14:30 〜 14:45
[AAS01-04] Feasibility of upstream weather intervention for downstream heavy rainfall mitigation based on ensemble sensitivity analysis
キーワード：極端降水、気象制御、アンサンブル感度解析
The frequency of heavy rainfall events is increasing due to global warming. One possible approach to reducing the impact of heavy rainfall disasters on society is to artificially generate or enhance rainfall over upstream (windward) oceanic regions to suppress heavy rainfall downstream over land. In this study, we report the results of an ensemble sensitivity analysis on rainfall over a target land area using an ensemble numerical experiment, aiming to assess the possibility of "realistic" weather intervention.
The RIKEN SCALE-RM model was used to reproduce a heavy rainfall event that occurred in Kyushu in mid-August 2021. A 100-member ensemble experiment was conducted by introducing random noise into the specific humidity of the boundary layer at the initial time. To evaluate the dependence on grid spacing, experiments were performed at several horizontal grid spacings ranging from 3.2 km to 800 m.
The area-averaged six-hour precipitation from 00 UTC on August 12, 2021, over a 50 km x 50 km region near Kumamoto Prefecture was used as the target variable. We calculated the lagged correlation and regression coefficients between the target variable and meteorological variables, such as six-hour precipitation at each grid point over the ocean southwest of Kyushu. The results show that statistically significant negative correlations emerge 24 to 12 hours before the target time over an upstream oceanic region approximately 100-200 km southwest of the target area. This finding implies that increasing precipitation in the negatively correlated region through weather intervention could potentially reduce rainfall over the target region half a day to one day later.
Since this analysis is currently based on a single case, additional cases are needed to establish robustness. In the presentation, we will also report on the downstream effects of meteorological intervention, such as those induced by placing kite-like objects.
The RIKEN SCALE-RM model was used to reproduce a heavy rainfall event that occurred in Kyushu in mid-August 2021. A 100-member ensemble experiment was conducted by introducing random noise into the specific humidity of the boundary layer at the initial time. To evaluate the dependence on grid spacing, experiments were performed at several horizontal grid spacings ranging from 3.2 km to 800 m.
The area-averaged six-hour precipitation from 00 UTC on August 12, 2021, over a 50 km x 50 km region near Kumamoto Prefecture was used as the target variable. We calculated the lagged correlation and regression coefficients between the target variable and meteorological variables, such as six-hour precipitation at each grid point over the ocean southwest of Kyushu. The results show that statistically significant negative correlations emerge 24 to 12 hours before the target time over an upstream oceanic region approximately 100-200 km southwest of the target area. This finding implies that increasing precipitation in the negatively correlated region through weather intervention could potentially reduce rainfall over the target region half a day to one day later.
Since this analysis is currently based on a single case, additional cases are needed to establish robustness. In the presentation, we will also report on the downstream effects of meteorological intervention, such as those induced by placing kite-like objects.