Sentinel-2 Bands at Level-1 and Level-2: Wavelengths, Uses, and Comparisons
2025-08-07 · 8 min read · Sentinel-2 · TOA · BOA · Landsat · Bands · Red edge · HLS · Sen2Cor · Band 10 · B8 vs B8A

TL;DR: Sentinel-2 MSI provides 13 bands from coastal blue to shortwave infrared at 10, 20, and 60 m. Level-1C is top-of-atmosphere (TOA) reflectance. Level-2A is bottom-of-atmosphere (BOA) surface reflectance. For most cross-date analytics use L2A; for time-critical tasks or scenes where correction may struggle (snow, heavy aerosols, very low sun), L1C can be more stable. Sentinel-2 stands out with three red-edge bands and a narrow NIR.
Level-1C vs Level-2A in one paragraph
Level-1C is radiometrically calibrated TOA reflectance in UTM tiles. Level-2A is BOA surface reflectance derived from the same L1C scenes. Choose L2A for consistent comparisons across dates and scenes. Choose L1C when you need the lowest latency or expect the atmospheric correction to struggle (e.g., polar winter, smoke, bright sand, very low sun).
BOA algorithms in practice
Sentinel-2 L2A uses Sen2Cor. Landsat 8/9 L2 uses LaSRC. For cross-sensor time series at 30 m, the Harmonized Landsat-Sentinel (HLS) products place both on a shared grid with harmonized correction.
Sentinel-2 bands at a glance
The table lists nominal (design) center wavelengths and native resolutions for Sentinel-2A/B (and S2C when applicable). The L2A column shows whether the band is present as surface reflectance in Level-2A. Actual bandpasses are defined by each instrument’s spectral response function (SRF — the curve showing how sensitive the band is across wavelengths); SRFs differ slightly between A/B/C but are cross-calibrated so most workflows can treat them interchangeably.
Band | Center (nm) | Native res | L2A? | What it is useful for |
---|---|---|---|---|
B1 Coastal aerosol | 443 | 60 m | Yes | Aerosol & coastal water context, haze over bright targets |
B2 Blue | 490 | 10 m | Yes | Bathymetry in clear water, true-color composites, turbidity |
B3 Green | 560 | 10 m | Yes | Vegetation vigor near the green peak, suspended matter |
B4 Red | 665 | 10 m | Yes | Chlorophyll absorption, indices with NIR |
B5 Red-edge 1 | 705 | 20 m | Yes | Early chlorophyll sensitivity, stress onsets |
B6 Red-edge 2 | 740 | 20 m | Yes | Chlorophyll content & canopy structure |
B7 Red-edge 3 | 783 | 20 m | Yes | Dense canopy discrimination, red-edge indices |
B8 NIR (wide) | 842 | 10 m | Yes | NDVI, EVI2, biomass proxies |
B8A NIR (narrow) | 865 | 20 m | Yes | BRDF (bidirectional reflectance distribution function) handling; fine canopy signals |
B9 Water vapour | 945 | 60 m | Yes | Water vapour estimation to aid correction |
B10 Cirrus | 1375 | 60 m | No | Thin cirrus detection and masking at TOA only |
B11 SWIR 1 | 1610 | 20 m | Yes | Moisture, burn severity, cloud screening |
B12 SWIR 2 | 2190 | 20 m | Yes | Mineralogy, residue, moisture gradients |
Note on S2A/B/C bandpasses
MSI units have slightly different SRFs (spectral response functions — per-band sensitivity curves), with nm-level center shifts and bandwidth differences. ESA publishes SRF curves per instrument and maintains cross-calibration across A/B/C. If you rely on narrow-band features (e.g., tight red-edge indices) or compare to lab/airborne spectra, convolve with the instrument-specific SRF for best accuracy.
B8 vs B8A: which NIR should you use?
- B8 (wide NIR, 842 nm @ 10 m) is the workhorse for NDVI/EVI2 at parcel scale. Use it for 10 m vegetation and biomass proxies, NIRv, and most browse composites (e.g., false-color with B8/B4/B3).
- B8A (narrow NIR, 865 nm @ 20 m) sits closer to Landsat’s NIR band center and is useful for BRDF-aware workflows, red-edge indices that prefer a narrower NIR, and certain aerosol/geometry handling tricks.
- Don’t mix B8 and B8A inside the same time series unless you re-tune thresholds and resample to a fixed grid. The ≈23 nm center shift and bandwidth difference can introduce small but real index offsets.
- Heads-up for Landsat users. Landsat OLI’s NIR (~865 nm) behaves more like Sentinel-2’s B8A than B8. If you port thresholds from Landsat to S2, validate with care or prefer S2 B8A for closer NIR alignment (accepting the 20 m native resolution).
Common band combinations & quick formulas
A few “no-surprises” recipes that behave well on Sentinel-2 (assume reflectance scaled to 0–1; if using scaled integers, divide by 10,000 first):
- True color (RGB): B4, B3, B2
- False color vegetation: B8, B4, B3
- False color SWIR (burn/moisture): B12, B8, B4
Indices
- NDVI: (B8 − B4) / (B8 + B4)
- EVI2 (S2-friendly, no Blue): 2.5 × (B8 − B4) / (B8 + 2.4 × B4 + 1)
- NDRE (red-edge vitality): (B8A − B5) / (B8A + B5) Swap B5→B6/B7 to probe deeper into dense canopy.
