Satellite Fundamentals

Every satellite image you analyze was captured by a machine in a specific orbit, at a specific altitude, with a specific sensor. Understanding these parameters tells you what the image CAN show and what it CANNOT — which is half the battle in intelligence analysis.

Key References

Bate, Mueller, White — Fundamentals of Astrodynamics (Dover, 1971) — The classic text on orbital mechanics

Hoots & Roehrich — Models for Propagation of NORAD Element Sets (1980) — SGP4/SDP4 propagator algorithm

Schowengerdt — Remote Sensing: Models and Methods for Image Processing (3rd ed., 2007) — Comprehensive sensor and image processing reference

ESA Copernicus Documentation: https://documentation.dataspace.copernicus.eu/

Orbital Mechanics for Analysts

You don’t need to solve Kepler’s equations. You need to understand what orbit parameters mean for collection.

Altitude Determines Resolution and Coverage

A satellite at lower altitude sees a smaller area but in higher detail. This is optics: angular resolution = wavelength / aperture diameter. From lower altitude, the same angular resolution covers less ground = smaller GSD.

Orbit	Altitude	Examples	Characteristics
LEO	300-2000 km	Sentinel-2 (786km), ISS (408km)	High resolution, limited swath, short orbital period (90-120 min)
MEO	2000-35786 km	GPS (20200km), Galileo	Navigation, not imaging
GEO	35786 km	GOES, Meteosat	Fixed position over equator, hemisphere coverage, poor resolution (1-4 km)
HEO	Varies	Molniya, Tundra	Loitering over high latitudes, used by Russia for comms

Key insight: GEO satellites cannot do high-resolution imaging because they’re too far away. At 36000 km altitude, achieving 1m GSD would require a mirror ~45m in diameter. GEO is for weather (km-scale) and communications, not reconnaissance.

Sun-Synchronous Orbits (SSO)

Most Earth observation satellites use sun-synchronous orbits. The orbit plane precesses (rotates) at exactly the rate Earth orbits the Sun, so the satellite crosses the equator at the same local solar time every orbit.

Why this matters for imagery:

Consistent illumination conditions — every pass has the same sun angle
Shadows are consistent between revisits, making change detection easier
Optical sensors typically cross at 10:00-10:30 local time (low enough sun for shadows to reveal 3D structure, high enough for good illumination)

The mechanism: Earth is not a perfect sphere — it bulges at the equator. This bulge causes the orbital plane to precess. By choosing the right inclination (~97-98 degrees for typical EO altitudes), the precession rate matches Earth’s orbital rate around the Sun.

Revisit Time

How often a satellite can image the same location. Depends on:

Orbital period — ~90-100 minutes for LEO
Swath width — wider swath = faster coverage of all longitudes
Constellation size — more satellites = more frequent revisits
Latitude — orbits converge at poles, so high latitudes get MORE frequent coverage

System	Satellites	Swath	Revisit (equator)	Revisit (60N, Estonia)
Sentinel-2	2	290 km	5 days	3-4 days
Landsat 8+9	2	185 km	8 days	5-6 days
Planet Dove	~200	~25 km	1 day	1 day
MAXAR WV-3	1	13 km	<1 day (off-nadir)	Variable (tasked)
Sentinel-1	2	250 km	6 days	3-4 days

Tasked vs. systematic: Sentinel-2 images systematically — same orbits, same timing. Commercial satellites can be TASKED to point off-nadir at a specific target, which gives more frequent revisit but at worse geometry (distortion from viewing angle).

Orbital Geometry

Nadir — the point directly below the satellite. Imagery at nadir has the best geometry (least distortion).

Off-nadir angle — the angle between nadir and the actual pointing direction. Commercial satellites regularly point 20-45 degrees off-nadir to image targets not directly below. Consequences:

Longer atmospheric path = more haze/scatter
Geometric distortion = buildings lean, measurements less accurate
Different shadow geometry

Swath width — the strip on the ground that the sensor covers in one pass. Wider swath = faster global coverage, but usually lower resolution (larger footprint per detector element).

The Four Resolutions

Every satellite sensor is characterized by four resolution dimensions. See The GEOINT Mind Map for the strategic view.

