Big Data Analytics Laboratory



Decision Support System for Localized Torrential Rain

Disasters of localized torrential rain have increased dramatically and become a serious issue because of global warming and heat-island phenomena. As a practical application of our real-space information analytics technology, a decision support system for localized torrential rain is being developed. This system uses phased-array weather radar (PAWR) installed in Osaka and in Kobe for early detection of vor-texes that indicate development of cumulonim-bus clouds (localized torrential rain baby cells). It then predicts areas on the ground where a rainfall exceeding 50 mm/h will occur within 30 minutes, visualized on a digital map (localized torrential rain early detection). It also sends warnings to pre-registered e-mail addresses, indicating areas such as water catchments that rainfall will flow into, underpasses and other areas susceptible to flood damage, before the localized torrential rain occurs. We are currently conducting a field test of this system in cooperation with the Kobe city office. We also develop AI technology for predicting various risks co-occurring with localized torrential rain such as traffic hazards. Our R&D scope also includes behavior support for avoiding the predicted risks, such as map navigation for safe route guidance.

Decision support system for localized torrential rain

Unauthorized copying and replication of the contents of this site are strictly prohibited.

Related Topic

Integrated Analysis of Environmental and Social Big Data on IoT

To make social systems more efficient and optimized utilizing IoT data, it is important to organically link data collection, analysis, and behavior support on the IoT. We are developing a data mining technology, called Event Data Warehouse, that will gather and integrate heterogeneous sensing data from IoT, and extract event information representing environmental phenomena or people's behaviors. This will be used to discover and predict spatial, temporal, and thematic associations among those data. We are also developing user interface and interaction technology for effective behavior support using various IoT devices in conjunc-tion with the data mining. Our research aims at optimizing information visualization, presentation, and user interactions based on extrinsic factors such as the types and situations of behavior (working, driving, walking, interacting, etc.), as well as intrinsic factors such as attentiveness and psychological stress.

Safe route navigation based on torrential rain x traffic data

Disaster risk analysis based on torrential rain x SNS data

in collaboration with Social Big Data Research and Collaboration Center

Unauthorized copying and replication of the contents of this site are strictly prohibited.

Multi-scale, Multi-modal Air Pollution Data Synthesis and Prediction Technology

We are developing a scalable method for air pollution prediction using atmospheric simulation models, which integrates analysis of multi-scale, multi-modal atmospheric data captured by satellite sensors, Lidar and other means at scales from global down to the level of towns and roadways. This is a world-leading technology realizing air pollution predictions for local living areas, from several hours to several days in advance, by taking into consideration the widespread cross-border pollution in Asia. We are also developing applications that provide pinpoint forcasts of adverse affects on health from the prediction results, as well as behavior support for prevention and mititgation. In the future, we plan to integrate data from mobile and personal sensors in order to provide even more detailed predictions.