- Time of the year
- Cloud coverage
- Spatial resolution
- Spectral resolution
- Financial constraints
Time of the Year
Time is very important factor to decide whether to pick a satellite image or not. For example, if you want to study deciduous vegetation one must not go for image that is acquire by the sensor in autumn. Or if you want to map water holes and have procured summer time imagery it will be disastrous.
One should not misunderstand that autumn-time images are useless; instead these are very useful if evergreen vegetation are to be studied and mapped. In fact these images are helpful in differentiating deciduous and evergreen vegetation type.
Cloud is also important factor as these may mask the ground features partially or completely. That’s why rainy season images are not preferred (until unless one wants to do some specific study for this season). If you are lucky you can get cloud-free image in rainy season too. Sometimes winter & summer time images may also contain clouds. So always see preview of images before procuring them and avoid images having more than 10% cloud coverage.
Clouds are the main constraint in various types of land use/land cover studies for North-East India and Andmans & Nicobar Islands as most of the time these areas remain covered with clouds.
Spatial & Spectral Resolution
These resolutions are important in the sense- what kind of study you want to do? If you want satellite images for cartographic purpose then high resolution panchromatic images are good (e.g., CARTOSAT-1 & 2 images, IKONOS panchromatic images etc.). On the other hand for forest mapping, multispectral images are required which may be low in spatial resolution like Landsat ETM+ Images or IRS-1D LISS III images. In some cases when built-up area & vegetation types are mapped to be togather- then multispectral images with high spatial resolutions are used (e.g. IRS-P6 MX LISS IV images).
If you have selected a cloud free satellite image for a right season with right resolution then you must see you pocket also- whether you have sufficient money or not. Going for high resolution images are definitely going to cost much higher. So where a Landsat ETM+ or IRS-1D images can work don’t go for IKONOS multispectral images. Also processed images cost high. If you have good image processing software and skilled team then don’t buy processed images- as processing you can do in your lab itself. If you can spend some extra pennies then procuring georeferenced images from your vendor (like NRSA, Space Imaging etc.) can be a good idea as it will save your time and also you get data with acceptable accuracy.