Spatial Resolution (GSD)

Ground Sample Distance: the area on the ground represented by one pixel.

GSD = (pixel_size * altitude) / focal_length

GSD	What You Can See	What You Cannot See
30m (Landsat)	Cities, large forests, lakes	Individual buildings
10m (Sentinel-2)	Building blocks, large ships, fields	Cars, small buildings
3m (Planet)	Individual large buildings, roads	Vehicles
1m (Commercial)	Vehicles, aircraft, small structures	Vehicle type
0.3m (MAXAR)	Vehicle type, aircraft type, people (as dots)	Faces, license plates

Spectral Resolution

The number and width of wavelength bands the sensor measures.

Panchromatic: 1 wide band (usually 450-900nm), highest spatial resolution
Multispectral: 4-13 bands, standard for analysis (Multispectral Analysis)
Hyperspectral: 100-200+ narrow bands (~10nm wide), material identification
Ultraspectral: 1000+ bands (experimental)

Sentinel-2 spectral bands:

Band	Wavelength (nm)	Resolution	Purpose
B01	443 (Coastal)	60m	Aerosol
B02	490 (Blue)	10m	Water, atmosphere
B03	560 (Green)	10m	Vegetation peak reflectance
B04	665 (Red)	10m	Chlorophyll absorption
B05	705 (Red Edge 1)	20m	Vegetation stress
B06	740 (Red Edge 2)	20m	Vegetation stress
B07	783 (Red Edge 3)	20m	Vegetation stress
B08	842 (NIR)	10m	Vegetation structure
B8A	865 (NIR narrow)	20m	Water vapor, vegetation
B09	945 (Water vapor)	60m	Atmospheric correction
B11	1610 (SWIR 1)	20m	Moisture, minerals, snow
B12	2190 (SWIR 2)	20m	Moisture, minerals

Temporal Resolution

See revisit table above. Key distinction: revisit (how often the sensor CAN image a location) vs effective revisit (how often you get a USABLE cloud-free image).

In Estonia (60N), winter cloud cover means optical effective revisit can be 2-4 weeks. This is why SAR (cloud-penetrating) is essential for Northern European GEOINT.

Radiometric Resolution

Bit depth determines how many brightness levels the sensor can distinguish.

Bits	Levels	System
8	256	Standard cameras, older satellites
11	2048	SPOT
12	4096	Sentinel-2, Planet
14	16384	Landsat 8/9
16	65536	Some hyperspectral

Higher radiometric resolution lets you distinguish subtle differences — shadow vs dark surface, shallow vs deep water. Critical for quantitative analysis like atmospheric correction and surface reflectance retrieval.

Major Satellite Systems

Free Data (Use These)

Sentinel-2 (ESA Copernicus) — Your workhorse for optical analysis.

2 satellites (A+B), 786 km altitude, sun-synchronous
13 bands, 10/20/60m resolution
5-day revisit at equator, 2-3 days at high latitudes
Data: Copernicus Data Space (dataspace.copernicus.eu)

Sentinel-1 (ESA Copernicus) — Your workhorse for SAR.

C-band SAR (5.405 GHz), VV+VH polarization
IW mode: 5x20m resolution, 250 km swath
Works through clouds, day and night
Data: Copernicus Data Space

Landsat-9 (NASA/USGS) — Longest running EO program (since 1972!).

11 bands, 30m multispectral, 15m panchromatic
16-day revisit (8 days combined with Landsat-8)
50+ year archive — incredible for historical change analysis
Data: USGS EarthExplorer (earthexplorer.usgs.gov)

MODIS (NASA) — Wide-area monitoring.

36 bands, 250m-1km resolution
Daily global coverage (Terra and Aqua satellites)
Best for: fire detection, vegetation phenology, ocean color

Commercial (Know About Them)

MAXAR (WorldView/GeoEye) — Highest optical resolution commercially available.

30-50cm GSD, multispectral
Tasked collection, expensive ($15-35/km^2)
Used by national intelligence agencies worldwide

Planet (Dove/SkySat) — Daily global coverage.

Dove: 3m, 4-band, ~200 satellites, daily revisit
SkySat: 50cm, video capable
Commercial but research/education access available

Capella Space — Commercial SAR.

