Cross-Analog Comparison: Hawaii, Iceland, and Lanzarote Lava Tubes
Problem
Hawaii, Iceland, and Lanzarote are not interchangeable analogs, even though every planetary mission architecture memo treats them as a loose set. The NASA BASALT program record describes a 4-year $4.2M PSTAR program at Mauna Ulu that picked Hawaii for a specific set of basaltic-volcanic reasons. The Wiley JGR Solid Earth paper on La Corona lava tube system, Lanzarote documents La Corona's 7.6 km length and up to 28 m width, which are different morphological targets than Hawaii's narrower thermal-erosion tubes. And the Astrobotic writeup on drone mapping the icy lava tube in Iceland describes Lofthellir as a water-ice-filled tube whose acoustic and thermal regime does not overlap Hawaii or Lanzarote.
The cross-analog comparison problem is not whether the mapping engine works at each site — that has been demonstrated enough times across each site individually. The problem is whether the same engine, with the same calibration, produces comparable quilts across sites. Without that comparability, planetary analog researchers cannot aggregate the findings into a single cross-site result, and the NASA Science Planetary Analog Explorer registry becomes just a catalog of one-off results rather than a portfolio of defensible priors.
A cross-niche parallel makes the comparability bar concrete. The karst-lava comparison work for hibernacula biologists tackles the same comparability problem at terrestrial scale: a single mapping engine has to produce comparable acoustic signatures across karst and lava-tube cave systems even though the underlying rock chemistry, thermal regime, and biological occupation differ substantially. Methods proven in that work for normalizing acoustic signatures across geologically distinct caves transfer directly into the cross-analog problem the planetary community faces, and the cross-pollination of methods has saved both communities time on calibration design.
Solution
EchoQuilt ran the same stitching engine and the same 1.8 W sensor head across Mauna Loa, Lofthellir, and La Corona during the 2025 analog season, and published a cross-site residual comparison. The headline result: the engine produced comparable quilts at all three sites without site-specific retuning, but the residual distributions differed in informative ways. Mauna Loa delivered a 4.2 cm median residual across 180 patches; Lofthellir came in at 5.8 cm across 134 patches, with a heavier tail driven by ice-surface specular reflections; La Corona hit 3.9 cm across 220 patches, helped by more benign thermal conditions and the larger cross-section allowing more patch geometry per station.
Each site taught the quilt something different. Mauna Loa's thermal-erosion tubes produced the most varied wall geometry per unit length, which stress-tested the stitching engine's patch-density scaling. The Nature Scientific Reports paper on secondary minerals near HI-SEAS Hawaii as Martian subsurface analogues documents the gypsum and coralloid mineralogies that drive the Mauna Loa secondary-acoustic texture, and those textures became the hardest calibration targets we had. La Corona's constructional-tube regularity let the quilt build up longer clean seams — sometimes 40 m of single-stitch path — which anchored the cross-site baseline. Lofthellir's ice surfaces forced the quilt to learn which reflections it should treat as "wall" and which as "ice roof," a problem we solved with a small ice-mask prior that waitlist partners receive as a standard part of the onboarding kit.
The campaign produced three secondary outcomes that have proven valuable beyond the cross-site comparison itself. First, the cross-site residuals became a usable benchmark for any new analog site that subsequently joins the EchoQuilt ecosystem; teams running first-time campaigns at Tenerife, Etna, or Snæfellsnes can compare their initial residuals against the established Mauna Loa-Lofthellir-La Corona baseline to assess whether their site is producing typical or atypical mapping difficulty. Second, the campaign identified two specific stitching engine bugs that only manifested under cross-site conditions, and the fixes for those bugs improved single-site performance at all three sites by small but measurable amounts. Third, the campaign established a multi-agency collaboration pattern that has since been extended to include cross-validation visits between European and US-led analog programs, which has tightened the international analog community's working relationships.
Three site-pairing patterns emerged from the campaign. First, the ESA piece on Lanzarote and ESA, united by space makes the case for PANGAEA's use of Lanzarote, and our La Corona quilts slotted cleanly into the PANGAEA reference frame without retrofitting. Second, the Mauna Loa quilts cross-referenced directly to our multi-sol tube surveys work, which was originally scoped as a duration test but ended up giving us the thermal-cycle priors we needed for the cross-site comparison. Third, Lofthellir's ice-surface patches required the most calibration adjustment of any site, and that adjustment cost was worth tracking because it revealed which assumptions in the engine were most site-fragile.
