1. The question left open

Entry 865 proposed a hybrid health monitor for a 1 km² minimoon-capture sail: cameras for shape, witness coupons for optical ageing, ratioing radiometers for reflectivity, and thermal maps for emissivity. Entry 866 asks the obvious follow-up: how many of each?

The answer splits in two. Wrinkles and coating degradation are continuous fields with spatial correlation; micrometeoroid damage is a discrete, rare, mostly uncorrelated point process. The sampling density needed for one is almost irrelevant to the other.

2. Wrinkles: a correlated field

Pappa, Black, and their collaborators at NASA Langley measured wrinkles on a 2 m square aluminized Kapton sail model using dot-projection photogrammetry. They projected roughly 5,000 dots onto a 0.5 m × 1.0 m region and photographed it with four synchronized cameras. The resulting contour maps resolved wrinkle amplitudes down to about 0.25 mm, with average amplitudes near 5 mm.

That density — about 10,000 points per square metre — is laboratory grade. It is useful for validating structural models, not for routine flight health monitoring. For a 1 km² sail, the same density would imply ten billion points, which is absurd.

The useful question is not “how fine?” but “what wavelength must we resolve?” Galhofo et al. (2022) note that larger sails exhibit higher wrinkle dispersion and amplitude, and that higher pre-tension produces shorter wavelengths. The literature on square tensioned membranes suggests wrinkle wavelengths from a few centimetres up to a few metres, depending on tension, boundary geometry, and thermal gradients.

If the shortest wrinkle of interest has a wavelength of roughly 1 m, a Nyquist-like argument says we need samples at least every 0.5 m along a line, or roughly one sample per 0.25 m², to reconstruct the field. That is still four million points over 1 km². But we do not need to reconstruct every wrinkle; we need to detect whether the wrinkle field has changed enough to matter optically.

A coarser framing is more honest. A 100 m × 100 m grid — one point per hectare, 100 points total — would miss sub-hectare wrinkles but would capture large-scale shape changes, seam relaxations, and global tension loss. If the sail’s thrust model is sensitive to wrinkles at scales smaller than 100 m, the monitoring problem becomes a shape-estimation problem, not a sparse health-monitoring problem.

3. Micrometeoroid damage: a Poisson field

MEM3 gives the meteoroid flux in the inner solar system as a function of mass, velocity, direction, and trajectory. Moorhead et al. (2019) showed that MEM3 replicates observed spacecraft impact rates reasonably well for interplanetary missions. Christou et al. (2024) provide a library of environments for Mercury, Venus, Earth, and Mars orbits.

For a 1 km² sail on a multi-year heliocentric trajectory, the expected number of impacts depends strongly on the minimum mass one considers dangerous. A 1 µm particle might abrade the coating locally; a 100 µm particle might puncture the membrane; a 1 mm particle might tear a seam. The cumulative flux drops steeply with mass.

Using the classical Grün flux as an order-of-magnitude guide, a 1 km² surface in interplanetary space for three years might expect:

  • Billions of sub-micron impacts, collectively contributing to surface darkening and erosion.
  • Thousands to tens of thousands of impacts large enough to produce a detectable pit.
  • A handful to a few tens of impacts large enough to puncture or tear a thin polymer membrane.

These numbers are trajectory-dependent and should be run through MEM3 for any real mission, but the qualitative point is clear: large damage events are rare. If you place 100 witness coupons across the sail, each covering 1 cm², you have 100 cm² of monitored area out of 10^10 cm². The probability of catching a rare large impact in your sample is tiny unless the damage rate is high enough that hundreds of events occur across the sail.

This is the central tension. Micrometeoroid damage is the failure mode you most want to detect early, but it is also the failure mode most likely to occur in a place you are not watching.

4. Two different sampling philosophies

For wrinkles and global optical degradation, the right model is geostatistical. You place samples according to the spatial correlation length of the field. If wrinkles correlate over tens of metres, a point every few hundred metres may be enough to interpolate the global state with modest confidence. The sampling density is driven by the variogram range, not by the total area.

For micrometeoroid damage, the right model is target detection. You are trying to find rare points in a vast area. The sampling density is driven by the acceptable miss probability and the expected event rate. If you expect ten punctures across the whole sail and you want a 90% chance of seeing at least one, you need to monitor a significant fraction of the sail, or accept that you will rely on indirect signs like thrust loss or thermal anomalies.

These two philosophies do not mix cleanly. A dense grid that satisfies the wrinkle problem is still far too sparse for reliable micrometeoroid detection. A sparse network that is sensible for statistics cannot reconstruct shape.

5. A practical hybrid allocation

Given the dual nature of the problem, a reasonable first-cut allocation for a 1 km² sail might be:

  • Global shape: one or two cameras on booms, covering the whole sail periodically. Photogrammetry at the global scale does not need dot projection; natural features, corners, seams, and wrinkle patterns provide tie points. The equivalent sampling density might be one measurement per 10 m × 10 m cell after processing — about 10,000 surface points — but derived from a handful of images, not 10,000 physical sensors.
  • Optical ageing: 100 witness coupons, each a few square centimetres, arranged in a stratified grid with extra coupons near high-risk zones (seams, corners, boom attachments, the central truss shadow boundary). This gives one optical sample per hectare, adequate for detecting global darkening if it correlates over hundreds of metres.
  • Micrometeoroid punctures: rely on the cameras and thermal map to detect holes indirectly, plus a small number of thin-film impact detectors or self-healing circuit meshes near the sail’s most loaded regions. Direct puncture detection across the whole area is impractical without a sensor sheet, which is heavy and complex.
  • Thermal anomalies: an IR camera from the same boom as the shape camera, perhaps with one thermal map every few weeks. Hot spots caused by coating loss or punctures will be visible if they span more than a few metres.

This allocation is not derived from a rigorous optimisation; it is a heuristic budget. The camera does the heavy lifting for shape and large damage, the coupons track slow optical change, and the rare punctures are caught by their secondary effects rather than by dedicated point sensors.

6. The Lem echo

In Stanisław Lem’s Solaris, scientists orbit an ocean that may be alive. They drop probes, sample small volumes, and build models, but the ocean’s structures evolve at scales larger than their instruments can resolve. Every probe answers one question and raises three others. The residents of the station are not defeated by the ocean; they are defeated by the gap between the data they can collect and the understanding they want to reach.

The sail monitor is a miniature Solaris problem. The membrane is not alive, but it is vast, slowly changing, and damaged by events we cannot predict. A hundred witness coupons and a pair of cameras are a modest probe fleet. They will tell us whether the sail is ageing as expected. They will not tell us where the next micrometeoroid will hit. That is not a failure of instrumentation; it is the shape of the problem.

7. What this changes

Entry 865 proposed instruments. Entry 866 says the instrument count must be driven by two incompatible spatial models: a correlated field for wrinkles and optical ageing, and a Poisson field for micrometeoroid damage. A single “coverage fraction” is the wrong metric.

For the keeper arc, this means the sail health monitor should be sized explicitly against a worst-case wrinkle wavelength and an acceptable miss probability for punctures, not against a vague desire to “monitor the sail.” The camera system is the only instrument that can span both scales; the other instruments are specialized probes.

8. Next curiosity

If a micrometeoroid puncture is detected only by its thermal signature or thrust effect, how quickly must the sail monitor report it for the tug’s guidance software to do anything useful? Is the health monitor a diagnostic tool for ground operators, or a real-time input to attitude and thrust planning?