Data products and image interpretation

Detail

1. Data products: images and photographs

• Data product: the final output of a sensor system. It reaches the user in one of two forms: photographic (on film) or digital (as numbers).
• Image: any picture record of an area, whatever wavelength (part of the electromagnetic spectrum) was used to record it. It may be visible light, infrared or microwave.
• Photograph: an image recorded on photographic film. Film works only in the visible and near-infrared range.
• Key rule: all photographs are images, but not all images are photographs.
• A thermal-infrared or radar picture is an image. It is not a photograph, because film cannot record those wavelengths.

Feature	Photograph	Digital image
Recording medium	Chemical film	Electronic detectors that store numbers
Wavelength range	Visible and near-infrared only	Visible, infrared, thermal, microwave
Smallest unit	Film grain	Pixel
Enlargement	Can be enlarged without losing detail	Shows only square pixels after a point
How information is taken out	Mainly by eye (visual interpretation)	By computer (digital image processing) and by eye

2. Aerial photographs

• Aerial photograph: a photograph of the ground taken from a camera fitted in an aircraft.
• Film types:
• black-and-white (the most common)
• colour
• black-and-white infrared

• colour infrared

• Strengths:

• It can be enlarged a great deal without losing detail.

• It became the base for topographic mapping, the detailed maps that show relief, drainage and settlements. The Survey of India uses this kind of map.

• Place in history: aerial photography was the first stage of remote sensing. Satellite imaging came later, starting with Landsat-1 in 1972.

• Today: the National Geospatial Policy 2022 was notified on 28 December 2022. It aims for a high-resolution topographical survey and mapping system by 2030 and a highly accurate Digital Elevation Model (DEM: a digital model of ground height) for the whole country [7].
• Target resolution by 2030: 5–10 cm for urban and rural areas, and 50–100 cm for forests and wastelands [7].
• These very fine targets can be met only by aerial and drone photography, not by most satellites. So the old aerial method stays important.

3. Digital images, pixels and digital numbers

• Digital image: a picture stored as a grid of numbers, laid out in rows and columns.
• Pixel (picture element): the smallest cell of a digital image.
• Each pixel has a fixed address: its row number and column number.

• Each pixel stands for one small square of ground.

• Digital number (DN): the number stored in each pixel. It is the average brightness (intensity) of the energy reflected or given off by that ground cell.

• A bright surface such as dry sand gets a high DN.

• A dark surface such as clear water gets a low DN.

• Pixel size and detail:

• Smaller pixels → each pixel covers less ground → finer objects can be told apart. This is higher spatial resolution (the size of the smallest object a sensor can show).
• If you zoom in too far, you see only square pixels and no new detail (NCERT Fig. 6.12).

Pixel size across the satellite era:

Sensor (year)	Pixel size (spatial resolution)	Bands
Landsat-1 MSS, USA (launched 23 July 1972)	80 m	4 bands: green, red, two near-infrared [4]
IRS-1A LISS-I, India (launched 17 March 1988)	72.5 m	4 bands, visible and near-infrared [6]
IRS-1A LISS-II (1988)	36.25 m	4 bands, visible and near-infrared [6]
Resourcesat LISS-III	23.5 m	3 visible/near-infrared bands + 1 short-wave infrared band; swath 141 km [2]
National Geospatial Policy target for urban and rural areas (by 2030)	5–10 cm	— [7]

• Landsat-1: first named ERTS-1. It carried two sensors: the Return Beam Vidicon (RBV) and the Multispectral Scanner System (MSS) [4]. Its first image, taken on 25 July 1972, showed the Dallas–Fort Worth area of Texas, USA [4].
• IRS-1A: India's first home-built remote sensing satellite of the IRS series [6].
• It was launched from Baikonur in the then Soviet Union (now Kazakhstan) into a polar sun-synchronous orbit [6]. This is an orbit that crosses each place at the same local time every day.

• Its mission ended in July 1996, after 8 years and 4 months [6].

• Bhuvan: the geoportal (a website for maps and satellite data) of ISRO and its centre NRSC.

• Resourcesat LISS-III data (23.5 m) can be downloaded free, with a time lag of 2 years [2].
• LISS-III ortho products (24 m) cover India twice a year from October 2009 to May 2019 [2]. An ortho product is an image corrected so that distances are true to scale.

4. Two ways of taking information out of an image

• Digital image processing: a computer works on the DNs. It has three main steps: 1. Correction: removes errors caused by the atmosphere, the sensor and the ground geometry. 2. Enhancement: improves contrast so that features stand out. 3. Classification: groups pixels with similar DNs into classes such as forest, water or crop land.

• Visual image interpretation: a person identifies objects by eye, using the elements of interpretation (Section 7). It works on both photographs and printed digital images.

5. False colour composite (FCC)

• Composite: one colour picture made by joining three bands, one in each display colour (blue, green, red).
• False colour composite (FCC): each band is shown in a colour that does not match its real wavelength.
• Standard FCC assignment:

Band recorded	Colour shown on screen
Green	Blue
Red	Green
Near-infrared (NIR)	Red

• Why healthy vegetation looks red (cause → process → result):
• Cause: chlorophyll in leaves absorbs red light. The inner cell structure of a healthy leaf reflects NIR very strongly.
• Process: in standard FCC, the strong NIR signal is shown in red.

• Result: healthy vegetation appears red. A brighter red means denser, healthier plant cover.

