Spatial vs Temporal Resolution: The Fundamental Trade-Off in Satellite Imaging

A project manager once asked me for daily 1-meter imagery of a construction site. When I quoted the cost and explained that no single satellite could deliver that, his response was: "But don't satellites fly over every day?"

They do. Just not the same satellite, at the same resolution, over the same spot, every single day. That conversation was my first lesson in explaining the resolution trade-off to non-specialists.

The Physics Behind the Trade-Off

Imagine you're taking photos from an airplane. If you use a telephoto lens (high spatial resolution), you capture fine detail — but only a small area. To cover a larger region, you'd need more passes. If you switch to a wide-angle lens (lower spatial resolution), you capture a wider area per frame but lose the fine detail.

Satellites face the same constraint. A sensor designed to resolve 30-centimeter objects has a narrow field of view — perhaps a 15 km swath width. That narrow strip means the satellite needs many orbits to cover the same area that a wide-swath sensor captures in one pass.

Sentinel-2, with its 10-meter resolution, has a 290 km swath width. That wide coverage, combined with two satellites, achieves 5-day revisit at the equator. A commercial satellite at 30 cm resolution might only cover a 10-15 km strip, requiring deliberate tasking (pointing the satellite) to image a specific location.

Resolution Spectrum in Practice

Satellite/Sensor	Spatial Resolution	Temporal Resolution	Swath Width
MODIS (Terra/Aqua)	250 m – 1 km	Daily (global)	2,330 km
Landsat 8/9	30 m (15 m pan)	16 days (8 with both)	185 km
Sentinel-2A/B	10 m	5 days (both satellites)	290 km
Sentinel-1A/C	5×20 m (IW mode)	6 days (both satellites)	250 km
SPOT 6/7	1.5 m (0.5 m pan)	Daily (tasked)	60 km
Planet SkySat	50 cm	On demand	8 km
Maxar WorldView	30 cm	1–3 days (tasked)	13 km

The pattern is clear: as spatial resolution improves, temporal frequency drops and swath narrows.

When Spatial Resolution Wins

Some questions demand detail:

Infrastructure monitoring: Counting vehicles in a parking lot, measuring building footprints, detecting small-scale construction changes — you need sub-meter imagery. A 10-meter pixel would blend the parking lot with its surroundings.

Urban mapping: Distinguishing individual buildings requires 1–2 meter resolution. At Sentinel-2's 10 meters, a city block becomes a handful of mixed pixels.

Precision agriculture (field-scale): Identifying variation within a single field — which rows are stressed, where drainage is poor — benefits from 1–5 meter resolution. At 30 meters, those within-field patterns disappear.

Legal evidence: Satellite imagery used in court proceedings or insurance claims often needs the highest available resolution to be unambiguous.

When Temporal Resolution Wins

Other questions demand frequency:

Crop phenology: Tracking growth stages — germination, flowering, senescence — requires observations every few days throughout the growing season. Missing the flowering window by a week means missing the data entirely. MODIS or Sentinel-2 time series work well here.

Flood dynamics: Floodwaters can rise and recede within days. Daily or near-daily coverage captures the full extent and recession. Waiting 16 days between Landsat passes means potentially missing the peak entirely.

Fire monitoring: Active fire detection needs multiple observations per day. VIIRS on Suomi NPP and NOAA-20 provides this, albeit at 375-meter resolution. You can't resolve individual trees, but you can detect the fire.

Atmospheric events: Weather systems, smoke plumes, dust storms — these phenomena evolve hourly. Geostationary satellites provide 10–15 minute imaging at kilometer-scale resolution.

Breaking the Trade-Off: Constellations

The traditional trade-off between spatial and temporal resolution assumed a single satellite. Constellations change the equation by distributing coverage across many platforms.

Planet's Dove constellation (200+ satellites) achieves daily global coverage at 3-meter resolution. Each individual satellite isn't remarkable — it's the sheer number that makes daily revisit possible.

Capella Space is building a SAR constellation aiming for hourly revisit at sub-meter resolution.

The cost of constellation data varies significantly. Planet's daily imagery requires a subscription; it's not freely available like Sentinel or Landsat data.

The Practical Choice

When planning a project, I ask three questions:

1. What's the smallest feature I need to detect? This sets the minimum spatial resolution. If you need to identify individual trees, you need sub-5m. If you need field-level patterns, 10–30m is fine.

2. How fast does the phenomenon change? Deforestation is detectable with monthly imagery. Flood extent changes daily. Crop stress evolves weekly. Match your temporal needs to the available data.

3. What's the budget? Free data (Sentinel, Landsat, MODIS) covers most research and monitoring needs. Commercial high-resolution data is necessary for specific applications but costs money.

Most of the time, freely available Sentinel-2 (10m, 5-day) data hits the sweet spot. It's detailed enough for field-level vegetation analysis and frequent enough for seasonal monitoring. When it isn't sufficient, that's when you start considering the higher tiers — and their associated costs.

Quick Reference: Choosing the Right Resolution

For the most common Earth observation use cases, here's a practical decision guide:

Use Case	Minimum Spatial	Minimum Temporal	Best Free Option	Notes
Regional crop monitoring (field blocks)	10–30 m	5–10 days	Sentinel-2	NDVI time series
Individual field precision management	5–10 m	5–7 days	Sentinel-2 (10 m)	Within-field variation needs ≤10 m
Flood extent mapping	10–20 m	1–3 days	Sentinel-1 (SAR)	Clouds defeated by radar
Wildfire active detection	375 m	Twice daily	VIIRS (FIRMS)	Need frequency, not detail
Wildfire burn scar mapping	10–30 m	Weekly	Sentinel-2	After fire season
Urban construction monitoring	1–5 m	2–4 weeks	None free; Planet/Maxar	Sub-meter for small features
Deforestation detection	10–30 m	Monthly	Sentinel-2 / Sentinel-1	SAR for cloud-affected tropics
Volcano deformation	N/A (InSAR phase)	6–12 days	Sentinel-1	mm-level precision
Coastal change / shoreline	3–10 m	Monthly	Sentinel-2	Annual shoreline comparison
Large-scale land cover mapping	10–30 m	Annual composite	Sentinel-2 / Landsat	Seasonal median composite

The key insight: most environmental monitoring doesn't require sub-meter resolution. The common mistake is assuming that higher resolution is always better. A 10-meter Sentinel-2 pixel covers 100 m² — more than sufficient to detect stress, change, or classification differences at the field or ecosystem level. You only need sub-meter when the target feature is itself smaller than a few meters.

Combining Resolutions: Data Fusion

An increasingly common approach is to combine data from multiple sensors:

• Use Sentinel-2 for its spectral richness and consistent temporal coverage
• Add commercial imagery for occasional high-resolution snapshots at critical moments
• Incorporate MODIS/VIIRS daily data to fill temporal gaps

This fusion approach gives you the best of both worlds — temporal continuity from frequent, moderate-resolution sensors and spatial detail from occasional high-resolution acquisitions. It requires more sophisticated processing but delivers analysis that no single sensor could provide alone.

The resolution trade-off isn't a problem to solve — it's a design constraint to work within. Understanding it helps you choose the right data for the right question, rather than always reaching for the highest resolution available.