Mangrove species and stand mapping in Gazi bay (Kenya) using quickbird satellite imagery

Abstract

This paper presents an automated method for mangrove stand recognition (delineation and labeling) and species mapping based on fuzzy per‐pixel classification techniques of a QuickBird satellite image. The four dominant mangrove species in Gazi Bay (Kenya) are mapped with an overall accuracy of 72 percent, where the two socio‐economically most important species are mapped with user accuracies above 85 percent. Mangrove stand maps were compared to visual delineations done by an expert interpreter and the quality was based on the quantity of overlap one has with the other. An overall correspondence up to 86 percent was achieved.