ISRU Site Selection Informed by Passive Cave Mapping

The Problem

A NIAC concept study evaluating lava tube ISRU at Mare Tranquillitatis ranked candidate skylight entry points using orbital imagery and LROC DEM data alone. The top-ranked site had a clean 100-meter pit rim, visible regolith shelving, and nearby LEND hydrogen enrichment. When the team commissioned a ground-based acoustic reality-check at a Mauna Loa analog tube with equivalent geometry, the passive survey revealed the "clean" candidate's interior collapsed into a choke point 40 meters past the skylight, cutting usable volume by 80%. The site ranking shifted entirely, and the team's Phase II proposal had to be reworked in three weeks.

This is the core risk for ISRU site selection from orbital data alone. NASA's ISRU overview explicitly states that ISRU reduces mission mass, cost, and risk, but those savings depend on selecting the right site. Lava tube sites promise thermal and radiation shielding advantages that make them strategically attractive, but the advantages only apply when the interior geometry actually supports hardware deployment. Orbital imagery sees the skylight; it does not see the tube.

The scoring frameworks exist but lack cave-scale geometry. AIAA ASCEND's ROI scoring methodology quantifies Artemis regions of interest against water-mine potential, and LPI's Lunar South Pole whitepaper documents criteria including PSR access and water ice distribution. Both treat lava tubes as coarse candidates because the data to score tube-interior geometry was not available at scoring time. NASA's LCROSS mission confirmed water ice in Cabeus PSR, validating the orbital signal, but the leap from orbital signal to tube-interior site remains a gap.

The same scoring methodology applies in adjacent domains. Our rille habitats work uses the same multi-layer overlay logic for habitat siting inside rille-associated tubes, and the two scorecards share enough underlying machinery that a candidate site evaluated for ISRU can be re-scored for habitat suitability with minimal additional data collection. This dual-use capability is decisive when a single tube target may be evaluated against multiple mission concept families.

The financial cost of late-stage site downgrades is substantial. NIAC Phase II proposals run on 18-24 month cycles with funding commitments measured in the millions; a Phase II that has to re-rank candidates after the proposal is locked typically loses a full funding cycle while the team rebuilds the scoring narrative. The Phase III gate is even less forgiving — concept teams that arrive at Phase III with a preferred site that fails interior-geometry validation often see the entire concept return to Phase I rather than proceed to flight implementation. Accurate interior-geometry data at the Phase I scoring stage avoids that failure mode at the gate where it costs least to fix. The same logic applies to Artemis architect working groups: a habitat siting decision made on orbital-only data can survive early review, but the moment ground-truth scouting reveals a mismatch, the entire architecture has to absorb the change downstream.

The Solution: Acoustic-Geometry ISRU Scoring

EchoQuilt's ISRU scorecard accepts passive acoustic geometry as an input layer alongside hydrogen abundance, thermal regime, illumination, slope, and mission-specific constraints. Each candidate site gets a quilt of its interior, and the scorecard produces quantitative metrics including usable volume, longest continuous pass-through length, average floor roughness, and thermal variance estimate derived from acoustic energy attenuation. Sites that look equivalent on orbital data quickly diverge once interior geometry enters the scoring.

The scorecard runs in two modes. In scouting mode, it consumes a partial quilt (maybe one EVA or rover excursion's worth of patches) and returns scores with explicit confidence bounds. In commit mode, it demands a quilt that covers the full intended deployment zone and returns scores suitable for Phase II/III mission decisions. Concept teams use scouting mode to rank candidates coarsely before committing scarce rover-time, then commit mode for the final selection. This two-stage approach matches the cadence of the NASA asteroid and lava tube ISRU prospecting concept, which treats prospecting as a prerequisite to extraction architecture design.

The overlay-stitching capability is what elevates the scorecard above a checklist. LEND-derived hydrogen maps sit on one layer; Diviner thermal maps on another; orbital slope estimates on a third; the passive acoustic quilt on the fourth. Each layer carries its own resolution and uncertainty. The scorecard stitches them at matched grid cells and flags discrepancies (for example, a thermally stable zone that acoustic data shows is geometrically unusable, or a hydrogen-rich zone with no acoustic confirmation of floor accessibility). Planners see the discrepancies as candidate downgrades rather than silent averages, which keeps scoring honest.

For lunar sites, the thermal case is especially strong. Diviner measurements of lunar pit interiors show interior temperatures near 17°C with less than 1°C diurnal variation, versus surface swings of hundreds of degrees. EchoQuilt's scorecard weights thermal stability heavily because a site that scores well on hydrogen but sits in the diurnal swing regime is a worse ISRU candidate than a slightly lower-hydrogen site inside a thermally stable tube. The acoustic quilt confirms which parts of the interior actually sit in the thermally stable regime, not just which parts are inside the pit rim.