- NBR (burn severity & moisture): (B8 − B12) / (B8 + B12) B11 works too; B12 is classic.
- NDMI (vegetation moisture): (B8 − B11) / (B8 + B11)
Grid discipline
Pick a target resolution and grid once per project and stick to it. Resample bands once (e.g., bring 20 m to 10 m for per-pixel alignment) and avoid repeated reprojection to keep time-series consistent.
Where PlanetScope and EarthDaily class sensors fit
Commercial constellations often trade band count for revisit or resolution. Compare by band families rather than exact nanometers.
- PlanetScope SuperDove class typically offers ~8 SR bands (coastal blue, blue, two greens, yellow, red, one red-edge near 705 nm, one NIR near 865 nm). No SWIR or thermal: NDVI/EVI2 and basic red-edge signals work, but moisture and burn-severity proxies that rely on SWIR won’t transfer directly.
- EarthDaily constellation class targets agronomy/monitoring with multiple red-edge bands and SWIR at meter-class resolution. Many Sentinel-2-style use cases move across more directly (validate thresholds).
How Sentinel-2 maps to Landsat 8/9 (for later)
Keep this as a reference when you cross-walk models; it’s intentionally lower on the page so S2 readers get what they came for first.
Purpose | Sentinel-2 band (nm, res) | Landsat 8/9 band (nm, res) | Notes |
---|---|---|---|
Coastal aerosol | B1 443, 60 m | B1 443, 30 m | Similar role |
Blue | B2 490, 10 m | B2 ~482, 30 m | Close analogs |
Green | B3 560, 10 m | B3 ~562, 30 m | Close analogs |
Red | B4 665, 10 m | B4 ~655, 30 m | Indices & mapping |
Red-edge | B5/6/7 (705/740/783), 20 m | — | S2 advantage (no OLI analog) |
NIR | B8 842, 10 m | B5 ~865, 30 m | Landsat NIR is closer to S2 B8A |
NIR narrow | B8A 865, 20 m | — | Useful for BRDF & harmonization |
Cirrus | B10 1375, 60 m | B9 ~1375, 30 m | Thin cirrus detection |
SWIR 1 | B11 1610, 20 m | B6 ~1610, 30 m | Moisture/burn |
SWIR 2 | B12 2190, 20 m | B7 ~2200, 30 m | Moisture/residue |
Panchromatic | — | B8 Pan 15 m | Landsat only |
Thermal | — | B10/11 TIRS 100 m | Landsat only |
Which band is not in Level-2A?
B10 (1.38 µm cirrus) is not included in L2A because it’s not a surface-reflectance signal. If you need cirrus screening, use B10 at L1C or rely on L2A scene classification/QA in your masking.
Practical notes on resolution & consistency
- 10 m bands: B2, B3, B4, B8
- 20 m bands: B5, B6, B7, B8A, B11, B12
- 60 m bands: B1, B9, B10
Hold the processing level constant across dates. Resample once to a fixed grid for per-pixel alignment. Prefer bilinear/cubic for continuous reflectance, nearest-neighbor for categorical masks.
Data handling
L1C/L2A store reflectance as integers with a scale factor of 10,000. Divide by 10,000 to convert to 0–1. Keep your grid and resampler consistent across dates.
FAQ
›Which Sentinel-2 bands are 10 m?
B2, B3, B4, and B8 are 10 m.
B5, B6, B7, B8A, B11, B12 are 20 m.
B1, B9, B10 are 60 m.
›Is Band 10 available at Level-2A?
No. B10 (cirrus) is only in L1C (TOA) and is used for thin-cloud detection.
For surface-reflectance workflows, use L2A with appropriate cloud/shadow masking or pre-screen with B10 at L1C.
›Should I use B8 or B8A for NDVI?
Use B8 (10 m) for standard NDVI/EVI2 at parcel scale.
B8A (20 m) is narrower and handy for some red-edge/BRDF workflows, but avoid mixing B8 and B8A in one time series unless you recalibrate thresholds.
›How close is Landsat’s NIR to Sentinel-2?
Landsat OLI NIR (~865 nm) is closer to Sentinel-2’s B8A than to B8. If you port thresholds from Landsat to S2, consider using B8A (20 m) for closer spectral alignment or re-validate when using B8 (10 m).
ClearSKY in practice
All ClearSKY deliveries come at 10 m across the stack so bands line up per pixel. Bands that are natively 20 m or 60 m are provided at 10 m using deep-learning super-resolution or carefully chosen resampling. If you want indices alongside multispectral data, set the Bandnames
field in your composite request to include the needed source bands and normalized-difference indices in one go. Use all
to receive the full Sentinel-2-style stack, or specify exactly what you need, for example B8,B4,[B8_B4]
to return NIR, Red, and a ready-to-use NDVI layer in one GeoTIFF. You can request Level-1C style inputs for speed or Level-2A style inputs for comparability and keep that choice consistent across a series.