Why satellite details matter
Every remote sensing task starts with a choice of data. A map of city lakes needs pixels small enough to show them. A study of crop growth needs images often enough to follow a season. A flood map has to work through cloud. Each of these needs depends on details of the satellite, and most mistakes in remote sensing start with data that was never suited to the question.
Three words are easy to mix up.
- The platform is the satellite itself, such as Landsat 8.
- The sensor is the instrument on it, such as the Operational Land Imager.
- The product is the processed data you download, such as a surface reflectance scene.
Orbits
An orbit decides where the satellite goes and how often it returns.
| Orbit | Height | What it does | Used for |
|---|---|---|---|
| Geostationary | About 35,786 km | Stays above one point on the equator | Weather, as in the Indian INSAT series |
| Sun-synchronous polar | About 600 to 800 km | Passes over each place at the same local solar time | Most land imaging: Landsat, Sentinel-2, Resourcesat |
| Other low orbits | A few hundred km | Varies | Some commercial and research missions |
A sun-synchronous orbit matters for GIS work because the sun is at about the same angle in every image of a place. Shadows and brightness repeat from date to date, so images taken in the same season compare well. Landsat 8 crosses the equator at about 10:00 local time, and Sentinel-2 at about 10:30.
Swath, repeat cycle and revisit
The swath is the width of ground the sensor sees on one pass. A wide swath covers more area per image and needs fewer scenes to cover a region.
The repeat cycle is the time the satellite takes to return to the same track. Landsat 8 repeats every 16 days. The revisit time is how often you can get an image of a given place, which can be shorter than the repeat cycle when two satellites share an orbit or when the tracks overlap. Clouds make the useful revisit longer. In a monsoon region, a five-day revisit on paper can leave you with a few cloud-free images in a season.
Sensors and bands
Most land imaging uses a passive optical sensor that records reflected sunlight in several bands. Radar sensors are active. They send their own pulses and work through cloud and at night. The note on remote sensing fundamentals explains the difference. A sensor is described by its bands, its pixel size and its radiometric depth, which is how many brightness levels it records.
Which bands a sensor has decides which tasks are possible. Without a shortwave infrared band you cannot calculate a water index that uses it. Without a thermal band you cannot map surface temperature.
Comparing the main missions
| Mission | Sensor type | Pixel size | Revisit | Swath | Access |
|---|---|---|---|---|---|
| Landsat 8 and 9 | Optical, thermal | 30 m, with a 15 m panchromatic band and 100 m thermal | 16 days each, 8 days together | 185 km | Free |
| Sentinel-2 | Optical, 13 bands | 10, 20 and 60 m depending on band | About 5 days with two satellites | 290 km | Free |
| Sentinel-1 | C-band radar | 10 m pixels, with about 20 m resolution in the common wide-swath mode | 12 days per satellite | 250 km in that mode | Free |
| MODIS on Terra and Aqua | Optical, 36 bands | 250 m, 500 m and 1 km | Near daily | 2,330 km | Free |
| Resourcesat (India) | LISS-IV, LISS-III and AWiFS | About 5.8 m, 23.5 m and 56 m | Several days to weeks, by sensor | 70 km, 141 km and 740 km | Free through ISRO Bhoonidhi, with a login |
| Commercial very high resolution | Optical | Below 1 m | Tasked on request | Narrow | Paid |
The numbers are approximate. They differ a little between sensors on the same family and between modes, so check the mission documents for the exact figure you need.
Processing levels and products
Raw data is processed in stages. The names differ between missions, but the idea is the same.
| Level | What it contains | Notes |
|---|---|---|
| Level 1 | Calibrated and located measurements | Landsat Level-1 stores digital numbers. Sentinel-2 Level-1C stores reflectance at the top of the atmosphere |
| Level 2 | Corrected for the atmosphere, as surface reflectance | Landsat Collection 2 Level-2 and Sentinel-2 Level-2A. Landsat Level-2 also has surface temperature |
Use surface reflectance when you compare dates, calculate indices or classify. Top-of-atmosphere data still carries the effect of haze.
