Sentinel-2: The Complete Guide to Europe's Free Multispectral Satellite

If someone asked me to recommend a single satellite dataset for learning remote sensing, I'd say Sentinel-2 without hesitation. Free, global, 10-meter resolution, 13 spectral bands, 5-day revisit. Five years ago, this combination would have been a premium commercial product. Today it's available to anyone with an internet connection.

Mission Overview

Sentinel-2 is part of the European Union's Copernicus Earth observation program, managed by the European Space Agency (ESA). The mission consists of two identical satellites:

• Sentinel-2A: Launched June 23, 2015
• Sentinel-2B: Launched March 7, 2017

Flying in the same sun-synchronous orbit but offset by 180°, the pair achieves a 5-day revisit cycle at the equator (less at higher latitudes — as short as 2–3 days at 45°N).

Each satellite carries the MultiSpectral Instrument (MSI), a pushbroom sensor that images a 290-kilometer-wide swath. The equatorial crossing time is approximately 10:30 AM local solar time (descending node), chosen to balance solar illumination with minimal cloud cover.

The 13 Bands

What makes Sentinel-2 particularly useful is its band configuration. Unlike Landsat, which was designed primarily for land applications, Sentinel-2's bands were specifically chosen to serve agriculture, forestry, and environmental monitoring:

Band	Central λ (nm)	Bandwidth (nm)	Resolution	Primary Use
B1	443	20	60 m	Coastal aerosol
B2	490	65	10 m	Blue
B3	560	35	10 m	Green
B4	665	30	10 m	Red
B5	705	15	20 m	Red edge 1
B6	740	15	20 m	Red edge 2
B7	783	20	20 m	Red edge 3
B8	842	115	10 m	NIR
B8A	865	20	20 m	NIR narrow
B9	945	20	60 m	Water vapor
B10	1375	30	60 m	Cirrus detection
B11	1610	90	20 m	SWIR 1
B12	2190	180	20 m	SWIR 2

The Four 10-Meter Bands

Bands 2, 3, 4, and 8 (Blue, Green, Red, NIR) are the workhorses. At 10-meter resolution, they're detailed enough to distinguish individual agricultural fields, detect small water bodies, and map urban features at the block level.

The combination of Red (B4) and NIR (B8) gives you NDVI. Add Green (B3), and you can compute several additional vegetation indices. These four bands alone cover 80% of typical analysis needs.

The Red Edge Advantage

Bands 5, 6, and 7 — the "red edge" bands at 20-meter resolution — are where Sentinel-2 truly differentiates itself. The red edge region (700–780 nm) is where vegetation reflectance transitions sharply from low (chlorophyll absorption) to high (leaf structure reflection).

These three narrow bands sample this transition zone, enabling:

• Early detection of vegetation stress before visible symptoms appear
• Better discrimination between crop types
• More accurate chlorophyll and nitrogen content estimation

Landsat doesn't have red edge bands. Neither do most commercial satellites. This makes Sentinel-2 uniquely valuable for precision agriculture.

SWIR Bands

Bands 11 and 12 at 20 meters provide shortwave infrared information critical for:

• Moisture detection: Wet surfaces absorb SWIR, so moisture differences are clearly visible
• Burn scar mapping: The Normalized Burn Ratio uses NIR and SWIR
• Snow/ice discrimination: Snow is bright in visible but dark in SWIR
• Mineral identification: Clay and iron oxide have diagnostic SWIR absorption features
• Cloud/snow separation: Both appear bright in visible, but clouds are bright in SWIR while snow is dark

Processing Levels

Sentinel-2 data comes in three processing levels:

Level-0: Raw instrument data. Not distributed to users.

Level-1C: Top-of-atmosphere (TOA) reflectance. Radiometrically calibrated and geometrically corrected, orthorectified to a global reference grid. This is what the sensor measured after accounting for solar geometry.