Advanced tactics
Three tactics extend the cross-analog comparison past the default three-site run. First, build a per-site noise-floor prior and feed it into the quilt's novelty detector. Lanzarote has a quieter ambient regime than Mauna Loa or Lofthellir, so the novelty detector set with Mauna Loa's threshold will over-report at La Corona. We now ship three noise-floor priors with the engine and recommend crews tune them against a 30-minute baseline recording on campaign day one.
Second, run the cross-site comparison at matched station density, not matched traverse length. Matched traverse length over-weights the larger La Corona system and under-represents the tighter Lofthellir geometry. Matched station density — patches per meter of axis — keeps the comparison fair and turned out to be the statistic reviewers asked for most.
Third, report the site-over-site transfer error explicitly. When you apply a Mauna Loa-calibrated quilt configuration to La Corona data, how much does the residual degrade? In our campaign, Mauna Loa-to-La Corona transfer degraded residuals by 1.1 cm — small enough to defend — while Mauna Loa-to-Lofthellir degraded by 3.4 cm, which is where the ice-mask prior earns its place.
Fourth, cross-stitch unexpected patches across sites and surface the matches. Lofthellir's ice-surface patches cross-stitched surprisingly well with patches from our Hadley analog campaign, revealing a shared reflection-signature that motivated a joint follow-up paper. EchoQuilt's cross-stitch tool runs over the entire patch library across sites and flags any matches that exceed an a priori similarity threshold; some of these matches are uninteresting (similar geometry happens by chance), but the meaningful ones often surface unexpected scientific connections that drive follow-up campaigns.
Fifth, capture site-specific operational metadata alongside scientific metadata. Each site has its own access constraints, weather windows, permit conditions, and crew logistics, and operational data feeds into the scientific data in subtle ways. A campaign that documents both kinds of metadata produces a richer dataset for cross-site comparison, because reviewers can attribute quilt-quality variation to operational factors rather than misinterpreting it as scientific signal. EchoQuilt's metadata schema supports both kinds of capture in a single record, and analog teams that have adopted the unified schema have produced cross-site papers that consistently survive peer review more cleanly than papers that capture metadata in fragmented form.
Sixth, schedule cross-site campaigns to overlap calendar months where possible. Even with site-specific calibration, a cross-site campaign that runs Mauna Loa in summer and Lofthellir in winter introduces a seasonal confound that can be hard to disentangle from site differences. Where logistics allow, scheduling all sites in similar Earth seasons reduces the seasonal confound and tightens the cross-site comparison.
Seventh, share the cross-site campaign's data publicly with sufficient lead time for downstream teams to use it before their own field windows. The analog community is small, and a campaign that releases data only after publication often releases it after the next analog season's field windows have already closed. EchoQuilt's cross-site campaign released its raw patch library within two months of campaign close, which let multiple downstream teams seed their own sol-zero calibration against the cross-site baseline before their next field season. This kind of release timing has become a community expectation that strengthens the cross-site dataset's collective value.
CTA
If you are planning a multi-site analog campaign, a NIAC study that spans Hawaii, Iceland, and Lanzarote, or an ESA PANGAEA-adjacent concept, EchoQuilt's cross-analog engine is already calibrated to all three sites. Each pilot ships with the three site-specific noise-floor priors (Mauna Loa, Lofthellir, La Corona), the ice-mask prior for Lofthellir's water-ice-filled tube regime, the matched-station-density comparison protocol that reviewers consistently ask for, the cross-site residual baseline (4.2 cm Mauna Loa, 5.8 cm Lofthellir, 3.9 cm La Corona) as a benchmark for first-time campaigns at Tenerife, Etna, or Snæfellsnes, the cross-stitch tool that flagged unexpected matches between Lofthellir ice-surface patches and Hadley analog patches, and a unified site-and-operational metadata schema that captures permit conditions and crew logistics alongside scientific metadata.
Pilot teams shape the cross-site site-pairing transfer-error reporting format and the public 2026 season La Corona patch library schema that the 2027 reference release will publish. Priority goes to NIAC PIs scoping multi-site analog concept proposals in the 2026 cycle, ESA PANGAEA campaign coordinators integrating Lanzarote with NASA Hawaiian sites, JAXA-collaborated analog program leads scoping cross-agency reference deployments, and joint NASA-ESA federated analog publication leads running cross-site residual reporting under PDS4 and ESA PSA-conformant archival standards. Join the Waitlist for Planetary Analog Researchers and we will share the three noise-floor priors, the ice-mask prior for Lofthellir, and the matched-station-density comparison protocol.