13:45 〜 14:00
[AAS02-01] Exploring global 1-km hydrological simulations using the Integrated Land Simulator
★Invited Papers
キーワード：全球陸域水循環、超高解像度陸域水循環シミュレーション、統合陸域シミュレータ、汎用カプラー
Accurate simulation of terrestrial processes is crucial for climate projections, impact assessment studies, and forecasting floods and droughts. The 1-km spatial resolution aligns closely with the human experience scale, offering the potential for realistic flood forecasts and water resource information. To this end, we have investigated the possibility of conducting global 1-km scale terrestrial hydrological simulations. This study utilizes the Integrated Land Simulator to couple a land model, MATSIRO, with an independent I/O module through an ILS coupling interface, which employs the general-purpose coupler, Jcup.
During our presentation, we will initially describe the computational methods employed by each module. Subsequently, we will address the challenges faced in executing hyper-resolution simulations and the strategies to overcome them. The original coupling system was designed for lower-resolution models, where the computational demand was sufficiently extensive compared to the coupling processes. This system required adjustments for our high-resolution study, which involves considerable computational loads around the coupling interface. Following these modifications, we achieved a setup that allows for completing a 1-year simulation within a practical time frame of 12.8 hours using the Fugaku supercomputer.
A global 1-km experiment, with 1-km resolution land parameters (land cover, soil classification, leaf area index, etc.) and 0.5-degree meteorological data, showed that the canopy interception, which has a significant impact from LAI, can be represented at 1-km. We evaluated the impact of high-resolution land parameters on various variables by comparing the ratio of the mean of standard deviations within each 0.5-degree grid cell to the overall 0.5-degree mean. This analysis revealed that snow cover-related variables were minimally impacted, whereas soil moisture, canopy intercepted water, and latent and sensible heat fluxes experienced more substantial effects.
During our presentation, we will initially describe the computational methods employed by each module. Subsequently, we will address the challenges faced in executing hyper-resolution simulations and the strategies to overcome them. The original coupling system was designed for lower-resolution models, where the computational demand was sufficiently extensive compared to the coupling processes. This system required adjustments for our high-resolution study, which involves considerable computational loads around the coupling interface. Following these modifications, we achieved a setup that allows for completing a 1-year simulation within a practical time frame of 12.8 hours using the Fugaku supercomputer.
A global 1-km experiment, with 1-km resolution land parameters (land cover, soil classification, leaf area index, etc.) and 0.5-degree meteorological data, showed that the canopy interception, which has a significant impact from LAI, can be represented at 1-km. We evaluated the impact of high-resolution land parameters on various variables by comparing the ratio of the mean of standard deviations within each 0.5-degree grid cell to the overall 0.5-degree mean. This analysis revealed that snow cover-related variables were minimally impacted, whereas soil moisture, canopy intercepted water, and latent and sensible heat fluxes experienced more substantial effects.