Accurate tsunami prediction is ofimportance for many coastalcities to anticipate and mitigate any disastrous impacts oftsunami oceanwaves. These disastrous wavesarecommonly induced by thetectonicearthquake or landslides that impacts to the oceanwater body. Thecollision ofthesetsunami waves to the urban areas can lead catastrophic economic infrastructure damages and human casualties. Physically based or numerical modeling for tsunami typically provides offline and imprecise tsunami prediction due to time delays on acquiring new oceanwave observation after the earthquake and intensivetsunamimodelcomputation that requests much time. It often occurs that the modeling prediction resultsareavailableconsiderably longtime duration after the tsunamievent has elapsed. Consequently, some national tsunami prediction centers used to initiate some warnings of tsunami event at certain coastalarea, but the actual disaster will never happen. These warnings could come up fromthe rough estimation with a simplified calculation and sometimes withoutsufficient variables, supported dataand computation power. With therecentadvancements ofinformation technologies, many contributionscan be offered towards the improvements ofthe current tsunami early warning systems. Some of these promising IT developments include the sub fields of data science, internet of things, parallel computing, immersive dashboard, intelligent data analytics, deep learning and so forth. This research focuses on developing immersive dashboard and intelligent data analytics for an accurate tsunami early warning system, that can be divided into four work packages. The first research task will be to build efficient and accurate physically based or numerical tsunami predictive models, which the parameter settings have been optimized for best characterizing the tsunami wave behavior and its disastrous impact risk analysis on the selected coastalarea, like inAceh, Sunda Strait, Palu or in Japan. Some existing data, like bathymetry, coastal area map, boundary conditions can be utilized in the tsunami modelling and the new incoming wave data can be collected and incorporated into the tsunami model predictions through data assimilation. Secondly, the tsunami models will be deployed and run on parallel, cluster orcloud computing infrastructureand they should beready for generating fast tsunami predictions with impacted coastalarea maps based on therisk analysis oncetheearthquake or landslide occurs. The parallelcomputing usingGPU processors will beexploited with CUDA programming and the usage of cloud technology in this effort will be supported by Amazon Web Services (AWS) Cloud. The third researchwork package will be to construct data-driven tsunami predictive models usingAI and machine learning to be integrated with the physically based or numeral tsunami models to enhance the prediction accuracy and capabilities. In this task of data-driven modelling, we will employ the parallelcomputing using GPU processors and cloud technology usingAWS Cloud in provisioning the accurate and fast tsunami model predictions. The last research work aims at developing immersive dashboard and intelligent data analytics to support decision making and for more accurate tsunamiearlywarning system. The immersive dashboard and intelligent analytics here are not only offering the interactive visualizations of physically based and data-driven model predictions but also provide tsunami risk map analysis of impacted coastalarea to direct the people rightly routing to the safety areas. One of the expected research deliverables is to provide immersive dashboard and intelligent data analytics running in tsunami control room, personal computer, or mobile systems, which can be used as an efficient and effective tool for tsunami decision makers fromthe national tsunami prediction centers, localmunicipalitiesand the peopleliving in variouscoastalareas with tsunamirisks.

AWS, Tsunami, Earthquake

• Samuel Mahatmaputra Tedjojuwono, S.Kom., M.Info.Tech

• Dr. Michael Baskara Laksana Adi Siek, S.Si., S.Kom., M.Sc.