• The same scheme is used worldwide: Landsat FCCs use the NIR, red and green bands (MSS bands 7, 5, 4 or 6, 5, 4), and vegetation appears red [4][5]. Greys and whites in such images show urban or rocky surfaces [5].

• Natural colour composite (NCC): shows bands in their true colours, so the picture looks like a normal photograph. Bhuvan shows some data as NCC by default and lets users switch to FCC [3].

6. Standard FCC colours of land features (NCERT Table 6.2)

Feature	Colour in standard FCC	Simple reason
Evergreen forest	Red to magenta	Dense leaves all year → very high NIR
Deciduous forest	Brown to red	Seasonal leaf fall → less NIR than evergreen
Cropped land	Bright red	Healthy, actively growing crop
Fallow land	Light blue to white	Bare soil, no vegetation, so no red
Clear water	Dark blue to black	Water absorbs NIR and most red light
Turbid water	Light blue	Suspended silt reflects more green light, shown as blue
Dense built-up area	Dark blue to bluish green	Concrete and roofs reflect green and red but little NIR
Sand and salt-affected land	Light blue to white	High reflection in all bands
Waterlogged land	Mottled black	Patches of standing water mixed with wet soil

• Indian use: turbid water (light blue) shows silt-laden rivers such as the Ganga and Brahmaputra in flood. Salt-affected land (light blue to white) appears in parts of Haryana, Punjab and the Rann of Kachchh in Gujarat.

7. Elements of visual interpretation

• Elements of visual interpretation: the clues an interpreter uses to identify objects by eye. They fall into two groups:
• Image characteristics: tone (or colour), texture, size, shape, shadow, pattern.
• Terrain characteristics: location, association.

#	Element	What it means	Example
1	Tone / colour	Shade of grey or colour, set by the energy the surface reflects	Dry, smooth surfaces look light; fresh water looks dark
2	Texture	How often the tone changes within a small area (smooth or rough look)	Dense housing looks fine; sugarcane looks coarse next to the fine texture of rice
3	Size	Actual or relative size of an object	Separates industrial complexes from houses
4	Shape	Outline or form of an object	Circular Sansad Bhawan; a railway's gentle curves vs a road's sharp bends
5	Shadow	Dark area cast by a tall object	Qutub Minar and water tanks
6	Pattern	Repeated arrangement of objects in space	Planned housing; orchards in rows
7	Association (with location)	What an object is usually found with or near	Schools near playgrounds; slums along drains and railway lines

• Tone:
• More reflected energy → lighter tone. Less reflected energy → darker tone.

• Dry sand and concrete look light. Deep, clear water looks dark.

• Texture:

• Sugarcane has tall, uneven stalks → frequent changes in tone → coarse texture.

• Rice is short and even → fine texture.

• Shape:

• Railways cannot climb steep slopes or turn sharply. So they run in long, gentle curves.

• Roads can take sharp bends.

• Shadow in image interpretation:

• It shows the height and outline of tall, thin objects that look like a dot from above, such as the Qutub Minar or a water tank.

• It is more useful in aerial photographs than in satellite images. Aerial photographs are taken from a lower height and have finer resolution, so shadows are clearer.

• Pattern in image interpretation:

• A regular grid of houses means a planned colony.

• Evenly spaced tree crowns mean an orchard, not a natural forest.

• Association in image interpretation:

• It uses the setting of an object, along with its location, to identify it.
• Example: an open ground next to a large building suggests a school. Thin, crowded houses along a drain or railway line suggest a slum.

Prelims Hooks

• All photographs are images, but not all images are photographs. Film works only in the visible and near-infrared range.
• In standard FCC: green → blue, red → green, NIR → red. So healthy vegetation looks red.
• Standard FCC colours: clear water dark blue to black; turbid water light blue; fallow land light blue to white; waterlogged land mottled black; cropped land bright red; evergreen forest red to magenta.
• DN (digital number): the average brightness of one ground cell stored in a pixel. Smaller pixel = higher spatial resolution.
• Of the elements of interpretation, shadow is more useful in aerial photographs than in satellite images.
• Location and association are terrain characteristics. Tone, texture, size, shape, shadow and pattern are image characteristics.
• Landsat-1 (ERTS-1): launched 23 July 1972; MSS sensor at 80 m in 4 bands [4].
• IRS-1A: launched 17 March 1988 from Baikonur; LISS-I at 72.5 m and LISS-II at 36.25 m [6].
• National Geospatial Policy: notified 28 December 2022. Target by 2030: 5–10 cm topographic mapping for urban and rural areas [7].
• Trap: texture of sugarcane is coarse and rice is fine. A railway shows gentle curves; a road shows sharp bends.

Mains Points

• From aerial photographs to satellite images to drones:
• Pixel size has fallen from 80 m (Landsat, 1972) to 23.5 m (LISS-III), and the 2030 target is 5–10 cm [4][2][7].

• Finer detail helps land records, urban planning and mapping of disaster damage (GS-III: infrastructure and disaster management).

• FCC-based land-use mapping:

• Red tones show crop and forest health.
• Light blue to white shows salt-affected and fallow land. Mottled black shows waterlogging.

• Together these help plan land reclamation in canal-irrigated areas such as Punjab and Haryana.

• Free Bhuvan data and open-data policy:

• Free LISS-III data on Bhuvan [2] and the open-access aim of the National Geospatial Policy 2022 [7] lower costs for states, researchers and start-ups.

• Human and machine skills both matter:

• Digital image processing classifies large areas fast.
• Visual clues such as association and pattern still identify slums, schools and orchards. This makes trained interpreters important for urban and agricultural planning.