Lesson 3 of 6 · Earth observation
Read the color recipe
9 minute lesson · 3 checks along the way
Your goal for this lessonInterpret a satellite image through its bands, display mapping, and processing choices.
Words with a dotted underline open a short definition right here in the lesson.
Color is a mapping from numbers
A monitor mixes red, green, and blue light to create its colors. A satellite image assigns measured quantities to those display channels. A natural-color uses visible red, green, and blue measurements in their corresponding channels, aiming for a familiar appearance. Even then, , atmospheric effects, contrast adjustments, and the display influence what you see. It is a processed representation of measurements.
An assigned-color image may place or shortwave-infrared measurements into channels your eyes can see. “” names that mapping; it does not mean the observations are fictional. A red patch could represent strongly reflected near-infrared radiation, high temperature, or a category on a thematic map. Without the channel recipe or legend, the hue does not identify which of those meanings applies.
Keep the observation and its display separate
Healthy leafy vegetation often reflects much more near-infrared than visible red radiation. Assigning near-infrared to the screen’s red channel can therefore make vegetation look red. This does not show red-hot trees. Other combinations emphasize moisture, exposed ground, snow, or burn scars. Interpretation still requires the scene, wavelength behavior, and processing information; a single universal color key does not cover every satellite product.
ESA’s Sentinel-2B view of Semarang, Indonesia, is explicitly a image. It reveals coastal water, fields, and built areas through a familiar visible-light presentation. It can help identify spatial patterns, but it does not directly display every parcel’s crop, ownership, water quality, or . A useful caption identifies the acquisition and processing, while separating visible structure from claims requiring additional evidence.
Copernicus Sentinel-2 captured Semarang, Indonesia, on May 9, 2018. ESA’s processed true-color view shows the city and harbor along the lower-left coast and patterned agricultural land inland. Water color and field color alone do not establish depth, crop condition, or land-use history.
Look closely
Compare a built-up area, a field patch, and coastal water. Which distinctions are visible, and which proposed explanation would need a different band or another date?
See what to notice
Shapes, textures, and displayed visible colors separate parts of the scene. A depth estimate, vegetation-health claim, or claim of recent conversion needs an appropriate calibrated measurement, a model, and often observations at other dates; one attractive image does not provide those tests.
Check 1 of 3
10 pointsUse the reading above. You can try again without losing points.
Work through a decision
Remap a hypothetical sample
Consider invented calibrated reflectances: blue 0.05, green 0.20, red 0.10, and near-infrared 0.60. For this exercise, map each number linearly to its display channel using the same zero-to-one scale, with no gamma adjustment or extra . A natural-color triplet is screen red 0.10, green 0.20, blue 0.05. Green is the strongest of its three displayed components.
Now assign near-infrared to screen red, measured red to screen green, and measured green to screen blue. The triplet becomes 0.60,0.10,0.20, so red dominates. None of the underlying reflectances changed. Only the display assignment changed. The result demonstrates why comparing two pictures by hue alone can invent a physical change when their band combinations or display scales differ.
One measured spectrum, two color assignments
Original hypothetical band values, with a common linear zero-to-one display scale and no gamma or separate stretch. Colored swatches illustrate the stated mapping; no thermal measurement is included.
Open full-size diagram in a new tab: One measured spectrum, two color assignmentsCheck 2 of 3
10 pointsUse the reading above. You can try again without losing points.
Read a legend before drawing a boundary
A contrast stretch can also make a small of values occupy much of the available display brightness. Two images stretched independently can look equally bright despite different measured values, or differently bright despite overlapping values. For a quantitative comparison, use calibrated data and compatible scales. A screenshot’s color values generally cannot recover the original measurements without the complete display transformation.
When presented with a striking satellite image, first identify the sensor, bands, date, and whether colors encode measurements or classified categories. Then look for spatial patterns and compare them with independent context. A good interpretation can say both what becomes easier to see and what remains unknown. Clear disclosure of assigned color makes invisible wavelengths understandable without pretending they are ordinary human vision.
Check 3 of 3
10 pointsUse the reading above. You can try again without losing points.
Put it together
Your lesson progress
Each check connects to a part of the story. Complete all 3 to earn your 25-point lesson bonus.
Go to the source
Explore the evidence behind this lesson. These optional readings and videos open in a new tab.