ISRU scoring depends on PSR traverses because many high-value ISRU sites require PSR-capable routing, and the route confidence the traverse planner produces feeds directly into the ISRU scorecard's accessibility weight.

Advanced Tactics

Score candidate sites in tiers matched to mission decision gates. A Phase I concept study needs scoring resolution good enough to rank 10+ candidates against each other but can tolerate wide confidence bounds per candidate. A Phase II study needs tight bounds on the top 2-3 candidates. EchoQuilt's scorecard exposes scoring resolution as a configurable parameter, so concept teams can burn their limited acoustic data on the candidates that matter at their current gate rather than spreading it thin.

Treat the acoustic quilt as a negative-evidence instrument too. If a quilt patch reveals structural instability signatures, water ingress, or unexpected void geometry, the scorecard downgrades the candidate explicitly rather than just failing to upgrade it. This is how the Mare Tranquillitatis example at the top of this post would have been caught in scouting mode: the choke point shows up as a negative-evidence patch before it becomes a Phase II reproposal.

For federated concept studies with ESA or JAXA partners, publish your scorecard inputs alongside your outputs. The LPI whitepaper tradition shows how Artemis scoring is inherently collaborative, and teams that open their acoustic layers can cross-check each other's interpretations before concept review. EchoQuilt's scorecard exports are structured for exactly this kind of federated review and plug into standard PDS archival formats.

Use volatiles-context layers to disambiguate sites that score similarly on geometry alone. Lunar Volatiles Tracking products combine LEND, LRO LAMP, and Diviner data into a unified volatile-stability indicator that is more useful than any single layer in isolation. EchoQuilt's scorecard accepts this composite layer as a top-level input and weights it explicitly against tube-interior geometry. Two candidate sites with identical interior volume can score very differently when the volatile stability layer is added, and the resulting downselect captures information that orbital-only scoring would have missed entirely. This is especially valuable for ISRU concepts targeting water-ice extraction inside thermally stable cave segments, where the volatile signal is what drives the architecture decision in the first place.

A cross-domain analogue from terrestrial conservation extends the multi-layer overlay logic. Our access budgeting work shows how the same overlay logic governs terrestrial cave disturbance planning, where biologists weigh access requirements against disturbance budgets across multiple data layers. The conservation biology community has run this kind of multi-layer scoring for decades and EchoQuilt's planetary scorecard inherits several of their established weighting heuristics.

Calibrate scoring weights against historical concept review feedback. If past Phase II reviews have downgraded sites for thermal instability rather than for volume constraints, the scorecard weights should reflect that reviewer concern. EchoQuilt's weights are exposed as a configurable layer rather than baked into the algorithm, so concept teams can tune the scorecard to match the review committee they will face. This kind of explicit weight tuning makes the scorecard's outputs defensible during the actual review, because the team can demonstrate exactly how their site preferences derive from the inputs, rather than presenting a black-box ranking that reviewers have to take on faith.

Ready to Score Your ISRU Candidates?

NIAC concept teams and Artemis architects scoping lava tube ISRU need a scoring tool that treats interior geometry as a first-class input rather than a blank space in the orbital map. EchoQuilt's ISRU scorecard is built around that integration, with native LEND, Diviner, LROC, and LCROSS overlay importers and PDS-compatible exports. Each pilot ships with a Lunar Volatiles Tracking composite-layer importer that combines LEND, LRO LAMP, and Diviner data into a unified volatile-stability indicator, a two-mode scouting/commit configuration aligned to NIAC Phase I and Phase II decision gates, a candidate site ranking module pre-tuned to Mare Tranquillitatis, Marius Hills, and Hadley Rille geometries, and a negative-evidence patch class that explicitly downgrades sites with structural instability or water ingress signatures.

Pilot teams shape the scoring weight defaults and the composite volatile-layer integration that the 2027 reference release will adopt for Artemis architect working group review. Priority goes to NIAC concept teams targeting Phase II proposals in the 2026 cycle, Artemis III architect working groups scoping water-ice extraction sites within thermally stable cave segments, JAXA-collaborated lunar pit mission planners, and ESA PANGAEA-adjacent ISRU researchers integrating with NASA reference frameworks. Join the Waitlist for Planetary Analog Researchers to test the scorecard against your current candidate site list and help us prioritize which orbital layers to integrate next.