Current Science

R. Gayathri ¹, Prasad K. Bhaskaran ¹, Felix Jose ²

Affiliations
1 Department of Ocean Engineering and Naval Architecture, Indian Institute of Technology, Kharagpur 721 302, IN
2 Department of Marine and Ecological Sciences, Florida Gulf Coast University, Fort Myers, FL 33965, US

Abstract

Coastal inundation is the flooding of coastal zone resulting from increased river discharge, spring tides, severe storms, or generation of powerful waves from tectonic activity (tsunami). This article discusses the critical factors that contribute to coastal inundation. Among the probable factors that cause coastal flooding and destruction, storm surge is the most frequent, and hence this article provides a detailed evaluation of the progress made in storm inundation research. Recent advances in coastal inundation modelling include efforts to understand the nonlinear dynamic interaction of near-shore waves, wind and atmospheric pressure with still water sea level and coastal currents, and their combined effects on storm surge along the coast and interaction with coastal morphology. An advanced storm-surge model comprises different modules, viz. an atmospheric component, and two ocean components for surge and wave simulations; these modules are coupled with each other. The nesting of regional coastal model with an ocean-wide model captures the far-field boundary forcing of extreme events that usually originate from the warm open ocean. Even though significant advancements reported on the efficiency and accuracy of storm surge and inundation prediction, further studies are required to understand the nonlinear interaction of storm surge with coastal landforms and their vegetation (land cover). In the context of rising sea level, increased tropical cyclone activity and rapid shoreline change, it is pertinent to evaluate the future flooding risk associated with landfall of tropical cyclones in densely populated coastal cities.

Keywords

Coastal Inundation, Coupled Models, Storm Surge, Tropical Cyclones.

Full Text

   

C. Sherin Shibi ¹, R. Gayathri ¹

Affiliations
1 Department of Electronics and Communication Engineering, Sri Venkateswara College of Engineering, Sriperumbudur 602 117, IN

Abstract

Deep learning approaches have received major interest in the field of remote sensing. Hyperspectral imaging has rich data that are distributed in multi-dimensions. It is challenging to apply deep learning algorithms due to the limited amount of labelled data. So, unsupervised feature extraction approaches are used to overcome this limitation. In this study, we propose an unsupervised feature learning approach using deep belief network (DBN). In the proposed framework, the input hyperspectral image is segmented using entropy rate superpixel segmentation and filtered by domain transform recursive filter which extracts spatial and spectral information effectively. Then the features are learned by improved DBN. In the traditional methods, DBN is stacked with restricted Boltzmann machine (RBM) which is suitable for only binary value data. In the proposed framework, we used Gaussian–Bernoulli RBM which was constructed for real value data such as images. The experiments were carried out using Pavia University dataset. The results show that the proposed network has good performance in terms of classification accuracy and computation time.

Keywords

Deep Belief Network, Hyperspectral Image, Remote Sensing, Spatial–Spectral Classification, Superpixel Segmentation.

Full Text