Case study · Satellite mission & orbit design

Designing a constellation to watch the world's water.

Orbit and constellation design for AquaScope — a four-satellite, multi-instrument mission for monitoring Earth's water — and a spaceborne GNSS-R simulation testing how much of the planet reflected GNSS signals could map from that constellation.

World map of specular GNSS reflection points over three days, blue LEO ground tracks woven across the globe with orange ocean reflection hits filling the gaps
Simulated specular GNSS-R reflection points over three days: blue LEO ground tracks with orange ocean reflection hits, sampling the globe between the tracks.
Mission
AquaScope (concept)
Work
Orbit design · GNSS-R simulation
Constellation
4 satellites · 1 plane
Programme
Satellite Mission Design seminar, TUM

One mission, many ways to measure water

AquaScope — Advanced Quantitative Understanding of Aquatic Systems via Comprehensive Observations and Precise Environmental sensing — is a concept mission for monitoring water in every form: oceans, lakes, ice, soil moisture and the atmosphere. It carries four complementary instruments: KaRIn (a Ka-band radar altimeter with SWOT heritage for ~2 cm water-height precision), additional SAR for extent and backscatter, LiDAR for terrain and vegetation, and GNSS for precise orbit determination and reflectometry.

Two questions fell to this study. First, what orbit and constellation let those instruments observe the same scene almost simultaneously, often enough to be useful? And second — could passive GNSS reflectometry, riding along on the same satellites, add a genuinely valuable water-mapping layer?

A single satellite sees a place rarely and one way at a time. The design problem is to see often, and to see the same water several ways at once.

Why a distributed constellation

Packing everything onto one spacecraft is heavy, costly and fragile — a single failure loses the mission. Spreading the instruments across four coordinated satellites is lighter, cheaper and more resilient, and lets each sensor fly an orbit that suits it. The catch is coordination: the satellites must cross a region close enough together in time that their measurements describe the same surface state, minimising temporal decorrelation between sensors.

Satellite 1 WATER HEIGHT
KaRIn + GNSS precise orbit determination — high-precision water-surface heights.
Satellite 2 TERRAIN · VEGETATION
LiDAR + GNSS POD — elevation mapping, vegetation structure and atmosphere.
Satellite 3 SOIL · FLOOD
GNSS-Reflectometry + SAR — soil moisture and flood extent.
Satellite 4 REDUNDANCY
KaRIn + GNSS — redundancy and improved coverage.

The orbit that ties it together

The constellation settled on a single orbital plane carrying four evenly-phased satellites (mean-anomaly spacing ΔM = 90°), at roughly 890 km altitude and 77.6° inclination for near-global reach. At that altitude one revolution takes about 103 minutes — 14–15 orbits a day — and the 90° phasing puts neighbouring satellites ~25–26 minutes apart along the same ground track, so a region is swept by the whole cluster within about an hour and a half. The target was a global revisit of seven days or better, tightening to one–three days over a regional focus area (the Iberian Peninsula).

Ground track of the AquaScope satellites woven densely across a world map between roughly 77 degrees north and south
Fig. 1 — Ground track of the constellation: the 77.6° inclination weaves dense coverage between ~77°N and 77°S, the basis for the revisit targets.
Three-dimensional plot of the satellite orbit in Earth-centred inertial coordinates
Fig. 2 — The orbit propagated in Earth-centred inertial (ECI) coordinates — the geometry behind the ground track above.

Does the orbit hold?

A propagator simulated the orbit over three days and out to thirty, tracking how the elements drift. The behaviour matched theory closely: the right ascension of the ascending node (RAAN) drifts about −1.36°/day under Earth's oblateness (the J2 effect), inclination stays essentially constant (≈0.0076° over three days), and the semi-major axis decays only a few centimetres a day from residual atmospheric drag. In short, the orbit is stable over the timescales that matter, and the drifts that exist are the predictable, manageable kind.

4sats · 1 plane
Constellation — evenly phased at ΔM = 90° for coordinated overpasses.
890km · 77.6°
Orbit — altitude and inclination chosen for near-global coverage.
≤7days
Global revisit — tightening to 1–3 days over the regional focus area.
−1.36°/day
RAAN drift — matches the J2 prediction; orbit otherwise stable.
~4cm/day
Altitude decay — small, drag-driven, as expected for medium LEO.
~1.8M / 72h
Reflections — cumulative GNSS-R hits across all LEO satellites in the sim.

Could reflected GNSS map the water too?

The second half of the study tested spaceborne GNSS Reflectometry — a passive technique that listens to GNSS signals after they bounce off the surface, reading roughness, soil moisture and water extent without any transmitter of its own. A simplified simulation modelled Earth as a sphere with a chessboard reflectivity pattern, placed 24 GNSS-like satellites in medium Earth orbit, and flew a LEO receiver beneath them, computing each specular reflection point by bisector approximation and logging only the valid, in-view ocean hits.

Illustration of spaceborne GNSS reflectometry: several GNSS satellites send signals that reflect off Earth up to a GNSS-R receiver satellite
Fig. 3 — The GNSS-R principle: signals from many GNSS satellites reflect off the surface and are caught by a single receiver — coverage for free, from constellations already in orbit.

Sweeping the parameters — one to thirty days, one or three GNSS systems — showed how quickly coverage fills in. With the full constellation and multiple GNSS systems, specular points blanket the oceans densely; the gaps left by one system are filled by the next. A density map over all four AquaScope satellites makes the pattern plain.

Global heat map of log specular-point density for all four AquaScope satellites, oceans saturated red with bright high-density bands across the tropics
Fig. 4 — Specular-point density (log scale) for all four AquaScope satellites with three GNSS systems over 30 days: the oceans are sampled densely and near-continuously.

What the study concluded

The orbit design delivers what AquaScope needs — four evenly-phased satellites giving frequent, near-simultaneous multi-sensor coverage with predictable, stable orbits. And reflectometry earns its place on board: passive, all-weather, day-and-night, and effectively free coverage that scales with every GNSS satellite in the sky. It is highly sensitive to surface roughness, soil moisture and inland water — exactly the variables that matter for floods, wetlands and coastal monitoring — and complements the mission's radar, LiDAR and altimetry rather than competing with them.

The simulation was packaged as a standalone, parameter-driven tool (run_gnss_sim) that prompts for its inputs at runtime — so coverage for a new orbit, duration or set of GNSS systems is one run away.

Where it goes next

  • 01

    Long-term coverage

    RAAN keeps drifting, so the coverage pattern shifts after a month or two — a 6–12 month analysis is needed to judge it.

  • 02

    Orbit-control strategies

    Drag compensation and RAAN management to hold the coordinated overpasses over the mission lifetime.

  • 03

    Expand the constellation

    More than four satellites where even higher temporal resolution is required.

  • 04

    Toward real DDMs

    Advance from geometric specular points to modelled delay-Doppler maps and measured reflectometry data.

Sources & data

This case study is a condensed summary of independent research. For full data sources, datasets and references, please get in touch.

Work with GWO

Designing a mission that has to measure?

If you are shaping an Earth-observation concept and need its orbit, coverage and sensing performance worked out before commitment, tell us what the mission must see and how often.

Global map of simulated specular reflection points from the initial GNSS-R model