Detecting and monitoring aquacultural patterns through multitemporal SAR imagery analysis

The inventory and monitoring of aquaculture areas are essential tools for decision-making at a governmental level in developing countries. With the use of satellite imagery, these tasks can be performed in an accurate, rapid and objective way. This approach is also economically viable, as the worth of aquaculture far outweighs its cost. This paper describes a methodology for identifying and monitoring shrimp farms by means of multi-temporal satellite SAR data. SAR offer all-weather capabilities, an important characteristic since shrimp farms exist in tropical and sub-tropical areas. Moreover, the backscatter effect created by the dykes surrounding the ponds produces a typical pattern which allows the interpreter to distinguish them from other types of water-covered surfaces. However, the presence of speckle noise limits the interpretability of SAR imagery. To increase it, a multi-temporal set of four scenes covering the study area was processed by using a method that enhances time-invariant spatial features and reduces speckle without compromising the geometrical resolution of the images. The enhanced SAR imagery has proved to be valuable in identifying shrimp farm patterns with a field-tested accuracy of more than 90 percent. The methodology reported in this study has been tested with the ground truth obtained under operative conditions in Sri Lanka, thanks to the support of the FAO TCP/SRL/6712 project.