top of page
Session : 5 Ocean color algorithms and applications handled by Dr. SeunghyunSon , NOAA/NESDIS Star/Colorado State University - USA
Highlights of the Session - 5
• The importance of satellite data for Oceanography was discussed with appropriate example of global distribution of chlorophyll.
• The use of satellite data in biological oceanography and significance of ocean color in identification of harmful algal bloom (red tide), floating green algae bloom, cyanobacterium bloom, Coccolithosphores bloom was narrated.
• The history of Satellite Ocean Color Sensors and also about current satellite ocean color sensors. Monthly variation in Chlorophyll -a was also discussed with CZCS images (1978-1986).
• Application of ocean color remote sensing such as Ocean Productivity and Global Carbon Cycle, Ocean Acidification, Marine Biodiversity and Function, Flow of material through the marine food webs, implications for marine resources, Marine pollution and water quality, Validation and Improvement of Earth System and Ocean, Biogeochemical models, Data Assimilation to improve model performance, assessing impact and adaptation of marine ecosystem to climate change were addressed.
• The validation of monthly variation and seasonal variation in chlorophyll-a, TSS using MODIS data were discussed with a case study. Time series for Interannual Variability of Chlorophyll-a and Decadal Changes in Chlorophyll-a identification using MODIS and SeaWiFs data were also shortly explained.
• The various Primary production models like, Vertically Resolved PP model, Chesapeake Bay Production Model were explained in detail.
• New Inverse Semi-Analytical Kd (490) Model for Turbid Waters, New Blended Kd (PAR) Conversion Algorithm were elaborated.
• He illustrated the New OCI Chl-a algorithm for VIIRS, Diurnal Changes of Ocean Biogeochemistry from GOCI with an example. Human-induced Environmental Changes were also analyzed with time series for changes in SST, Chl-a, TSS from MODIS data.
• The information regarding interpretation of Ocean dump site, Typhoon-induced Biogeochemical Changes, Changes in Phytoplankton Community Structure, changes in Water Properties in inland Lakes were unfolded with examples.
• Various algorithms for Turbidity, Chl-a, Secchi Depth, and explained Phytoplankton size composition (PSC) and changes in PSC responding to ocean warming.
• The future Satellite Ocean Color Sensors which are scheduled for launch and depicted the information about PACE (plankton, Aerosol, Cloud, Ocean ecosystem), SBG (Surface Biology and Geology), Geosynchronous Littoral Imaging & Monitoring Radiometer (GLIMR), Geostationary Extended Observations (GeoXO), Small Size Satellites (CubeSat) and their future applications.
Session : 6 Space technology for Flood and Cyclone Management - Indian Experience handled by Dr S V Shiva Prasad Sharma, Scientist Engineer/SF, NRSC - ISRO, Hyderabad
Highlights of the Session - 6
• The terms Hazard, Vulnerability and risk were defined and India’s Disaster profile including earthquakes, landslides, floods, drought, Tsunami, cyclones, landslides and average annual losses was briefed.
• Insight into India’s Disaster reduction Strategy including Disaster Management Act, Risk Financing and Insurance. People are educated about the disaster and given insurance if they shift to low disaster prone areas, Vulnerability reduction, Fund ,Relief and Rehabilitation are discussed.
• The proactive ways of dealing with disaster management like Indigenous coping mechanism, Community based disaster management, National Response mechanism were explained.
• The role of various committees of Disaster management in India like Cabinet Committee, High Level Committee, Ministry of Home Affairs, North East council, State Disaster Management authorities, Community services, early warning mechanism, State Police and disaster response force were highlighted.
• The association of ISRO with DMS of our country in Monitoring and damage assessment, National Database for Emergency management, VSAT based VPN, Emergency Communication, Early warning systems were elaborated.
• India’s satellites which are scheduled for future were named and their applications were listed.
• Triggering mechanism of Decision Support Centre was addressed and the role of space technology in Disaster Risk Reduction like Strategy for Disaster reduction, Sendai Framework, early warning, Risk information, Impact Assessment, Preparedness, Emergency communication were explained in detail.
• Various phases of disaster management cycle like preparedness, response, mitigation, R&D, capacity building was explained. The use of space technology in disaster management was explained well with examples of Uttarakhand floods of 2021, Cyclone Yass, Heavy rain in Odisha, Beirut chemical Disaster.
• The disaster assessment using optical and SAR sensor was differentiated with a suitable example. The flood mapping using optical and microwave images was also distinguished with an illustration.
• The case studies for Uttarakhand floods, Urban flood modelling, Tamilnadu floods 2020, Hyderabad rains 2020, assam floods, Mumbai heavy rains 2017, heavy rain of Tamilnadu 2015 were addressed and the mechanisms like Digital elevation model, continuous flood mapping were also explained.
• The various application of Bhuvan portal was given an insight and few examples like Assam hazard zonation map, HUDHUD cyclone, Phailin cyclone were illustrated.
Session 7 : Machine and deep learning for remote sensing image classification and detection - Our Experience handled by Dr Anil Kumar, Head - Scientist Engineer/SG, IIRS -ISRO, Dehradun
Highlights of the Session - 7
• Application of AI that provides systems the ability to automatically learn, with the help of ML Classifications (grouping of pixels), Prediction (by regression mode) can be done.
• ML algorithms – SUM, Random Forest, Decision tree, C-means, MLC, SVM, ANN, etc were informed to the participants
• Different aspects while handling Remote Sensing images are briefly discussed with mixed pixels, which can be handled by Fuzzy set theory and the noisy and isolated pixels, which can be handled by contextual approach
• Advantage of ML to handle non-linearity between classes by kernel approach and for spectral overlap between two class pixels, temporal data are to be used were conveyed.
• Paddy field transplanted at different dates was discussed in bi-sensor and multi-sensor approach. CNN based model; integrated CNN-LSTM(RNN) based model was explained.
• Optimum training samples are to be used in DL based Classification. Temporal change and how to use this information was discussed with the example of sunflower crop growth-Shahabad area temporal changes.
• Importance of temporal images, heterogeneity within class were briefed with specific examples. ML to DL, DL Classification algorithms- CNN, RNN and their hybrid versions were highlighted with outputs of various classifiers were elaborated.
• GAN (Generative Adversarial Network) based DL to remove cloud from optical temporal data- its structure and model were discussed.
• Detection based DL algorithms are R-CNN, Fast R-CNN, Faster R-CNN, SSD, Mask R-CNN, etc. SOME were discussed with examples.
bottom of page