Values in a product are usually integers stored with a scale factor. Landsat Collection 2 Level-2 reflectance is the stored value times 0.0000275 minus 0.2, and its surface temperature in kelvin is the stored value times 0.00341802 plus 149.0. Sentinel-2 Level-2A reflectance is the stored value divided by 10,000. The original files from processing baseline 04.00 onward also carry an offset of 1000. Check the metadata, and look at a dark target such as a clear lake, because some copies have already removed the offset.
Reading scene names and tiles
A scene name holds the details of the product. The Landsat scene used in this series is LC08_L2SP_144051_20250329_20250401_02_T1.
| Part | Meaning |
|---|---|
| LC08 | Landsat 8, with the optical and thermal sensors |
| L2SP | Level-2, with surface reflectance and surface temperature |
| 144051 | Path 144 and row 051 of the Landsat reference grid |
| 20250329 | Date of acquisition, 29 March 2025 |
| 20250401 | Date of processing |
| 02 and T1 | Collection 2 and the highest quality tier |
Sentinel-2 data is cut into tiles on the military grid reference system. The tile 43PGQ means UTM zone 43, latitude band P and square GQ. A tile is about 110 kilometres square, and its pixels are in the UTM zone of the tile, which is EPSG:32643 here.
From details to decisions
| If the task is | Look at | A likely choice |
|---|---|---|
| Mapping city land cover | Pixel size, bands | Sentinel-2 at 10 m, or Landsat at 30 m for a wider area |
| Following crops through a season | Revisit, cloud cover | Sentinel-2, or MODIS for very large areas |
| Mapping surface temperature | Thermal band | Landsat thermal data |
| Mapping a flood under cloud | Active sensing | Sentinel-1 radar |
| Looking back several decades | Archive length | Landsat, which has recorded since 1972 |
| Mapping individual buildings or trees | Sub-metre pixels | Commercial very high resolution imagery |
| A wide, low-cost regional view | Swath and cost | MODIS, or Resourcesat AWiFS |
Example: the data in the QGIS series
The tutorials use a Sentinel-2 Level-2A scene of tile 43PGQ from 15 March 2025 and a Landsat 8 Level-2 scene of path 144, row 051, from 29 March 2025. The two are two weeks apart, which keeps the season the same. The Sentinel-2 bands give 10 metre pixels and the Landsat bands give 30 metre pixels, so the same city looks sharper in one and smoother in the other. The Landsat thermal band is what makes the surface temperature map in the interpolation tutorial possible.
Common mistakes
- Choosing data by habit instead of by the task.
- Forgetting that a wide swath at coarse resolution cannot show small objects.
- Comparing images from different seasons or sun angles.
- Using top-of-atmosphere data where surface reflectance is needed.
- Applying a scale factor or offset the file does not need, or skipping one it does.
- Relying on a revisit time without counting the cloud.
Glossary
| Term | Meaning |
|---|---|
| Platform | The satellite that carries the instruments |
| Sensor | The instrument that records the data |
| Swath | The width of ground seen on one pass |
| Repeat cycle | The time a satellite takes to return to the same track |
| Revisit time | How often a given place can be imaged |
| Surface reflectance | The fraction of light reflected by the ground, after atmospheric correction |
Practise it
The tutorial Band combinations and spectral indices uses Sentinel-2 and Landsat 8 side by side and applies their scale factors.
Test yourself
1. Why do most land imaging satellites use a sun-synchronous orbit?
The satellite passes over each place at the same local solar time, so the sun angle is similar from one image to the next and images compare well.
2. What is the difference between repeat cycle and revisit time?
The repeat cycle is the time for one satellite to return to the same track. The revisit time is how often a place can be imaged, which can be shorter when satellites share an orbit or tracks overlap.
3. Which mission suits a flood map during the monsoon, and why?
Sentinel-1, because radar is active and works through cloud.