One of the most devastating effects caused by climate change has been the increased intensity and frequency of extreme rainfall. This has made the application of AIoT innovative technology in developing a more intelligent disaster prevention warning system an important work for all urban areas. Using the Taiwan Computing Cloud (TWCC), National Center for High-performance Computing (NCHC) of NARLabs is able to integrate IoT real-time water regime forecasts, high-efficiency physical model computations for water regime trends, as well as big data related to historical floods and rainfall. NCHC has successfully developed the “Smart Flood-Proofing System” to process water regime data from all counties and cities in Taiwan; a system capable of generating water regime changes and forecasts for the coming hour in just three minutes.
The “Smart Flood-Proofing System” brings the benefits of integrating traditional hydraulic engineering and forward-looking technology to the fore. In flood prevention management, every second counts, and this system excels when it comes to saving time. The NCHC uses TWCC to access water monitoring data for each county and city on the IoT, executing high-efficiency physical model computations which, in conjunction with AI-calibrated historical big data, have enabled the system to rapidly extend perceptual information pROVided by the IoT to water regime predictions. It compiles and analyzes the data, intelligently converting the information into effective disaster prevention actions.
With its petaFLOP capacity, TWCC has greatly impROVed forecasting capabilities in Taiwan by reducing the time needed to conduct predictions, and enabling island-wide, large-scale water regime forecasts. In terms of forecast efficiency, it can finish generating spatial high-resolution, next-hour water regime forecasts in just three minutes. As for the requirements placed on large-scale computational resources, TWCC, with its ability to rapidly reach virtual machines, can quickly deploy its forecasts to computers in each county and city. TWCC can also tailor forecast frequencies (4, 6 or 24 forecasts per day) and durations (3, 8 or 12 hours into the future) to meet the needs of local governments. With such information, governments can make comprehensive disaster prevention plans well in advance.