X-band SAR, 50cm resolution
On-demand collection, very high resolution SAR

Airbus (Pleiades Neo/SPOT) — European commercial.

Pleiades Neo: 30cm, 4-band + 6 NIR-SWIR bands
SPOT-7: 1.5m, 6m swath, daily revisit

Python: Orbital Calculations

Compute Revisit from Orbital Parameters

import numpy as np
 
def compute_revisit(altitude_km, swath_width_km, inclination_deg, n_satellites=1):
    """Estimate revisit time for a sun-synchronous satellite."""
    R_earth = 6371  # km
    mu = 398600.4418  # km^3/s^2
 
    r = R_earth + altitude_km
    period_s = 2 * np.pi * np.sqrt(r**3 / mu)
    period_min = period_s / 60
 
    # Orbits per day
    orbits_per_day = 86400 / period_s
 
    # Ground track separation at equator
    earth_circumference = 2 * np.pi * R_earth  # km
    # Earth rotates under satellite by one orbit spacing
    track_separation_km = earth_circumference / orbits_per_day
 
    # Number of adjacent strips to cover equator
    strips_needed = track_separation_km / swath_width_km
 
    # Revisit time (days)
    revisit_days = strips_needed / n_satellites
 
    return {
        "period_min": period_min,
        "orbits_per_day": orbits_per_day,
        "track_separation_km": track_separation_km,
        "revisit_days": revisit_days,
    }
 
# Sentinel-2
result = compute_revisit(786, 290, 98.6, n_satellites=2)
print("Sentinel-2:")
for k, v in result.items():
    print(f"  {k}: {v:.1f}")
 
# Landsat-9
result = compute_revisit(705, 185, 98.2, n_satellites=2)
print("\nLandsat 8+9:")
for k, v in result.items():
    print(f"  {k}: {v:.1f}")
 
# Planet (approximate)
result = compute_revisit(475, 25, 97.5, n_satellites=200)
print("\nPlanet Dove constellation:")
for k, v in result.items():
    print(f"  {k}: {v:.1f}")

Predict When a Satellite Passes Over Your Location

import numpy as np
from datetime import datetime, timedelta
 
def next_pass_estimate(lat, lon, altitude_km=786, swath_km=290,
                       revisit_days=5, last_known_pass=None):
    """
    Rough estimate of when a sun-synchronous satellite next passes over
    a given location. For precise predictions, use TLE-based propagators
    (e.g., sgp4 library + space-track.org TLEs).
    """
    R_earth = 6371
    r = R_earth + altitude_km
    mu = 398600.4418
    period_s = 2 * np.pi * np.sqrt(r**3 / mu)
    orbits_per_day = 86400 / period_s
 
    # Sun-synchronous satellites cross equator at ~10:30 local time
    # Local time offset from UTC
    utc_offset_hours = lon / 15  # rough approximation
 
    # Ascending/descending node times
    local_cross_time = 10.5  # hours, descending node (daytime)
    utc_cross_time = local_cross_time - utc_offset_hours
 
    # The satellite revisits with period = revisit_days
    if last_known_pass is None:
        last_known_pass = datetime(2024, 1, 1, int(utc_cross_time),
                                   int((utc_cross_time % 1) * 60))
 
    now = datetime.utcnow()
    dt = (now - last_known_pass).total_seconds() / 86400
    passes_since = dt / revisit_days
    next_pass_offset = (1 - (passes_since % 1)) * revisit_days
 
    next_pass = now + timedelta(days=next_pass_offset)
    # Set time to approximate crossing time
    next_pass = next_pass.replace(hour=int(utc_cross_time) % 24,
                                   minute=int((utc_cross_time % 1) * 60))
 
    return next_pass
 
# Tallinn, Estonia
tallinn_lat, tallinn_lon = 59.44, 24.75
next_s2 = next_pass_estimate(tallinn_lat, tallinn_lon)
print(f"Approximate next Sentinel-2 pass over Tallinn: {next_s2}")
print("NOTE: For real predictions, use TLE data from space-track.org with sgp4")

