Coastal landscape management is becoming a critical issue around the world, particularly where the conservation of coastal environments and anthropic pressures compete for sustainable use of natural resources. Recent advances in remote sensing provide opportunities to map and classify biophysical parameters at scales and resolutions relevant to the management issue. Airborne hyperspectral images provide a wide range of quantitative data that allow obtain synoptic information on vegetation structure and composition, water quality, sediment composition and distribution. Research objective aims at providing spectral libraries as benchmark against which future hyperspectral images classification could give ecological variability quantification. In order to get structural and functional indicators useful in coastal dynamics analysis at habitat level an accurate field survey has to be carried out and a series of different biophysical parameters have to be achieved. The purposes of the field observations are twofold. First, field observations allowed to evaluate ecological variability useful to validate hyperspectral image classification. Secondly, field reconnaissance allows to build spectral libraries of vegetation and sediment involved in optical interpretation of the shallow water body and of the emerged sandy beach.
Results/Conclusions:
The constructions of different spectral libraries, collected in two different coastal sandy beaches during three years (2009-2010-2011) provided ecological variability quantification on the hyperspectral images. Vegetation image based analysis in submerged areas, provides percent cover of vegetation. By the integration of the shallow coastal seabed spectral library, vegetation composition is obtained and divided in two classes: P. oceanica and seaweeds. In the beach dune habitat, image based analysis lead to the identification of three vegetation structural classes that are defined as trees, bushes and grasses. Spectral library of species and different ecological succession in heterogeneous areas provided the possibility to classify each structural class in functional level. Sand occurrence has been identified on the base of the inherent reflectance properties in image processing where sand spectral library allowed to better classify its grain size distribution: coarse and fine sands. Still uncertainties remains about effects of mineralogy and moisture content that has to be further investigated. High water column transparency determines optical response in which seabed influences water reflectance properties. Integration of spectral library and field waters quality measurements allowed training and validation of bio-optical modelling, in order to improve accuracy of mapping and classification of [Chl], [TSM], [CDOM].