This article detailed a study in which researchers used a series of Landsat MSS and Landsat TM images from between 1973-1998 to assess the impact of land use/cover change on temperature change and air quality in Atlanta. The study indicated disturbing trends of deforestation and urban sprawl, but it also yielded worth-while information about the use of satellite photograph interpretation.
- Provides an excellent overview of a region
- Allows a retrospective view
- Can provide highly accurate results
- Relies on interpretation
- Limited by resolution, image quality, atmospheric haze, and contrast
How to (Steps):
- Geometric Rectification- align landmarks (necessary for analysis over time)
- Radiometric Normalization (RRN)- necessary when images were taken by different sensors, to standardize radiometry, or contrast within an image
- Design of Classification Scheme- decide on classification categories ex. cultivated land or grassland
- Image Classification- divide the photograph into "clusters", spatial areas with the same characteristics, then decide into which category they should be placed
- Spatial Reclassification- an effort to reduce errors in classification involving techniques such as having a human check portions of computer-classified work
- Accuracy Assessment- compare classification results to what the ground areas truly are as revealed by aerial photographs
Accuracy of Method:
The study results stated that user accuracy was over 80% in 5 out of 6 land-type classification categories. The accuracy for the sixth was in the high seventies. Landsat TM images which have a higher resolution yielded slightly more accurate results than those from the Landsat MSS images especially for high and low density urban areas and forests. User accuracy for these categories was 96%, 98%, and 98% respectively.