Visualize Orbit Ground Track

import numpy as np
import matplotlib.pyplot as plt
 
def plot_ground_track(altitude_km=786, inclination_deg=98.6, n_orbits=3):
    """Plot ground track of a sun-synchronous satellite."""
    R_earth = 6371
    mu = 398600.4418
    r = R_earth + altitude_km
    period = 2 * np.pi * np.sqrt(r**3 / mu)
    inc = np.radians(inclination_deg)
    earth_rot = 360 / 86164  # deg/s (sidereal)
 
    t = np.linspace(0, period * n_orbits, 10000)
    orbit_angle = 2 * np.pi * t / period
 
    lat = np.degrees(np.arcsin(np.sin(inc) * np.sin(orbit_angle)))
    lon_inertial = np.degrees(np.arctan2(
        np.cos(inc) * np.sin(orbit_angle), np.cos(orbit_angle)
    ))
    lon = (lon_inertial - earth_rot * t) % 360 - 180
 
    fig, ax = plt.subplots(figsize=(14, 7))
    scatter = ax.scatter(lon, lat, c=t/60, cmap="plasma", s=0.5)
    plt.colorbar(scatter, label="Time (minutes)", ax=ax)
 
    # Mark Tallinn
    ax.plot(24.75, 59.44, "r*", markersize=15, label="Tallinn")
 
    ax.set_xlabel("Longitude")
    ax.set_ylabel("Latitude")
    ax.set_title(f"Ground Track — {altitude_km}km SSO, {inclination_deg}° inc, "
                 f"{n_orbits} orbits")
    ax.set_xlim(-180, 180)
    ax.set_ylim(-90, 90)
    ax.grid(True, alpha=0.3)
    ax.legend()
    plt.tight_layout()
    plt.savefig("orbit_ground_track.png", dpi=150)
    plt.show()
 
plot_ground_track()

Exercises

Exercise 1: Which Satellite Can See Tallinn This Week?

Given that Sentinel-2A last passed over Tallinn on a Monday, determine:

When will the next Sentinel-2 pass occur?
If that pass is cloudy, when is the next opportunity?
Can Sentinel-1 fill the gap? What are the tradeoffs?

Exercise 2: Sentinel-2 vs Planet for Vehicle Detection

A military convoy is reportedly moving through a 50km stretch of road.

Can Sentinel-2 (10m GSD) detect vehicles? What could it detect about the convoy?
What resolution do you need to count vehicles?
Planet Dove (3m) — can it count vehicles? What about vehicle types?
What revisit would you need to track the convoy’s movement?

Exercise 3: Why GEO Can’t Do High-Res Imaging

Using the GSD formula: GSD = (pixel_size * altitude) / focal_length

A typical detector pixel is 10 micrometers. At GEO altitude (35786 km), what focal length would you need for 1m GSD?
What mirror diameter does this imply? (f/D ratio of ~10 for space telescopes)
Compare to the James Webb Space Telescope mirror (6.5m) — is this feasible?

Self-Test Questions

Why do sun-synchronous orbits cross the equator at the same local time?
What is the difference between revisit time and effective revisit time?
A satellite has 12-bit radiometric resolution. How many brightness levels can it distinguish?
Why does spatial resolution improve at lower altitude?
An image was taken at 30 degrees off-nadir. What problems should you expect?

GEOINT Notes

Explorer

Satellite Fundamentals

Satellite Fundamentals

Orbital Mechanics for Analysts

Altitude Determines Resolution and Coverage

Sun-Synchronous Orbits (SSO)

Revisit Time

Orbital Geometry

The Four Resolutions

Spatial Resolution (GSD)

Spectral Resolution

Temporal Resolution

Radiometric Resolution

Major Satellite Systems

Free Data (Use These)

Commercial (Know About Them)

Python: Orbital Calculations

Compute Revisit from Orbital Parameters

Predict When a Satellite Passes Over Your Location

Visualize Orbit Ground Track

Exercises

Exercise 1: Which Satellite Can See Tallinn This Week?

Exercise 2: Sentinel-2 vs Planet for Vehicle Detection

Exercise 3: Why GEO Can’t Do High-Res Imaging

Self-Test Questions

Graph View

Table of Contents

Backlinks