Resolution: spatial, spectral, radiometric and temporal

Detail

What "resolution" means

• Resolution (resolving power) is how much detail a sensor can capture. NCERT's heading for this is "Resolving Powers of the Satellites".
• A sensor has four kinds of resolution. Each one answers a different question:

Type	Question it answers	Unit / measure	Set by
Spatial	How small an object can it see?	Ground size of one pixel (metres)	Sensor optics, height of the orbit
Spectral	How many colours (wavelength bands) can it tell apart?	Number of bands and how narrow they are	Detector design
Radiometric	How small a change in brightness can it tell apart?	Grey levels (bits)	Detector sensitivity, data coding
Temporal	How often does it see the same place again?	Revisit interval (days)	Orbit

Spatial resolution

• Spatial resolution: how well a sensor separates two objects that lie close together on the ground.
• It is given as the ground size of one pixel (pixel: the smallest square of a digital image). A 30 m sensor records one value for each 30 m × 30 m patch of ground.
• The spectacles analogy (NCERT): reading glasses help you tell apart letters that sit close together. In the same way, a sensor with finer spatial resolution can tell apart objects that sit close together.
• A smaller number means finer detail. 5.8 m is finer than 80 m.
• Table 6.1 (NCERT), coarsest to finest:

Sensor	Spatial resolution
Landsat MSS (USA)	80 m
IRS LISS-I (India)	72.5 m
IRS LISS-II (India)	36.25 m
Landsat TM (USA)	30 m
IRS LISS-III (India)	23 m
SPOT HRV-I (France)	20 m
SPOT HRV-II (France)	10 m
IRS PAN (India)	5.8 m

• LISS-II (36.25 m) is exactly half of LISS-I (72.5 m). So LISS-II sees twice as much detail along each side of a pixel.
• Later data:
• Resourcesat-2A was launched on 7 December 2016 by PSLV-C36 into a Sun-synchronous polar orbit. It carries three cameras [2]:
  • LISS-4: 5.8 m, three bands in the VNIR (visible and near-infrared) [2]
  • LISS-3: 23.5 m, three VNIR bands and one SWIR (short-wave infrared) band [2] (NCERT, older LISS-III: 23 m)
  • AWiFS (Advanced Wide Field Sensor): 56 m, three VNIR bands and one SWIR band [2]
• Landsat 8 was launched on 11 February 2013. Its multispectral bands are 30 m, the same as the older Landsat TM. Its panchromatic band is 15 m and its thermal band is 100 m (resampled, that is re-gridded, to 30 m) [3].
• Panchromatic (one wide black-and-white band) images are always sharper than the multispectral images from the same satellite. This is because one wide band collects more light, so each pixel can be made smaller.
• Cartosat-3 was launched on 27 November 2019 by PSLV-C47 from Sriharikota. It flies in a Sun-synchronous polar orbit at 509 km, has a mission life of 5 years, and ISRO calls it its "most complex and advanced" Earth observation satellite. It is a third-generation agile satellite, which means it can quickly tilt to point at a target [4].

Spectral resolution

• Spectral resolution: how many bands (wavelength ranges) a sensor records, and how narrow each band is.
• More bands and narrower bands mean finer spectral resolution. With them, the sensor can tell apart surfaces that look the same to the eye.
• The idea behind it is the dispersion of light (Box 6.2, NCERT):
• White light is a mix of many wavelengths.
• A prism, or raindrops in a rainbow, bend each wavelength by a slightly different amount.
• So white light splits into separate colours (VIBGYOR).

• A multispectral sensor does the same thing. It splits the reflected energy into bands and records each band separately.

• Why bands help: each surface has its own spectral signature, the pattern of how much energy it reflects in each band.

• NCERT example: Najafgarh, Delhi (Fig. 6.11, Resourcesat-1, 2005):
• Fresh water absorbs in band 4 (infrared), so water bodies look dark in band 4.
• Dry surfaces reflect in band 2 (green), so dry land looks bright in band 2.

• Putting the two bands side by side separates water from dry land clearly.

• Later data: Landsat 8 records 11 spectral bands across two instruments, OLI (Operational Land Imager) and TIRS (Thermal Infrared Sensor) [3]. Resourcesat-2A's LISS-3 and AWiFS add a SWIR band to the three VNIR bands [2]. SWIR is useful for mapping soil moisture and crop water stress.

Radiometric resolution

• Radiometric resolution: how well a sensor detects small differences in energy (radiance, the energy reflected or given off by a surface).
• It is counted in grey levels, the number of brightness steps between pure black and pure white.
• More grey levels mean finer detail in tone. For example, two kinds of crop with only slightly different brightness can still be told apart.
• Table 6.1 (NCERT):

Sensor	Grey-level range	Bits (data size)
Landsat MSS	0–64	6-bit
IRS LISS-I/II	0–127	7-bit
Landsat TM, SPOT HRV	0–255	8-bit
Landsat 8 OLI (2013)	4,096 values	12-bit [3]

• How bits work: n bits give 2ⁿ levels. 6 bits give 64 levels, 7 bits give 128, 8 bits give 256 and 12 bits give 4,096.
• Landsat 8 collects 12-bit data, up from the 8-bit data of Landsat 7 ETM+. That is 4,096 possible values per pixel instead of 256, so it can pick up more subtle detail on Earth's surface [3]. Most modern sensors work in 10–12 bits.

Temporal resolution

• Temporal resolution: how long a satellite takes to image the same area again (the revisit interval).
• The orbit sets it. Most Earth observation satellites fly in a Sun-synchronous polar orbit: a near-polar orbit that crosses each place at the same local time. Because the light is the same each time, images from different dates can be compared fairly.
• Examples:
• Landsat 8: 16 days on its own. Paired with Landsat 9, the two together give an 8-day repeat [3].

