5:15 PM - 6:45 PM
[AAS08-P06] Development and demonstration experiments of cumulonimbus cloud hazard prediction information to support business continuity decisions of companies.
Keywords:lightning forecast, demostration experiments with semi-conductor factories
A project under the Cabinet Office's 'Research and Development Bridging Program (BRIDGE) for Society 5.0', titled 'Research and Development of Cumulonimbus Cloud Hazard Prediction Information and Deployment of Social Implementation Models' (Lead Institution: National Research Institute for Earth Science and Disaster Resilience, Representative: Shingo Shimizu), aims to develop cumulonimbus cloud hazard prediction information that contributes to corporate business continuity decisions based on the industrial needs. This project focuses on various hazards associated with cumulonimbus clouds (primarily lightning) and creates and integrates three types of forecast information: imminent forecasts, a few hours ahead forecasts, and half-day ahead forecasts. The research project, running from October 2023 to March 2025, involves conducting demonstration experiments to validate the effectiveness of the developed 'cumulonimbus cloud hazard prediction information'. From an economic security perspective, the project focuses on semiconductor factories, referred to as the 'rice of industry', and plans to conduct demonstration experiments in the Kyushu region, which has the highest semiconductor shipment value share in Japan (21.6%), during the warm season of 2024.
Interviews with several semiconductor factories revealed that, while hardware measures such as Uninterruptible Power Supplies (UPS) have limitations due to cost, reducing losses through process management based on accurate forecast information is a challenge. Furthermore, it was found that losses are significant in the pre-manufacturing stages of semiconductor wafers, hence the investigation into the necessary lead time for operational rate adjustments in these stages. It was discovered that forecasts up to a few hours ahead for operational rate adjustments and up to half a day ahead for management planning are needed. For 12-hour ahead forecasts, the Japan Meteorological Agency's mesoscale model guidance on lightning probability (Figure 1 left) was utilized, and efforts to verify forecast accuracy began. Similarly, for 2-hour ahead lightning forecasts, the entire Kyushu region is predicted in real-time up to 2 hours ahead at a resolution of 1km. Initial value improvements were achieved by assimilating the radial wind and reflectivity from the Ministry of Land, Infrastructure, Transport, and Tourism's XRAIN using the 3DVAR+IAU method, and McCaul et al., 2009 (Wea, Forecasting) proposed lightning index was introduced to CReSS (Figure 1 right). As verification, lightning data from Franklin Japan was used. For the case of a large-scale power outage caused by lightning in Kumamoto city between 04:30 and 06:00 (all times JST) on July 5, 2022, mesoscale guidance using 18:00 on the 4th as the initial value predicted more than 12% lightning probability over a wide area of Kumamoto Prefecture, while CReSS predictions using 04:00 as the initial value could predict high lightning indices in areas concentrated with lightning locations, demonstrating the possibility of narrowing down lightning hazards in imminent forecasts. Future efforts will focus on improving forecast accuracy and enhancing process management support information.
Interviews with several semiconductor factories revealed that, while hardware measures such as Uninterruptible Power Supplies (UPS) have limitations due to cost, reducing losses through process management based on accurate forecast information is a challenge. Furthermore, it was found that losses are significant in the pre-manufacturing stages of semiconductor wafers, hence the investigation into the necessary lead time for operational rate adjustments in these stages. It was discovered that forecasts up to a few hours ahead for operational rate adjustments and up to half a day ahead for management planning are needed. For 12-hour ahead forecasts, the Japan Meteorological Agency's mesoscale model guidance on lightning probability (Figure 1 left) was utilized, and efforts to verify forecast accuracy began. Similarly, for 2-hour ahead lightning forecasts, the entire Kyushu region is predicted in real-time up to 2 hours ahead at a resolution of 1km. Initial value improvements were achieved by assimilating the radial wind and reflectivity from the Ministry of Land, Infrastructure, Transport, and Tourism's XRAIN using the 3DVAR+IAU method, and McCaul et al., 2009 (Wea, Forecasting) proposed lightning index was introduced to CReSS (Figure 1 right). As verification, lightning data from Franklin Japan was used. For the case of a large-scale power outage caused by lightning in Kumamoto city between 04:30 and 06:00 (all times JST) on July 5, 2022, mesoscale guidance using 18:00 on the 4th as the initial value predicted more than 12% lightning probability over a wide area of Kumamoto Prefecture, while CReSS predictions using 04:00 as the initial value could predict high lightning indices in areas concentrated with lightning locations, demonstrating the possibility of narrowing down lightning hazards in imminent forecasts. Future efforts will focus on improving forecast accuracy and enhancing process management support information.