Sunday, April 22, 2007

‘Reading’ a Satellite Image

  • Please keep visiting for updates,soon more topics will be covered.
  • Please post your comments if you are looking for a particular topic on Remote Sensing, GIS & GPS we will try to include them.


Satellite images are like books. To read and interpret these, we need to learn the language in which these are written. The more we practice the more conversant we become in interpreting images. Experience is the key for digging more and more information. Visual image interpretation comes with time and with understanding of elements which help to ‘decode’ the language of images. These elements of visual image interpretation are as follows:

  • Shape
  • Size
  • Tone
  • Texture
  • Pattern
  • Shadow
  • Location
  • Association
  • Shape

    Shape of ground objects and features is one of the most important elements to identify them. For example- Road and rail both appear like lines but road has sharp curves while railway track has smooth curves. Natural water bodies are with irregular shape while most of the man-made water bodies have definite shape (rectangular, circular etc.). Same is true for natural drainage and man-made canal.


    Sizes are important in identifying many objects, e.g. on the basis of size we can differentiate between trees and bushes. Length, width, height (i.e. dimensions) and area provide clue for many objects. We can interpret in terms of absolute and relative sizes. When we talk about absolute size of a feature we go for exact dimensions while in relative size we look in terms of smaller or bigger. In high resolution satellite image one can easily assume size of a building by comparing it with size of a car parked next to it.


    It refers to reflectance of features which we see in the form of tone of colours in a satellite image. As we discussed earlier different objects appear different in an image depending upon their spectral signatures. Hence this element gives firm evidence in identification of many features. On the basis of tone we can differentiate between plant species, age of plants, shallow & deep water bodies, dry & wet soil, crop types etc.


    Texture of ground feature refers to how tones vary in the image. In other terms, how frequently tone varies. Textures are often said to be coarse and fine/smooth e.g. young plants generally have smooth texture while mature ones appear coarse-textured. Crops have smoother texture than vegetation.
    Scale or resolution of satellite image should be considered while interpretation is being done on the basis of texture because low resolution images will show most of the features smooth-textured and after certain scale we can not differentiate between objects only on it.


    Arrangement of objects also helps in image interpretation. Most of the man-made features show definite pattern hence these can easily be differentiated from natural objects. For example, plantations have definite arrangement of trees with well defined pattern while natural vegetation will not have uniform pattern.


    Shadows are both good and bad for image interpretation. These are good for studying relief and identifying hilly regions. Tall objects (like clock tower, overhead water tanks etc.) which are sometimes difficult to locate, can easily identified with the help of their shadows.
    Shadows are bad because these mask most of the features coming in their zone. These particularly create problems in hilly terrain where hill shadows hide information about vegetation and many other features.


    Site of presence of an object helps in avoiding misinterpreting it as other same looking object. For example- tones of two vegetation species may appear similar in an image but their geographical location can help to identify them correctly.


    While interpreting an image we should always consider how a particular feature is associated with its surroundings. For example- one can identify a village by- its small number of settlements, connecting roads, agriculture land in its adjoining area and often a water body.

    Whether we are experts or novice- in image interpretation we should always consider a number of elements before concluding about features. Only considering one element may lead to erroneous identification of objects.


    Karim Hussein said...

    in an application to estimate the rate of desertification ,what would be the appropriate satallite images you would see? and what would be the time interval required?

    tom Barnes said...

    Very interesting article, i wonder if you found some similar idea about satellite image interpretation in here