New spectral metrics for mangrove forest identification

ABSTRACT

This study proposed three spectral metrics, namely spectral match degree (SMD), normalized difference mangrove index (NDMI) and shortwave infrared absorption depth (SIAD), to enhance the separability between mangrove forests and terrestrial vegetation in remote sensing imagery. The Landsat 8 OLI image of an interest area in Beilunhekou National Nature Reserve was used to test the spectral metrics. The derived spectral metrics and raw band reflectance data were classified using a support vector machine classifier. Mangrove forest maps were then identified from the classified images. Identification accuracies were compared and evaluated by determining the user’s accuracy (UA), producer’s accuracy (PA), overall accuracy (OA) and by conducting McNemar’s test. Results showed that the use of spectral metrics (UA = 85%, PA = 94%, OA = 95%) outperformed the use of raw band reflectance data (UA = 72%, PA = 82%, OA = 90%). McNemar’s test confirmed that the spectral metrics were significantly better than the raw band reflectance data (Z = 4.63, p < 0.05). Therefore, the proposed spectral metrics could improve the accuracy of mangrove forest identification.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by the Forestry Nonprofit Industry Scientific Research Special Project ‘The research of ecosystem service and evaluation techniques of coastal wetlands, China’ [Grant number 201404305] and the Scientific Research Foundation for Newly Introduced High-End Talents of Shenzhen University.