogy, formation pressure — into a common [0, 1] suitability scale and combine into a composite storage favorability surface with quantified uncertainty and full audit trail.
Unlike petroleum exploration — which asks "did nature accumulate hydrocarbons?" — CCS asks the engineering-inverse: can we deliberately place a buoyant fluid underground and guarantee it stays there for millennia? This reversal changes which elements matter. Trap geometry still matters — but the penalty for failure is not a dry hole; it is an atmospheric emission liability. The PFA application decomposes CCS suitability into three independent risk surfaces and combines them into a composite storage favorability map.
In petroleum PFA, a false positive is a dry hole — an economic loss, not a regulatory one. In CCS, a false positive on containment is a CO₂ leakage event that invalidates the carbon credits, triggers remediation, and potentially damages freshwater aquifers. The scoring framework must reflect this asymmetry: containment elements carry non-negotiable minimum thresholds that no amount of injectivity or capacity can compensate for.
Confidence surfaces are computed independently for each element via kernel density estimation with Sheather–Jones bandwidth. Well-characterized formations show high confidence; frontier saline aquifers with sparse well control show low. The composite output carries both favorability and confidence — regulators see exactly where data supports the interpretation and where it does not.
Can CO₂ be injected at commercial rate (≥ 1 Mt/yr per well) without fracturing the caprock? The system scores formation permeability, pressure differential from hydrostatic, and the co2_injectivity_class from the Hydrogeology Library. Drill-stem test data and core permeability provide direct evidence; lithology polygons scored via co2_storage_class provide indirect proxy.
Will CO₂ remain permanently sequestered over millennia? This is the non-negotiable element. The system evaluates caprock thickness and mineralogy via co2_seal_class, structural integrity via containment_integrity_class, and the density of legacy well penetrations that breach the seal horizon — each abandoned wellbore is a potential leakage pathway.
Is there sufficient accessible pore volume for the project's injection target? The system computes effective storage capacity from formation thickness × porosity × area × efficiency factor. Long-term security is assessed via co2_reactivity_class (mineral trapping potential) and pressure_management_class from the Hydrogeology Library — formations that dissipate pressure buildup score higher.
CCS site characterization draws from petroleum legacy (well logs, 2D/3D seismic, core analysis) and new-purpose datasets (formation pressure surveys, caprock integrity testing, geomechanical models, aquifer pressure monitoring). The PFA importer handles each format, projects into a chosen CRS, and tracks provenance. Missing data is surfaced honestly — the confidence surface reveals gaps rather than hiding them.
The same three-stage pipeline as other PFA domains: score evidence at its native geometry, propagate through a spatial model, combine across elements. The CCS-specific insight is that some evidence scores in the reversed direction — legacy well density, fault proximity, and baseline seismicity are all monotonic-decreasing: more is worse.
Continuous data (porosity, thickness, pressure) → configurable scoring functions. Categorical polygons (lithology,
structure, hydrogeology) → auto-populated from the three CCS scoring contexts across two attribute libraries:
co2_storage_class + co2_seal_class + co2_reactivity_class from the
Lithology Library (81 entries) and containment_integrity_class + injection_safety_class
+ storage_volume_class from the Structure Library (64 entries). The Hydrogeology Library adds four
more CCS contexts. All mappings editable, all overrides tracked per-scenario.
Point evidence propagated via Gaussian decay, IDW, or kriging. Anisotropy first-class — CCS formations
frequently have directional permeability controlled by depositional or structural fabric. Polygon layers
center-point rasterized. Background value: agnostic 0.5 by default, but CCS projects
often justify pessimistic 0.0 for containment — "absence of evidence on seal integrity is
not evidence of good seal."
pessimistic (0.0). Unexplored areas should not be assumed to have adequate seal.
This inverts the geothermal convention and must be set deliberately.
Per-element: weighted geometric mean across proxies. Across elements: the weighted geometric mean
is the CCS default because it is non-compensatory — a site cannot trade containment for capacity — but less
punishing than pure multiplication when data are sparse. The minimum operator is a conservative
alternative for regulatory submissions where every element must independently exceed a threshold.
minimum-with-threshold: any cell where the Containment score falls
below a user-specified cutoff (e.g., 0.3) is forced to zero regardless of other element scores.
weighted geometric mean with Containment at ≥ 0.40 weight.
For regulatory submissions, switch to minimum-with-threshold (Ï„ = 0.3 on Containment).
Always run weight-perturbation sensitivity — if rank order flips with ±15% weight shift, the discrimination is too weak for site selection.
A Class VI permit application requires demonstration that characterization data adequately constrains the subsurface model. The PFA system generates a per-element confidence surface via KDE, then combines them into a composite confidence map. Low-confidence zones become targets for additional characterization wells — the system tells you where to invest in more data, not just where to inject.
The ADM Decatur project injected over one million tonnes of CO₂ into the Cambrian Mt. Simon Sandstone, sealed by the Eau Claire Shale, with comprehensive monitoring confirming containment. The Illinois Basin is the best-characterized onshore CCS target in North America: dense well control, 2D/3D seismic, core data, and a complete stratigraphic framework. The PFA application loads these evidence layers natively and can reproduce — and extend — the published suitability analysis.
CCS characterization data is often commercially sensitive or subject to pre-competitive agreements. The application is fully offline-capable with an embedded spatial database — no cloud upload, no third-party dependency. Project files can be version-controlled and submitted alongside a Class VI permit application as auditable records.
.pfa project. 317 attribute library entries across Lithology (81), Structure (64), Alteration (52), Hydrogeology (46), and GDE (74) tables — 49 scoring contexts across 4 domains.We are working with CCS operators, Class VI permit applicants, and DOE-funded research partnerships. If you are characterizing a saline aquifer, evaluating a depleted reservoir, or building a regional storage atlas — we should talk.