Beyond True Color: Sentinel-2 Band Combinations Worth Knowing

Thirteen Bands, Infinite Combinations

Sentinel-2 carries a Multi-Spectral Instrument (MSI) with 13 spectral bands spanning visible light through short-wave infrared. Most users see the true color composite (red-green-blue) and stop there. That's like buying a 13-channel mixing board and only using the volume knob.

Each band measures reflectance at a specific wavelength range, and different surface materials have characteristic spectral signatures — patterns of high and low reflectance across wavelengths. Band combinations exploit these patterns to highlight features that are invisible or ambiguous in natural color.

The Essential Combinations

1. True Color (B4, B3, B2)

The baseline. This is what the scene would look like to human eyes (approximately). Useful for orientation and context, but limited by what our eyes can see.

• Best for: General landscape orientation, cloud identification, sharing with non-specialists
• Resolution: 10m
• Limitation: Vegetation appears uniformly green — you can't distinguish healthy from stressed crops

2. False Color Infrared (B8, B4, B3)

Probably the single most useful non-standard combination. Near-infrared (B8) is placed in the red channel, red (B4) in green, and green (B3) in blue.

Why it works: Healthy vegetation reflects strongly in NIR, so it appears bright red. Stressed or sparse vegetation appears pink or light red. Water absorbs NIR and appears dark blue/black. Urban areas appear cyan or gray.

• Best for: Vegetation health assessment, water body delineation, distinguishing vegetation types
• Resolution: 10m
• The "aha" moment: When you see a field that looks uniformly green in true color but shows clear red/pink variation in false color, revealing irrigation differences or crop stress

3. Agriculture Composite (B11, B8, B2)

Short-wave infrared (B11, 1610nm) in red, NIR (B8) in green, blue (B2) in blue.

Why it works: SWIR is sensitive to leaf water content, and combining it with NIR separates crop types effectively. Bare soil appears brown/tan. Actively growing crops are bright green. Mature or dry crops shift toward yellow/orange.

• Best for: Crop type identification, harvest monitoring, soil exposure mapping
• Resolution: 20m (SWIR bands are 20m native, upsampled when combined with 10m bands)

4. SWIR Composite (B12, B8A, B4)

Both SWIR bands with the red-edge NIR. This combination is particularly good for geological and moisture applications.

Why it works: Different rock and soil types have distinct SWIR reflectance patterns. Moist areas appear dark because water absorbs SWIR. Burn scars are clearly visible because charred vegetation has very different SWIR properties than green vegetation.

• Best for: Burn scar mapping, geology, mineral exploration, soil moisture (qualitative)
• Resolution: 20m

5. Natural Color Enhanced (B4, B3, B2 with contrast stretch)

Same bands as true color, but with the histogram stretched to use the full display range. This is sometimes underrated as a "combination" since it uses the same bands, but the visual improvement is significant.

The default rendering often leaves images looking hazy because atmospheric scattering compresses the useful data into a small portion of the dynamic range. A good contrast stretch can make a mediocre-looking scene quite informative.

Combinations I Use Less Often (But Others Swear By)

Atmospheric Penetration (B12, B11, B8A)

All infrared bands, no visible. Useful for seeing through thin haze, but the complete loss of natural color context can be disorienting. I find it most useful over desert terrain where visible wavelengths are dominated by atmospheric effects.

Vegetation Index (Color mapped NDVI)

Rather than a three-band composite, calculate NDVI as a single value and apply a color ramp. This is more quantitative than false color composites and works well for time-series comparison, but you lose the spatial detail that three-band composites provide.

Moisture Index (B8A, B11)

The Normalized Difference Moisture Index (NDMI) uses these two bands and is useful for drought monitoring and irrigation mapping. Like NDVI, it produces a single index value rather than a visual composite.

Practical Workflow

When I look at a new Sentinel-2 scene, I typically follow this sequence:

1. True color first — orient myself, assess cloud cover, get context
2. False color infrared — immediately shows vegetation distribution and health
3. Specific combination based on the question — agriculture composite for farmland, SWIR for burn scars or geology

The switch takes seconds and the additional information is substantial. There's no reason to limit yourself to one view.

A Note on Resolution

Sentinel-2's bands have three different native resolutions:

• 10m: B2 (Blue), B3 (Green), B4 (Red), B8 (NIR)
• 20m: B5-B7 (Red Edge), B8A (NIR narrow), B11-B12 (SWIR)
• 60m: B1 (Coastal), B9 (Water Vapor), B10 (Cirrus)

When you create a composite mixing 10m and 20m bands, the 20m bands are typically resampled to 10m. This doesn't add real detail — you're still limited by the coarser band — but it allows the combination to display at 10m pixel spacing.

For vegetation work, the 10m bands (especially B4 and B8 for NDVI) give you the finest spatial detail. For moisture and geological work, you're working at 20m resolution regardless.

On the Sentinel-2 viewer, you can switch between true color, false color infrared, and spectral indices like NDVI, EVI, NDMI, and NBR — each revealing different surface properties from the same scene. For custom three-band composites like the agriculture or SWIR combinations described above, desktop GIS software provides full flexibility.