Satellite Imagery Basics: Resolution, Bands and Revisit Time
When someone says "use satellite imagery" for a disaster project, the next question should always be: which satellite, which image, and why. Not all satellite images are the same. Some show fine detail but visit a place rarely. Some cover huge areas every day but show only a blur. Picking the wrong one wastes time or gives you a wrong answer with confidence.
This post covers three properties that define any satellite image: spatial resolution, temporal resolution and spectral resolution. Once you understand these, reading a satellite data page or a band list stops being confusing.
Spatial resolution: how much detail you get
Spatial resolution is the size of one pixel on the ground. A 10 m resolution image means each pixel represents a 10 by 10 meter square. Anything smaller than that square gets blended into a single average value.
- Coarse resolution (250 m to 1 km): good for watching weather, smoke plumes, or large-scale vegetation patterns. One pixel can cover several city blocks.
- Medium resolution (10 m to 30 m): the range of Sentinel-2 (10 m for some bands) and Landsat (30 m). Good for land cover, flood extent, burned area.
- High resolution (sub-meter to a few meters): commercial satellites like Maxar or Planet. Good for counting damaged buildings or spotting individual cars, but these images are not free and cover smaller areas per scene.
A common mistake is assuming "higher resolution is always better." A 30 cm image of an entire province would be an enormous file and take forever to process, and you often don't need that much detail to answer a regional question. Match the resolution to the size of the object you're actually trying to see. Mapping a flooded river valley needs maybe 10-30 m. Counting collapsed roofs after an earthquake needs sub-meter imagery, a topic covered in the post on assessing earthquake damage with satellites and AI.
Temporal resolution: how often you get a new look
Temporal resolution, often called revisit time, is how often a satellite passes over the same spot and takes a new image. This matters enormously for disaster work, because a disaster unfolds in hours or days, not months.
Revisit time depends on the satellite's orbit and how many satellites are in the same constellation. A single satellite might revisit a location every two weeks. A constellation of multiple satellites working together, like the Sentinel-2 pair, can cut that down to about five days. Commercial constellations with many small satellites can revisit daily or even multiple times a day.
Clouds complicate this further. An optical satellite (one that records visible and near-visible light, like a normal camera) cannot see through cloud cover. So the "revisit time" on a spec sheet is not the same as "time until you get a usable, cloud-free image." During a typhoon or a rainy flood event, you might wait many revisit cycles before the sky clears enough for optical imagery to be useful. This is one reason SAR satellites, which use radar and can see through clouds, are preferred for flood mapping — that method is explained in a separate post on how SAR satellites detect flooded areas.
For slow-moving hazards like drought or urban heat buildup, a revisit time of five to sixteen days is often fine. For fast hazards like flash floods or wildfire spread, you want the shortest revisit time you can get, and you may need to combine multiple satellite sources to fill the gaps.
Spectral resolution and how to read a band list
Spectral resolution refers to which wavelengths of light a satellite sensor can detect, and how finely it splits them. Human eyes see red, green and blue light. Satellite sensors often see those same colors plus wavelengths we cannot see, like near-infrared or shortwave infrared.
Each of these recorded wavelength ranges is called a band. A data product's band list might look like this for Sentinel-2:
Band 2 - Blue - 10 m Band 3 - Green - 10 m Band 4 - Red - 10 m Band 8 - NIR - 10 m Band 11 - SWIR - 20 m Band 12 - SWIR - 20 m
A few things to notice when you read a list like this:
- Band number is just an ID, specific to that satellite. Band 4 on Sentinel-2 is not the same wavelength as Band 4 on Landsat.
- Name or wavelength range tells you what the band actually measures: visible colors, near-infrared (NIR), shortwave infrared (SWIR), or thermal.
- Resolution per band can differ within the same satellite. Sentinel-2's visible and NIR bands come at 10 m, but its SWIR bands come at 20 m. If you combine bands, you need to resample them to match.
Why do extra bands beyond visible light matter? Healthy vegetation reflects a lot of near-infrared light, so comparing red and NIR bands can flag plant stress or burned land. Water absorbs infrared strongly, so NIR and SWIR bands help trace water bodies even when they look visually similar to land. These comparisons are called spectral indices, and the next post in this series, on NDVI and NDWI, explains exactly what those indices calculate and what they show.
More spectral bands generally mean more ways to distinguish ground features by their reflectance pattern, not just their shape or color as seen by eye. This is the core idea behind multispectral analysis.
Matching image choice to the question
In practice, choosing a satellite image means answering a few questions before you open any software:
- What size is the thing I'm looking for? A city-wide heat pattern needs far less spatial detail than a single collapsed bridge.
- How fast does it change? A wildfire front moves in hours. A land subsidence pattern moves over years. This sets your required temporal resolution.
- What physical property am I measuring? Vegetation health, water extent, surface temperature and burned area each rely on different bands. Check the band list before assuming a dataset has what you need.
- What's actually free and available? Sentinel and Landsat data are free and well documented on the Copernicus Data Space and USGS EarthExplorer. High-resolution commercial imagery usually is not free, and access may be limited during a live disaster.
These three properties also trade off against each other. A satellite that offers very high spatial resolution usually cannot also offer very frequent revisit at low cost, and a sensor with many finely split spectral bands often has a coarser spatial resolution. No single satellite is the best choice for every question. Part of doing this work well is picking the dataset that fits the specific hazard and the specific question, not the dataset that sounds the most impressive.
What to try next
Open the Copernicus Data Space or Google Earth Engine and look at the band list for a Sentinel-2 scene over your area. Note the resolution of each band and try to guess, before reading further, which bands you'd combine to detect vegetation versus water. Then move to the next post in this series, which explains NDVI and NDWI and shows what those index values actually mean on a map.
More materials (some in Korean) are on GeoAI for Disaster Safety.
This post was drafted with the help of AI tools and checked before publishing. Tool features and data portals change over time, so check the official documentation when you practice.
Comments
Post a Comment