Idea 2 — Volatile prospecting planner
15 November 2026. One of six ideas; read 00-overview.md
first. Source numbers refer to its section 8. Compiled with AI (Claude);
not re-checked by a person. "Judgement" marks an assessment.
Product in one line
For a prospecting area, the tool returns a map of where water ice is most likely in the top metre, with uncertainty, and ranks candidate measurement and drill stops by expected information gain under the rover's limits.
Decision it supports
A rover science team decides where to drive and drill first, and how to change the plan after each measurement.
Why now
- Prospecting missions planned for the South Pole: VIPER, launch August 2027 and landing September 2027 [3]; PROMISE, launch target late 2028, drill to about 1 m, 18 months [1]; DISCO for ice in micro cold traps [2]; LUPEX with instruments from JAXA, ISRO, NASA and ESA [5]; MAGPIE in 2029 and LUVMI-M in 2028 (C findings 3 and 4).
- Japan funds four prospecting instruments and a method for resource exploration at the University of Tokyo and partners until 2030 [7]. ispace is a core partner in a lunar water-sensing satellite project [39].
- ESA's community asked for prospecting missions before polar infrastructure planning, because the ice resource is uncertain (C finding 6).
What exists
- A prospectivity model for the VIPER site at Mons Mouton. It combines terrain, thermal and PSR layers, assumes relations between ice and those layers, and is built to be adjusted in near real time during the mission [4].
- The VIPER team evaluates thousands of traverse plans with Monte Carlo simulation [3].
- A comparable tool outside NASA and USGS: not found (one search; Europe and Japan not searched in depth).
Data basis and checks
- Inputs: LOLA terrain; Diviner temperatures at 240 m; the Diviner ice-depth model on a mesh of about 500 m (B finding 4); PSR maps. Orbital hydrogen, ultraviolet and infrared ice indicators were not reviewed in B.
- Ground truth: in-situ ice measurements at a South Pole site were not reviewed; expect very few before VIPER (judgement).
- Data class: every prospectivity value is inferred. Label it so.
Type A slice (TRL 3)
| CF | Function | Check |
|---|---|---|
| CF1 | Prospectivity map for one site from open inputs | Reproduction of the published map [4]; a comparison with a model, not with measured data |
| CF2 | Uncertainty of the map from terrain and thermal input uncertainty | Stability of the ranking across the terrain clones |
| CF3 | Ranking of stops by expected information gain | Hindcast: build the map without one orbital data set, then test the ranking against that data set [TBD which] |
Out of the Type A slice: update during a mission, traverse optimisation, other resources, any site on demand.
Users outside ESA (candidates; none contacted)
- Japan: the University of Tokyo consortium, the LUPEX science team, ispace.
- Europe, non-ESA: ispace-Europe (MAGPIE prime), Space Applications Services (LUVMI-M), resource-focused research groups.
- United States: NASA and USGS have their own model; commercial rover teams are possible users.
Four tests (judgement)
- Unique: pass with thin evidence. One model exists in the United States; test 1 of the grilling brief treats it as context.
- Feasible: weak. Inputs are open; measured data to check the result are scarce.
- Needed: pass. More named missions than any other idea.
- Fits Type A: weak. Two of three checks compare models with models. The TRL 3 definition asks for a proof of concept checked by experiment.
Link to ESA programmes and path beyond TRL 3
MAGPIE, the ESA instrument on LUPEX, and the Argonaut payloads. Next step: test against VIPER data once NASA publishes them, expected after the 2027 landing.
Risks and unknowns
- Planetary geology and geostatistics experience is not shown in the CV on file [TBD].
- Chandrayaan-2 and ESA archive data cannot go into a commercial product without permission (B finding 5).
- VIPER's delivery depends on the first Blue Moon MK1 flight; a delay moves the only strong validation data.
- Route optimisation must stay out: ESA funds it elsewhere (A2 finding 3).