visualizationtutorialfundamentalsremote sensingbest practices

Choosing the Right Colormap for Satellite Data Visualization

Kazushi MotomuraJuly 20, 20256 min read
Choosing the Right Colormap for Satellite Data Visualization

Quick Answer: Sequential colormaps (viridis, inferno) are best for continuous data like elevation or backscatter intensity. Diverging colormaps work for data with a meaningful center point, like NDVI difference maps. Never use rainbow/jet — it creates false boundaries and is unreadable for colorblind users. Viridis is the safest default: it's perceptually uniform, colorblind-friendly, and prints well in grayscale.

Why Colormaps Matter More Than You Think

A satellite image is fundamentally an array of numbers. The colormap translates those numbers into colors that your visual system can interpret. Choose poorly, and you'll see patterns that don't exist — or miss patterns that do.

This isn't just aesthetics. Studies in perceptual science have shown that colormap choice directly affects:

  • Pattern detection accuracy — Can you see the gradient, or does it look like discrete steps?
  • Quantitative estimation — Can you roughly estimate a value from its color?
  • Anomaly detection — Do real features stand out, or do they blend into colormap artifacts?

Types of Colormaps

Sequential Colormaps

Go from low values (dark) to high values (bright) through a smooth gradient. Best for data with a natural ordering from low to high.

ColormapCharacterBest For
ViridisPurple → green → yellowGeneral purpose, safest default
InfernoBlack → red → yellow → whiteHigh dynamic range data, dramatic visualization
PlasmaPurple → orange → yellowSimilar to inferno, slightly less contrast at extremes
HotBlack → red → yellow → whiteTemperature-like data, thermal imagery
YlOrBrYellow → orange → brownSoil moisture, elevation, earth-toned data
Greys_rWhite → blackTraditional remote sensing display, SAR imagery

Diverging Colormaps

Two sequential ramps meeting at a neutral center. Best for data with a meaningful zero or center point.

These are ideal for difference maps where negative = decrease, zero = no change, positive = increase.

Categorical Colormaps

Distinct colors for distinct classes. Used for classification maps (land cover types, crop types) but not for continuous satellite data.

The Case Against Rainbow (Jet)

The rainbow colormap is still widely used but has serious problems:

  1. Perceptual non-uniformity — Equal data differences produce unequal perceived color differences. The yellow band is much brighter than the blue and red bands, creating false "features" at yellow boundaries.

  2. Colorblind inaccessibility — Roughly 8% of men have some form of color vision deficiency. Red-green colorblindness makes large portions of the rainbow colormap indistinguishable.

  3. Grayscale failure — When printed in black-and-white, the rainbow colormap becomes a confusing mess with no consistent light-to-dark ordering.

  4. False boundaries — The human visual system perceives sharp transitions between hue categories (blue→cyan, green→yellow). These perceived "edges" in the colormap create false boundaries in the data.

Colormap Selection Guide

SAR Backscatter Intensity

Recommended: Greys_r (reversed grayscale)

SAR imagery is traditionally displayed in grayscale, where brightness corresponds to backscatter intensity. This is intuitive: bright = strong return (urban, rough surfaces), dark = weak return (water, smooth surfaces). Most SAR literature uses grayscale, making it easier to communicate your results.

Vegetation Indices (NDVI, EVI, SAVI)

Recommended: Custom green ramp or diverging red-green

For NDVI and similar vegetation indices, a colormap that goes from brown/red (low vegetation) through yellow (sparse vegetation) to green (dense vegetation) is intuitive because it matches the actual landscape colors.

Water Indices (NDWI, MNDWI)

Recommended: Blue sequential or diverging blue-brown

Blue for water is culturally universal and immediately understood. A diverging colormap with brown for land and blue for water detection makes the land-water boundary obvious.

Burn Severity (dNBR)

Recommended: Diverging green-red

Green for unburned/regrowth, neutral for no change, red/orange for burn severity. This matches the traffic-light intuition (green = good, red = bad).

Elevation and Terrain

Recommended: YlOrBr or terrain colormaps

Earth tones (greens for low elevations, browns for mid, whites for high) match the natural appearance of landscapes and are easily interpreted.

General Purpose / Unknown Data

Recommended: Viridis

When you're exploring unfamiliar data and don't know what patterns to expect, viridis is the safest choice. It's perceptually uniform (equal data steps produce equal perceived color steps), works for colorblind users, and maintains ordering when printed in grayscale.

Min/Max Scaling: The Hidden Variable

Even with the perfect colormap, poor min/max scaling ruins the visualization. The colormap maps the full range between your min and max values. If the range is too wide, subtle features disappear. If it's too narrow, you clip important data.

Auto-scaling

Most tools set min/max to the data's actual range. This works when data is well-distributed but fails when outliers stretch the range (a single very bright pixel makes everything else look dark).

Percentile-based scaling

Set min to the 2nd percentile and max to the 98th percentile. This clips the extreme outliers and uses the full colormap for the meaningful data range. This is usually the best starting point.

Domain-specific scaling

For standardized indices, use known ranges:

  • NDVI: -0.2 to +0.9
  • NDWI: -0.5 to +0.5
  • SAR backscatter (dB): -25 to +5

These fixed ranges make different scenes directly comparable.

Practical Tips

Test with Multiple Colormaps

Don't commit to the first colormap you try. Switch between 2-3 options to see if the patterns you're seeing are real or colormap artifacts. If a feature only appears with one colormap, it's likely an artifact.

Match Colormaps to Your Audience

Technical audiences familiar with SAR expect grayscale. General audiences may find viridis or inferno more engaging. Environmental reports benefit from intuitive green/brown/blue schemes.

Be Consistent Across Comparisons

When comparing multiple images (different dates, different areas), always use the same colormap with the same min/max scaling. Different colormaps make visual comparison meaningless.

Colormaps in Off-Nadir Delta

Off-Nadir Delta applies colormaps tailored to each data type:

Sentinel-2 optical — Each spectral index uses an optimized colormap automatically:

  • NDVI, EVI, SAVI, NBR → RdYlGn (Red-Yellow-Green) for intuitive vegetation mapping
  • NDWI, MNDWI → Blues for water detection
  • NDBI → Oranges for built-up areas
  • NDSI → CoolWarm (Blue-White-Red) for snow/ice
  • Iron Oxide → Hot (Black-Red-Yellow-White)
  • Clay Minerals → Terrain for geological context

Sentinel-1 SAR — Grayscale display for VV/VH polarization, with RdYlGn for SAR vegetation indices (RVI, RFDI).

VIIRS Nighttime Lights — User-selectable colormaps including Greys_r, Viridis, Plasma, Inferno, Hot, and YlOrBr. Switch between them using the visualization selector to find the best representation for nighttime light intensity data.

The key takeaway: you don't need to manually choose colormaps for most satellite data — the platform applies scientifically appropriate defaults. For nighttime lights, experiment with different options to highlight the patterns most relevant to your analysis.

Kazushi Motomura

Kazushi Motomura

Remote sensing specialist with 10+ years in satellite data processing. Founder of Off-Nadir Lab. Master's in Satellite Oceanography (Kyushu University).