top of page
Session : 3 Remote sensing applications in Coastal hazards and Vulnerability assessment - Dr Tune Usha - Scientist G, Group Head, Coastal Hazards division - National Centre for Coastal Research , Ministry of Earthsciences - Government of India
Highlights of the Session - 3
• The session dealt with the explanation of various coastal hazards & disasters and also the ways to reduce the predicted risk caused by four tools – Remote sensing-To collect data, GIS- To analyse, Numerical Modelling-To predict, the entire work can be formulated on a web GIS platform.
• CLIMATE CHANGE – Climate Change and Coastal Hazard Risk Information System (CHRIS) was discussed with its methodology, Spatial/Aspatial database, Hazard Assessment, Exposure Analysis, Vulnerability Analysis, Risk Analysis, Risk Reduction
• FLOOD – Two integrated flood warning systems have been developed by NCCR and MOES namely iFLOWS-Chennai and iFLOWS-Mumbai. A video on iFLOWS-Mumbai was played for the better understanding of the system. Seven different modules related to flood warnings are discussed as follows. Data dissemination module, Flood module, Inundation module, Vulnerability module, Risk module, Dissemination module, Discussion support system since they help to predict the flood scenarios at ward level well in advance to carry out the mitigation works. Red Alert Atlas has also been developed. iFlOWS-Kolkatta is in pipeline.
• TSUNAMI - The countries preparedness on tsunami early warning centre, was established at INCOIS, Hyderabad and has been started to predict the ocean state. Also, she has detailed about NCCR Geo Surveyor App which has been developed for the onsite data collections and validations
• (IBIS-TC) Impact Based Information System for Tropical Cyclones has been formulated to predict the exact area to be affected, so as to plan the mitigation measures was discussed.
• (IMD/WCSSP) Impact based forecasting is an Indo-UK based collaboration work involved in risk based forecasting and high impact weather/seasonal events and to explore the potential for multi-hazard forecasting tool for tropical cyclones.
• THOONDIL is an app developed for the decision support system for fishermen safety and security. Around 15000 people are using it today. It is also useful at the administrative level to know how many fishermen and boats are into the sea.
• Some of the case studies and NCCR contributions are briefly discussed by Dr.Tune Usha related to the fields of coastal hazards and disasters.
Session : 4 An overview of deep learning and its application in remote sensing handled by Mrs Minakshi Kumar - Scientist SG, Indian Institute of Remote Sensing, ISRO - Government of India
Highlights of the Session - 4
• Artificial intelligence, machine Learning and Deep Learning were defined and the differences between machine learning and deep learning were highlighted.
• The basic components of Neural Network and that of biological neuron were compared for easy understanding.
• The different layers of Artificial Neural Network such as input, hidden layers and output were explained with a demonstration.
• Various functions like Activation function, sigmoid function, Threshold function, hyperbolic function and ReLU function were briefed.
• It was explained in short, how the errors in network has been back propagated by using cost and loss functions.
• Convolution Neural Network (CNN) and the steps involved, like Convolution layers, Convolution mask, Convolution padding and Stride, reflection process, ReLU layers and max pooling mechanism were narrated in detail.
• Fully Convolutional Networks (FCN), semantic and instance segmentation were stated and information related to U-Net, VGG-16, RESNET, Alex NET and various datasets like MNIST, CIFAR were provided.
• The softwares for machine and deep learning , like ImageNet, Pattern Net, NWPU45 and deep learning frameworks like Tensor flow, ENVI deep learning module, cognition and ERDAS deep learning module and the applications of deep learning in remote sensing like, Image Fusion, Image Registration, Scene Classification, Object Detection, Land Use/ Land cover classification, Segmentation and object based image analysis were accounted.
bottom of page