Satellite PPP vs CORS Subscriptions: Total Cost and Coverage

A sober comparison of the two ways to feed your RTK fleet — recurring network fees versus one-time self-converging hardware.

Two architectures for the same centimetre

Both roads lead to RTK-grade positioning; they differ in where the reference lives. The CORS model densifies the ground: networks of permanent stations (state CORS, Trimble VRS Now, Leica SmartNet, national infrastructure) compute corrections and sell them per rover over NTRIP. The satellite PPP model densifies the sky: global tracking networks compute precise orbits and clocks, broadcast them from geostationary satellites on L-Band, and any receiver in the beam converges to absolute coordinates on its own — Trimble CenterPoint RTX, CHCNAV PointSky, and the service behind UAV Mate's self-converging base are the commercial faces. The architectures are not enemies — many operations run both — but their cost curves, coverage maps, and failure modes are nearly mirror images, and buyers deserve the comparison plainly.

Coverage: interpolation versus illumination

CORS accuracy is a function of distance: superb near stations, degrading past 30–40 km baselines, absent beyond the network's edge — and networks follow population, not projects. Cross a border and both the stations and your subscription may stop. Satellite delivery inverts the geometry: continental beams illuminate a mine lease and a metropolis identically, ~99% availability, no interpolation because nothing is being interpolated — the corrections describe the satellites, which you can see from anywhere with sky. The one symmetric weakness: heavy canopy or canyon walls that block sky degrade both models' rovers, and additionally delay a PPP receiver's convergence (the placement craft in L-Band PPP explained).

Cost structure: rent per rover versus own the source

Network RTK prices per consumer, per year: every drone, tractor, and rover with a login is typically several hundred to a few thousand dollars annually, and the meter never stops. Satellite PPP in the base-station architecture prices per source: one hardware purchase plus one correction subscription at the base, feeding unlimited rovers over standard RTCM 3.x. The crossover arithmetic is short — a three-rover fleet usually breaks even inside the first year, and every added rover widens the gap. Two honest exceptions run the other way: a single-rover urban operation inside excellent coverage may find one subscription cheaper than any hardware; and organizations that already own network subscriptions for other reasons should count them as sunk. The worksheet lives in the buyer's guide.

Failure modes and who controls the fix

CORS-model outages are someone else's: network maintenance windows, cellular dead zones, caster congestion, subscription lapses in accounting. When they bite, every rover on the job stops, and the fix is a phone call. Satellite-model outages are mostly physics you can see: blocked sky at the base (move five metres), severe ionospheric storms (rare, shared by all GNSS), or a lapsed correction service (one renewal to track instead of many). After convergence, a PPP base holds its coordinates and keeps serving even through correction-stream gaps — the mission-critical minutes are insulated. Operations people phrase the difference simply: one model's reliability is a contract, the other's is equipment in your truck.

Accuracy and datum: a subtle but decisive difference

Within coverage, both models deliver centimetre solutions — parity in the brochure. The datum behaviour differs. Network corrections are usually expressed in the national plate-fixed datum: convenient domestically, seams at borders, and the frame is whatever the operator chose. Satellite PPP delivers ITRF current-epoch coordinates — globally uniform, epoch-tagged, transformable to any national datum with documented parameters (the frames guide). For multinational programs, mixed fleets, and any workflow where datasets from different years must overlay, the single global frame is quietly worth more than the headline accuracy: it makes repeatability a property of the architecture rather than a procedure.

A decision rule that survives contact with reality

Score your operation on three axes. Geography: what fraction of jobs sit inside strong network coverage? Fleet: how many correction consumers, today and in two years? Longitudinality: how often must new data overlay old? Mostly-covered, single-rover, one-off jobs — stay on the network, happily. Anything remote, multi-rover, or repeat-visit — own the source: a self-converging PPP base turns corrections from an operating expense and a coverage lottery into a piece of kit that works the same everywhere. Many mature operations land on the hybrid: network logins for the odd urban rover, the base for everything that earns the real money.

The hybrid pattern, concretely

Mature fleets rarely evangelize either model; they assign each to what it does cheapest. The pattern that recurs: a self-converging PPP base as the primary correction source — every repeat-visit site, every remote job, the whole machine fleet — plus one retained network login for the genuinely roaming rover (utility locates across a metro area, say) where per-rover pricing matches per-rover usage. The datum plan makes the mix safe: everything exports through documented transformations into one delivery frame, so a network-datum locate and an ITRF-frame as-built coexist in the client's GIS without seams (the transformation discipline). Costed honestly, the hybrid usually runs one subscription instead of five and touches the network exactly as often as the network earns it.

Questions to ask each vendor before committing

For a network operator: coverage map with station spacing at your actual sites (not the marketing shade of green); per-rover pricing at your fleet size in year three; roaming and cross-border terms; uptime history, not SLA prose. For a satellite-PPP base vendor: convergence spec with conditions (open sky, minutes, centimetres — and the frame); beam coverage for every country on the schedule; what happens through correction-stream gaps after convergence; and the transports out — radio power and licensing, NTRIP, RC link. Both lists fit on one page, and both convert the architecture argument of this article into numbers specific to your operation — which is where the decision belongs.

The migration path, if you switch

Moving from network-first to base-first is undramatic when sequenced. Month one: run both — the base anchors one pilot site while the fleet stays on subscriptions; compare checkpoint residuals and correction-age logs side by side. Month two: migrate the repeat-visit and remote work, where the base's advantages are structural; document the ITRF-to-delivery transformations per client. Month three: review which subscriptions still earn their line item (usually the one roaming rover) and retire the rest at renewal. The datum discipline is the only real work; the hardware change is an afternoon.

One-line takeaway

CORS rents you corrections per rover inside its coverage; satellite PPP sells you the source once for anywhere with sky — count your rovers, map your jobs against real coverage, and the architecture chooses itself.

Frequently asked questions

Is network RTK more accurate than satellite PPP?

Inside good coverage they are comparable (centimetre class). Beyond ~30–40 km baselines network accuracy degrades; satellite PPP is uniform wherever sky is open.

Can I keep my CORS subscription and add a PPP base?

Yes — hybrid is common: the base anchors remote and repeat-visit work; the subscription covers roaming urban rovers. Both speak RTCM to the same fleet.

What happens to my data if I switch models?

Nothing, if you manage frames: transform network-datum data and ITRF data into one documented frame. The switch is a datum exercise, not a re-survey.

Which model is better for a drone service business?

Businesses selling accuracy anywhere tend to own the source: no coverage caveats in proposals, no per-rover fees scaling with the fleet, identical workflow at every client site.