III - WHAT IS A DIGITAL IMAGE?
2- CHARACTERISTICS OF AN IMAGE
2.2- Spatial resolution
Like images made by a camera or like your computer screen, earth observation images are composed of pixels. The more pixels for the same area of terrain, the more the details of an image are visible and the finer the "resolution".
By 'the resolution' of an image, we usually mean the spatial resolution. This corresponds to the number of pixels per unit length or, as is the case with remote sensing, to the size of the area covered by a pixel.
For example, each detector of the SPOT -5 satellite's HRG sensor captures an area of about 10 x 10 m on Earth. The spatial resolution or "ground resolution" is therefore 10 m. Since this sensor is equipped with 6000 light-sensitive elements, it can therefore scan an area of 6000 x 10 m = 60 km on the Earth's surface by passage.
|
The first civilianeEarth observation sensor, the Multispectral Scanner (MSS), was aboard Landsat-1 (formerly called ERTS) and had an Instantaneous Field of View (IFOV) of 83m in the scanning direction and 68m in the transverse direction. The spatial resolution was thus 68m x 83m, but the pixels were usually resampled into 60m x 60m square cells. From the 1980s and 1990s, low spatial resolution sensors, i.e. resolutions on the order of one kilometre, were also put into orbit (e.g. SPOT VEGETATION and its successor PROBA-VEGETATION, and MODIS). These sensors have the advantage of recording large areas and are ideally suited for global-scale analysis. Since the early 2000s, a new generation of Earth observation satellites has been producing images with submetric resolution - i.e. pixels whose side is less than one metre. The European Pleiades satellites are a good example of this major technological breakthrough with a spatial resolution of 50 cm. |
||
|
||
The resolution of some earth observation satellites is given in this table:
| Satellite | Sensor | Ground resolution |
| Meteosat 11 | Seviri | 1 km |
| PROBA-V | VEGETATION | 100 m - 300 m |
| Landsat 8 & 9 | TIRS | 100 m |
| Landsat 8 & 9 | OLI Multispectral | 30 m |
| Landsat 8 & 9 | OLI Panchromatic | 15 m |
| Sentinel-2 | Multispectral | 10 - 60 m |
| SPOT 6 & 7 | Multispectral | 6 m |
| Pléiades 1-A & 1-B | Multispectral | 2 m |
| SPOT 6 & 7 | Panchromatic | 1,5 m |
| Pléiades 1-A & 1-B | Panchromatic | 0,50 - 2 m |
To illustrate the importance of resolution for image quality, here is an animation showing 4 consecutive images of the same area seen by sensors of increasing resolution: the VEGETATION sensor (100 m resolution) on the PROBA-V satellite, the OLI sensor (30 m resolution) on the Landsat-9 satellite in multispectral mode, the MSI sensor (10 m resolution) on the Sentinel-2 satellite and the HiRI sensor (1.5 m resolution) on the Pléiades-1A satellite. The area is the Atomium site in Brussels.
The spatial resolution of a radar system (SAR) is calculated independently in the azimuth and "range" directions and depends on the characteristics of the radar system and sensor and not on the distance between the sensor and the observed scene. The resolution in the azimuth direction is determined by the antenna length. Resolution in the range direction is determined by the signal bandwidth of the transmitted pulse. As for optical sensors, radar images can reach 1-metre and even sub-metre resolutions