Mapping the Distribution of Exotic Mangrove Species in Shenzhen Bay Using Worldview-2 Imagery

Abstract

Shenzhen Bay is an important distribution area of mangrove, with Futian Reserve and Mai Po Reserve. Sonneratia (including Sonneratia caseolaris and Sonneratia apetala) are exotic mangrove species in Shenzhen Bay. Study on the Sonneratia potential impact on native mangrove species requires accurate and repeatable mapping of Sonneratia distribution. Several previous studies have mapped mangrove extent and species, but Sonneratia have been largely ignored. In this study, high resolution Worldview-2 (WV2) imagery was used in Shenzhen Bay Sonneratia mapping. Separability analyses with spectral and textural features were conducted based on Jeffries-Matusita (JM) distance. Results showed that Sonneratia caseolaris and Sonneratia apetala were inseparable in WV2 imagery, and therefore the two species were merged into a single species in the study. Maximum likelihood (ML) classifier, neural net (NN) classifier and support vector machine (SVM) classifier were applied to spectral and textural features, and six mangrove species classification results were obtained. Considering the six classification results together, the distribution of Sonneratia was mapped based on the criteria that, for each polygon, it was categorized as Sonneratia if and only if it was classified as Sonneratia in at least four classification results. The distribution results of Sonneratia showed an agreement with distribution characteristic based on field survey and past literatures. The producer’s accuracy was 79.24% and the user’s accuracy was 92.14%, which indicated great potential of using high-resolution multi-spectral data for distinguishing and mapping Sonneratia.