Identification and spatial mapping of mangrove species using SAM classification a case study from Aroor, Alappuzha District Kerala

Abstract

It gives an accuracy (overall 96%) by discriminating mangroves from non-mangrove species and obtained an overall accuracy of 85% in mangrove by supervised classification in species differentiation [2]. Verheyden. et al; (2002) experienced that the number of objects will be easily differentiated by higher spatial resolution data and similar profile of spectral signature, plays an important role in distinguishing mangrove species and any other vegetation that nearer to the mangroves have an essential role in the discrimination process [8]. Myint. et al; (2008) used an object-oriented approach with high resolution data to differentiate three mangrove species in Trang Province, Thailand. The result reveals the significance of high resolution data along with hard classification approaches [6]. Wang et al; (2009) acknowledged hyperspectral data have great probability for distinguishing mangrove canopies of various mangrove species composition and wave channels at 780, 790, 800, 1480, 1530, and 1550nm were recognized as the most suitable bands for mangrove species categorization [10]. Chakravortty; (2013) conducted study in Sunderban-West Bengal. From The Hyperion data, spectral profile of 7 species are taken and then library of spectra created [1]. Held et al; (2003) used CASI data for the exploration of mangrove species, hierarchical neural network classification and maximum likelihood classification performed. Based on structural information general mangrove zones are separated and then species are extracted using spectral differences [3]. Rodriguez et al; (2004) performed classification of wetland vegetation by considering intensity-hue-and saturation, Principal Component analysis and Normalized Vegetation Index to facilitate species discrimination of mangroves [7]. Vidyasagaran et al. (2014) studied the extent of mangroves in Kerala and its diversity. According to the estimation, the extent of mangroves of Kerala is 2502 ha and total of 15 pure mangroves species were recorded [9]. From the literatures it is understood that classification by PCA and SAM gives more accurate results when compared to other supervised classification techniques. Hence for the present study SAM classification technique has been adopted. III. STUDY AREA The study area (Mangrove patch) covers an extent of 0.0185 sq.km falls between latitude 9° 51ʹ 45ʺ-9° 51ʹ 32ʺ N and longitude 76° 19ʹ 02-76° 18ʹ 59ʺ E. Figure 3.1 showing the study area of the mangrove patch at Aroor, Alappuzha district. Alappuzha coastal area is rich with various types of mangrove habitats. Marshy areas and brackish water are very good system to grow wetland vegetation like mangroves in this area. Study area situated in the coastline of Vembanad backwaters near Aroor in Alappuzha district. Vembanad Lake is also known as Vembanad kayal and it is in number one position in India in terms of its length. This lake is the biggest lake in Kerala. Fig. 3.1 Study area of mangrove patch at Alappuzha IV. MATERIALS AND METHODS Sentinel-2B is a European optical imaging satellite with wide swath high-resolution multispectral imager with 13 spectral bands of spectral range 443-2190 nm and 10 m spatial resolution data used for the study. The data contained an applied radiometric and geometric corrections (including ortho-rectification and spatial registration). ASD HandHeld 2 Spectroradiometer used for spectra collection of mangrove species. Its wavelength ranges from 325-1075 nm with an accuracy of 1nm, spectral resolution of <3 nm at a wavelength of 700 nm. Environment for Visualizing Images (ENVI) is an image processing software fully integrated with ArcGIS. ENVI software is used for the processing and analysis of satellite imagery of the area under studied. ArcGIS software platform from ESRI using for making maps with spatial information. ArcGIS provides mapping tools and it has a unique capability for the manipulation of data, data editing, data analysis and mapping.