Coverage and revisit

Coverage answers a geometric question: from a given point on Earth, is a satellite visible above the elevation mask? Revisit answers the follow-up: how long between opportunities?

Neither says anything about whether a usable signal arrives. That is the link budget, and the two questions have different answers more often than not.

How it is computed

A grid is laid over the Earth, the constellation is propagated, and at every time step each grid point is tested against each satellite.

Two details in that flow matter more than they look.

The elevation mask is part of the definition. A satellite one degree above the horizon is geometrically visible and practically useless: the slant range is at its longest, and the signal passes through the most atmosphere. Coverage computed at a zero-degree mask flatters every design.

Statistics are area-weighted. A grid at constant angular spacing packs far more points near the poles than at the equator, because meridians converge. An unweighted mean would let a sliver of polar area dominate the result. Weighting by $\cos(\text{latitude})$ corrects for that.

Reading the output

orbitforge coverage \
  --constellation demo.json \
  --duration-hours 6 --step-seconds 60 \
  --grid-deg 10

Coverage over 684 grid points (10 deg), 60 sats, 6.0 h at 60 s:
  mean coverage 70.2% (area-weighted), min 0.0%, fully covered 10.5%, any 68.4%
  max fold 3, max coverage gap 5760 s

Fold, and why it matters

Fold is the number of satellites simultaneously visible from a point. Fold of one is bare coverage. Higher fold buys three distinct things:

A design that needs continuous service needs fold of at least two, because fold of exactly one means every handover is a gap.

Gaps are the requirement

For most missions the binding constraint is not average coverage but the longest gap. A ground station that can wait 90 minutes and one that can wait 10 minutes are different systems, even at identical mean coverage.

The 5760-second gap above is roughly one orbital period at 550 km, which is diagnostic: covered points are typically waiting for the same satellite to come around again, rather than being handed to a different one. That is what a constellation too sparse for continuous service looks like.

Sensor footprints

Coverage depends on what the payload can actually see, not just on line of sight.

A conical field of view of half-angle $\theta$ at altitude $h$ subtends a ground radius that grows faster than $h$, because the Earth curves away beneath it. This is why footprint area is not linear in altitude.

Choosing grid spacing and time step

Both are a cost-versus-truth trade, and both fail in the same direction: too coarse, and you under-report gaps.

A quick convergence check: halve both and re-run. If the maximum gap grows, your earlier numbers were optimistic and you have not converged yet.

Next steps

• Link budgets for whether the covered passes can carry data.
• Your first analysis for a worked example of coverage and link disagreeing.