5:15 PM - 7:15 PM
[AAS10-P02] Data Assimilation Experiment on Simultaneous Observations of Temperature/Water Vapor Lidar and Doppler Lidar
— Case Study of the Haeavy Rain event in Kyushu on June 24, 2024 —
Keywords:Data assimilation for QPF, Simultaneous Observations of Temperature/Water Vapor Lidar and Doppler Lidar
As part of the Cabinet Office BRIDGE project “Enhancing the Accuracy of Flood Forecasting Using Innovative Integrated Meteorological Data” (Principal Investigator: Prof. Yuji Sugihara, Kyushu University), efforts are being made to improve the forecasting accuracy of heavy rainfall events such as senjo-kousuitai and to advance river discharge prediction.
During the 2024 warm season, EKO INSTRUMENTS CO., LTD and Kyoto University installed a temperature/water vapor lidar and a Doppler lidar on Koshiki Islands, Kagoshima Prefecture, initiating continuous observations of temperature, water vapor mixing ratio, wind direction, and wind speed. The National Research Institute for Earth Science and Disaster Resilience (NIED) is responsible for investigating the impact of assimilating these observational data.
At around 8:00 AM on June 24, 2024, a linear precipitation system with a three-hour accumulated rainfall exceeding 100 mm was observed over the sea near Koshiki Islands, Kagoshima Prefecture. To evaluate the impact of lidar data assimilation on predicting this precipitation system, a cloud-resolving numerical simulation was conducted using the Cloud Resolving Storm Simulator (CReSS) with a horizontal resolution of 1 km, covering a 480 km × 464 km domain that includes the entire Kyushu region (vertical resolution: 50 layers, model top: 20.6 km).
The initial condition was set using the 6:00 AM output from the Japan Meteorological Agency's local model, with hourly forecast data provided as the boundary condition. The temperature, water vapor, wind direction, and wind speed vertical profiles at 6:20 AM, just before the precipitation started, were assimilated using 3DVAR. The three-hour accumulated rainfall from 6:00 AM to 9:00 AM was evaluated with radar observations (Fig. 1).
Fig1(left) shows observed three-hour accumulated rainfall from XRAIN (Ministry of Land, Infrastructure, Transport, and Tourism, MLIT), Fig.1 (center) shows forecast result with assimilation of water vapor and temperature. Finally, Fig. 1(right) indicates same as Fig.1 (center) but without any assimilation. By simultaneously assimilating water vapor and temperature, the prediction accuracy of the precipitation area was improved compared to the non-assimilated case.
Since the ultimate goal of this project is to enhance the accuracy of river discharge forecasting, evaluating watershed precipitation is essential. Therefore, the area-averaged precipitation over a 100 km² region centered on Koshiki Islands was compared: 25.12 mm (observation), 25.06 mm (with assimilation), and 8.43 mm (without assimilation). These results confirm that data assimilation significantly improved the accuracy of the area-averaged precipitation estimate.We plan to increase the number of case studies, continue investigating assimilation impacts, and develop polygon information for actual river basins to further verify the accuracy of watershed precipitation forecasting.
During the 2024 warm season, EKO INSTRUMENTS CO., LTD and Kyoto University installed a temperature/water vapor lidar and a Doppler lidar on Koshiki Islands, Kagoshima Prefecture, initiating continuous observations of temperature, water vapor mixing ratio, wind direction, and wind speed. The National Research Institute for Earth Science and Disaster Resilience (NIED) is responsible for investigating the impact of assimilating these observational data.
At around 8:00 AM on June 24, 2024, a linear precipitation system with a three-hour accumulated rainfall exceeding 100 mm was observed over the sea near Koshiki Islands, Kagoshima Prefecture. To evaluate the impact of lidar data assimilation on predicting this precipitation system, a cloud-resolving numerical simulation was conducted using the Cloud Resolving Storm Simulator (CReSS) with a horizontal resolution of 1 km, covering a 480 km × 464 km domain that includes the entire Kyushu region (vertical resolution: 50 layers, model top: 20.6 km).
The initial condition was set using the 6:00 AM output from the Japan Meteorological Agency's local model, with hourly forecast data provided as the boundary condition. The temperature, water vapor, wind direction, and wind speed vertical profiles at 6:20 AM, just before the precipitation started, were assimilated using 3DVAR. The three-hour accumulated rainfall from 6:00 AM to 9:00 AM was evaluated with radar observations (Fig. 1).
Fig1(left) shows observed three-hour accumulated rainfall from XRAIN (Ministry of Land, Infrastructure, Transport, and Tourism, MLIT), Fig.1 (center) shows forecast result with assimilation of water vapor and temperature. Finally, Fig. 1(right) indicates same as Fig.1 (center) but without any assimilation. By simultaneously assimilating water vapor and temperature, the prediction accuracy of the precipitation area was improved compared to the non-assimilated case.
Since the ultimate goal of this project is to enhance the accuracy of river discharge forecasting, evaluating watershed precipitation is essential. Therefore, the area-averaged precipitation over a 100 km² region centered on Koshiki Islands was compared: 25.12 mm (observation), 25.06 mm (with assimilation), and 8.43 mm (without assimilation). These results confirm that data assimilation significantly improved the accuracy of the area-averaged precipitation estimate.We plan to increase the number of case studies, continue investigating assimilation impacts, and develop polygon information for actual river basins to further verify the accuracy of watershed precipitation forecasting.