Level-2A: Bottom-of-atmosphere (BOA) surface reflectance. Atmospherically corrected using ESA's Sen2Cor processor. Also includes a Scene Classification Map (SCL) that identifies clouds, cloud shadows, vegetation, water, and other surface types.

Use Level-2A for any quantitative analysis. Level-1C is appropriate only when you need the raw measurement or when Level-2A isn't available.

The Scene Classification Map

The SCL band in Level-2A products is underappreciated. It provides a per-pixel classification:

Value	Class
0	No data
1	Saturated or defective
2	Dark area pixels
3	Cloud shadows
4	Vegetation
5	Bare soils
6	Water
7	Cloud low probability
8	Cloud medium probability
9	Cloud high probability
10	Thin cirrus
11	Snow/ice

This classification enables automated cloud masking without running your own algorithms. Filter pixels with SCL values 3, 8, 9, or 10 to remove clouds and shadows.

Data Access

Sentinel-2 data is freely available through multiple channels:

• Copernicus Data Space Ecosystem (dataspace.copernicus.eu): ESA's official portal, full archive
• AWS Earth Search: STAC-compatible access through Element 84, widely used in cloud-native workflows
• Google Earth Engine: Pre-processed collections ready for analysis
• Microsoft Planetary Computer: STAC catalog with integrated processing

The data is open access under the Copernicus license — free for all purposes including commercial use, with the attribution requirement: "Contains modified Copernicus Sentinel data [YEAR]."

Practical Applications

Vegetation Monitoring

NDVI time series from Sentinel-2 track crop growth, detect stress, and estimate yield. The 5-day revisit captures critical phenological stages. The red edge bands add sensitivity that standard NDVI misses.

Water Bodies

NIR absorption by water makes it straightforward to map lakes, rivers, and coastal features. The Modified Normalized Difference Water Index (MNDWI) using Green and SWIR bands provides even cleaner water body delineation.

Land Cover Mapping

The 13-band spectral richness supports detailed land cover classification. Combining visible, NIR, red edge, and SWIR bands, supervised classifiers can reliably distinguish 10–15 land cover classes.

Disaster Response

Post-disaster imagery for flood extent mapping (water vs. land), wildfire damage assessment (NBR), and earthquake building damage (texture changes). The 5-day revisit ensures timely acquisition.

Urban Analysis

At 10 meters, individual city blocks are visible. NDBI (Normalized Difference Built-up Index) using SWIR and NIR distinguishes built-up areas from vegetation. Time series reveal urban expansion patterns.

Limitations

No satellite is perfect:

• Clouds: Optical sensors can't see through clouds. In tropical regions, cloud-free composites may require 2–3 months of data
• Night: No nighttime observations (the sun needs to be up)
• 20/60m bands: Not all bands are at 10 meters. Red edge and SWIR are 20 m; atmospheric bands are 60 m
• Temporal gaps: Despite 5-day revisit, cloud cover can create weeks-long gaps in usable data for specific locations
• No thermal band: Cannot measure surface temperature (use Landsat or MODIS for thermal)

Sentinel-2 vs Landsat

Both are freely available, but they differ in important ways:

Feature	Sentinel-2	Landsat 8/9
Spatial resolution	10/20/60 m	15/30/100 m
Spectral bands	13	11
Red edge bands	Yes (3 bands)	No
Revisit	5 days	8 days (both)
Swath width	290 km	185 km
Thermal band	No	Yes
Archive start	2015	1972 (Landsat series)

For current monitoring with high temporal frequency, Sentinel-2 is generally the better choice. For long-term historical analysis or when thermal information is needed, Landsat is irreplaceable.

In practice, many projects combine both — using Landsat's longer archive for historical context and Sentinel-2's better resolution and revisit for current monitoring.

Sentinel-2 has democratized satellite remote sensing more than any other single mission. A decade ago, 10-meter multispectral data with 5-day revisit was a premium product. Today, it's the baseline that everyone starts from.