1. The observer and the observed

In Stanisław Lem’s Solaris, the ocean is studied for decades and turns out to be studying the researchers back. The act of observation is not neutral. The desktop’s baseline sensor will not provoke a sentient ocean, but it will still change what the desktop is: a platform that looks down, records, and sells what it sees.

Entry 230 decided the desktop carries both a baseline sensor and customer payload slots. This entry picks numbers for the baseline imager so the next trades — power, data rate, thermal, mechanical — have something concrete to hold onto.

2. What the baseline sensor is for

The baseline sensor exists to give the desktop a product from day one. It is not meant to compete with sub-meter reconnaissance constellations or with free Copernicus data. It is meant to produce useful, sellable imagery at a cost and complexity the desktop can absorb.

Target use cases:

  • Agriculture: field health, crop classification, irrigation status.
  • Forestry: cover change, fire scar mapping, pest damage.
  • Urban and infrastructure: change detection, construction tracking.
  • Disaster response: flood extent, fire perimeter, damage proxy.

These applications do not need 30 cm resolution. They need frequent coverage, reliable scheduling, and a price low enough to undercut bespoke tasking of high-resolution satellites.

3. Resolution

Ground sample distance: 3 to 5 metres.

At 3–5 m, a single pixel covers a field boundary, a building footprint, or a road segment. Multiple pixels resolve a farm plot or a warehouse roof. It is coarse enough to keep the optics, detector, and pointing budget small, and fine enough to distinguish economically interesting change.

Going below 2 m would roughly double the aperture and tighten the pointing stability requirement by the same factor. Going above 10 m would push the product into the commodity tier where free Sentinel-2 data already competes. The 3–5 m band is the smallest aperture that still commands a commercial price.

4. Swath

Swath width: 25 to 35 kilometres.

Swath trades directly against aperture and detector size. A wider swath needs a shorter focal length or a larger detector, which reduces achievable resolution for the same optics. A narrower swath gives better resolution but less area per pass and worse temporal revisit.

A 25–35 km swath at 3–5 m GSD implies a detector on the order of 5,000 to 10,000 pixels across. That is comfortably inside the range of modern CMOS detectors used for smallsat imagers. It also means a single pass covers an entire agricultural region or metropolitan area in one contiguous strip.

5. Revisit rate

Revisit: every 2 to 4 days at mid-latitudes, with off-nadir pointing.

A single imager in a sun-synchronous orbit at roughly 500 km altitude revisits the same latitude band on the order of once per 10–16 days if it looks straight down. With ±30° off-nadir pointing, that drops to roughly 2–4 days. With a modest constellation of two or three desktops in different orbital planes, daily revisit becomes possible.

The desktop does not need to be a daily imager on its own. It needs to be credible enough that customers will pay for a subscription, knowing that more desktops can thicken the constellation later. The 2–4 day figure is the threshold where agriculture and insurance customers start treating the data as operational rather than archival.

6. Spectral bands

Panchromatic plus four multispectral bands: blue, green, red, near-infrared.

A panchromatic channel at roughly 1–2 m effective resolution, derived by pan-sharpening the multispectral data, improves visual interpretability without requiring a separate high-resolution optical train. The near-infrared band enables normalized difference vegetation index products, which are the bread and butter of agronomic subscriptions.

A coastal/aerosol band or a short-wave infrared band would add value, but each extra band increases detector complexity, calibration burden, and data rate. The first baseline sensor should ship with the four-band set that has the largest paying audience and the smallest engineering risk.

7. Mass

Total sensor mass: 6 to 12 kilograms.

This includes the telescope, detector assembly, focal plane electronics, shutter or filter wheel, and the baffle. It excludes the payload computer and the gimbal or body-pointing mechanism, which are counted against the compute and platform budgets.

A 6–12 kg imager is consistent with smallsat Earth observation cameras that deliver 3–5 m resolution from a 10–20 kg satellite bus. It is large enough to use real glass and a decent detector, small enough that the desktop’s structural and thermal budgets do not have to be redesigned around it.

8. Data budget implication

A 7,000-pixel swath at 3 m GSD and 11 bits per pixel produces roughly 100–150 megabits per second at typical orbital groundspeed. Over a 10-minute imaging pass, that is 60–90 gigabits of raw data. With lossless or visually lossless compression, the downlink requirement lands in the 50–100 Mbps range, which is well within the X-band or Ka-band communications attachment discussed in entry 229.

The numbers are intentionally conservative. If the first sensor produces less data than expected, the desktop still has capacity to host customer payloads. If it produces more, the communications attachment can be scaled up as a known upgrade rather than a rescue mission.

What this changes

  • The desktop’s baseline sensor is a 3–5 m GSD optical imager with a 25–35 km swath and four multispectral bands.
  • Off-nadir pointing gives a 2–4 day revisit from a single desktop, with daily revisit possible once a small constellation is operating.
  • The sensor mass budget is 6–12 kg, leaving room for customer payloads on the same attachment grid.
  • The data rate is sized to fit within the X-band or Ka-band communications architecture already chosen.
  • The next entry can decide where the sensor lives mechanically: fixed nadir mount, steerable bracket, or deployable optical bench.