The performance of satellite images in mapping aquacultures

Abstract

Monitoring human pressures is the first step in the management of natural ecosystems, as well as a method to evaluate the effectiveness of the applied conservation measures. In this context, five commercial satellite images (QuickBird bundle, SPOT-5 multispectral, Landsat 7 ETM+, RADARSAT SAR, and ENVISAT ASAR) with various spatial and spectral characteristics have been assessed for their ability to map mussel farms off a coast of northern Greece, where the intensity and uncontrolled expansion of aquacultures is a pressure to a nearby wetland of international importance. The ability to identify the mussel farms on the images from background open water and accurately map these features was tested separately for the two types of mussel farms (pole and long line) present in the study area. The influence of waves on the mussel farms' identification was also investigated. Results indicate that the optimum satellite sensor varied according to mussel farm type, and is not necessarily the one with the highest spatial resolution. Pole farms were identified in all images bearing a spatial resolution superior to 10 m, but were better located and delineated with a high-resolution QuickBird image. Long line farms, on the other hand, were indistinguishable by passive optical sensors, and could only be identified on active microwave images. In addition to this, the findings show that surface waves drastically deteriorate the identification of mussel farms on an ENVISAT image, thus influencing its usefulness for monitoring.