• Resourcesat-2A LISS-4: it can be steered ±26° across its track (tilted sideways), which gives a five-day revisit [2].

• Tilting the sensor sideways is one way to see a place again sooner without changing the orbit.

Change detection

• Change detection: comparing images of the same area taken on different dates to find out what has changed. It needs good temporal resolution.
• How it works:
• Image the same area on date 1 and on date 2.
• Line the two images up pixel by pixel.

• Mark the areas where the spectral values have changed. These show new growth, flooding, damage, new buildings and so on.

• NCERT examples:

Example	Images	What changes
Himalaya (Fig. 6.9)	IRS, May and November	Coniferous forest stays green all year. Deciduous trees (which shed their leaves) and crops change tone between the seasons.
Banda Aceh, Sumatra, Indonesia (Fig. 6.10)	Before and after the 26 December 2004 Indian Ocean tsunami	Coastal land, buildings and vegetation washed away; the shoreline changed

• GIS overlay is another way to detect change. Maps of the same area from two dates are stacked as layers to show what changed (the Aligarh example, section 10).
• Uses: crop monitoring, deforestation, urban sprawl, flood and tsunami damage, glacier retreat.

The resolution trade-off (beyond NCERT)

• Finer spatial resolution usually means a narrower swath (swath: the width of ground imaged in one pass).
• The sensor sees a smaller strip of ground each time.

• So it takes more days to cover the whole Earth, and the revisit interval gets longer.

• Data from Resourcesat-2A [2] and Landsat 8 [3]:

Sensor	Spatial resolution	Swath	Note
AWiFS (Resourcesat-2A)	56 m (coarse)	wide	used for regional crop watch [2]
LISS-4 (Resourcesat-2A)	5.8 m (fine)	narrow	needs ±26° tilt for a 5-day revisit [2]
Landsat 8 OLI	30 m	185 km	16-day revisit [3]

• The solution is a constellation, a group of satellites working together. Landsat 8 with Landsat 9 halves the revisit from 16 to 8 days [3]. India runs the Resourcesat, Cartosat and RISAT series for the same reason.
• The other resolutions involve trade-offs too. Narrower spectral bands collect less energy each, so they need bigger pixels or more sensitive detectors. Finer radiometric resolution (more bits) creates more data to store and send to Earth.

Prelims Hooks

• Spatial resolution = ground size of one pixel. The smaller the number, the finer the detail. The finest sensor in NCERT Table 6.1 is IRS PAN at 5.8 m. The coarsest is Landsat MSS at 80 m.
• Order from coarse to fine (NCERT): Landsat MSS 80 m → LISS-I 72.5 m → LISS-II 36.25 m → Landsat TM 30 m → LISS-III 23 m → SPOT HRV-I 20 m → SPOT HRV-II 10 m → IRS PAN 5.8 m.
• Grey levels: Landsat MSS 0–64 (6-bit), IRS LISS 0–127 (7-bit), Landsat TM/SPOT 0–255 (8-bit). Landsat 8 OLI is 12-bit, giving 4,096 values [3].
• Spectral resolution is based on the dispersion of light (prism, rainbow). It is not the same as spatial resolution, a common trap.
• Najafgarh, Delhi (Resourcesat-1, 2005): fresh water absorbs in band 4 (IR). Dry surfaces reflect in band 2 (green).
• Temporal resolution is set by the orbit. Landsat 8 alone: 16 days. Landsat 8 + 9: 8 days [3]. Resourcesat-2A LISS-4: 5-day revisit using ±26° steering [2].
• Resourcesat-2A (PSLV-C36, 7 December 2016) carries LISS-4 (5.8 m), LISS-3 (23.5 m) and AWiFS (56 m) [2].
• Cartosat-3: PSLV-C47, 27 November 2019, Sun-synchronous orbit at 509 km, an agile satellite [4].
• Change detection, NCERT cases: Himalaya (May vs November: conifers vs deciduous trees and crops) and Banda Aceh, Indonesia (December 2004 tsunami).
• Trap: finer spatial resolution usually means a narrower swath and a longer revisit, not a shorter one.

Mains Points

• Each resolution suits a different job, so planners must pick the right mix:
• Fine spatial resolution (LISS-4 5.8 m [2], Cartosat-3 [4]) suits cadastral (land-ownership) maps, urban planning and infrastructure.
• Coarse but wide sensors (AWiFS 56 m [2]) suit crop estimates over a whole state.

• Many bands (spectral) and many grey levels (radiometric) help with water, soil-moisture and crop-stress mapping.

• Temporal resolution and change detection matter for disaster management (GS-III):

• Before-and-after images (Banda Aceh 2004) show where the damage is.
• A short revisit (5 days [2], 8 days [3]) helps track floods, cyclone landfall and forest fires as they happen.

• Seasonal images (Himalaya, May vs November) help monitor forests and agriculture.

• The resolution trade-off explains India's satellite strategy:

• One satellite cannot be fine, wide and frequent all at once.
• So India has built series and constellations: Resourcesat-2/2A, Cartosat and RISAT.

• Their data is shared through Bhuvan and opened up under the National Geospatial Policy 2022.

• India's progress in resolution shows its growing self-reliance in space:

• LISS-I at 72.5 m (IRS-1A, 1988), then 5.8 m (IRS PAN, LISS-4), then Cartosat-3 in 2019 [4].
• This cuts dependence on foreign imagery for defence, mapping and governance.