Severe rainfall events can cause significant harm to individuals, damage infrastructure, and result in substantial economic losses. If precipitation regulation could be realized, it could help mitigate the risks of disasters. However, controlling precipitation remains a formidable challenge due to the highly complex and uncertain dynamics of weather systems. To address this, we propose a novel control framework for precipitation management based on a numerical weather prediction (NWP) model and applied the framework for a series of warm bubble experiments, where the direction and amplitude of regional wind serve as the input and precipitation as the output. This approach investigates the potential of modifying regional wind patterns to effectively influence and regulate precipitation intensity and distribution.

    The proposed framework integrates a Sampling-Based Model Predictive Control (SBMPC) module to generate ideal control inputs for precipitation reduction and a novel Control Barrier Function (CBF) module to refine these inputs when discrepancies between the model and real weather systems are detected. The SBMPC combines the strengths of model predictive control and random sampling techniques to efficiently solve high-dimensional and nonlinear optimization problems. Inspired by ensemble prediction methods in numerical weather forecasting, the developed SBMPC module uses sampled control inputs to simulate potential system responses with a numerical weather prediction model and selects the input, whose corresponding output most closely aligned with the desired one, as the nominal control input. However, the effectiveness of the SBMPC module depends heavily on the accuracy of the NWP model, making it vulnerable to discrepancies between the simulated and real weather systems.

    To mitigate this limitation, our control framework incorporates a CBF module to ensure safety by enforcing constraints, such as maintaining precipitation intensities within predefined boundaries to prevent extreme weather events. Unlike conventional CBF methods, which rely on precise system dynamics, the CBF controller developed in this work reduces dependency on detailed models, making it particularly effective for managing the complexity and uncertainty inherent in weather systems.    

    The feasibility of the proposed framework is validated through simulations using the SCALE Regional Model (SCALE-RM), which emulates real-world weather systems. Results demonstrate that the proposed control framework effectively regulates precipitation to a safe level and maintains computational efficiency, offering a robust and practical solution for managing precipitation.

How to cite: Bai, Y., Ogura, M., and Kotsuki, S.: A Robust Control Framework for Precipitation Regulation under NWP Model Uncertainty, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-1846, https://doi.org/10.5194/egusphere-egu25-1846, 2025.

To support disaster prevention, it is essential to know in advance when scenarios start to distinguish one from the others, thus requiring the development of early detection methods of such separations. Ensemble prediction systems has been developed to provide scenarios of evolutions via their ensemble members, because the future state of the atmosphere predicted by a single ensemble member is less meaningful than the estimate of the future probability density from all the ensemble members. By construction, the primary function of an ensemble prediction system is to provide forecasters with a degree of uncertainty and level of confidence. For the last couple of years, we have developed different approaches which take advantage of the information provided by different ensemble prediction systems.

Tropical cyclone tracks forecasted by a prediction system sometimes group together into trajectories parting away from each other. An objective method, based on a robust clustering approach, has been created to detect such separation scenarios in the Mesoscale Ensemble Prediction System developed by the Japan Meteorological Agency. At each initialization time, when the number of clusters is greater than 1, local separation scenarios exist. Separations are related to different steering environments predicted by the different ensemble members.

On the same data used for the clustering, we also applied biological concepts such as the Waddington’s epigenetic landscape, and bioinformatics techniques like the graph-Hodge decomposition, to produce “MeteoScape”, an innovative way to characterize the evolution of a tropical cyclone. Particularly, “MeteoScape” can detect the possible paths and their associated probabilities of realization, as well as the regional separatrix, i.e., the spatial boundary between paths/scenarios, regardless of the initialization time (as in the clustering approach).

Using the cases of intense precipitations that occurred in western Japan in July 2018 and August 2021, we introduce a way to detect separation scenarios in a n-dimensional space. A dimensionality reduction technique is applied to the geopotential height at 500 hPa extracted from two different ensemble prediction systems (the Japanese Mesoscale Ensemble Prediction System and the North American National Centers for Environmental Prediction). In the resulting 3-dimensional latent space, trajectories of ensemble members at different initialization times can also part away. We show that these separations are linked to attracting atmospheric trajectories.

Further, the different techniques developed provide information on controllability, particularly where and when a manipulation could be performed.

How to cite: Oettli, P., Tokuda, K., Imoto, Y., and Kotsuki, S.: Detection of Separation Scenarios in Extreme Weather Events Using Regional Ensemble Prediction Data, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7791, https://doi.org/10.5194/egusphere-egu25-7791, 2025.

If the error growth of initial conditions in numerical weather systems is scale-dependent, then micro or mesoscale significantly affects the accuracy of a cyclonic weather system prediction. Thus, as the forecast skill improves (by including smaller-scale phenomena and reducing the error of the initial conditions), we would reach an intrinsic limit of predictability that we have set for the forecast of mid-latitude synoptic phenomena (geopotential height 500 hPa) at about three weeks. For scale-dependent initial error growth, it may turn out that small-scale phenomena that contribute little to the forecast product significantly affect the ability to predict that product. It is reasonable to study whether omitting these atmospheric phenomena will improve the predictability of the final value. The topic is studied in the extended system of Lorenz (2005). This system shows that omitting small spatiotemporal scales will reduce predictability more than modeling it. In other words, a system with model error (omitting phenomena) will not improve predictability. Orrell's hypothesis is extended to explain and describe this behavior, whereby the difference between systems (model error) produced at each time step is treated as an error in the initial conditions. The resulting model error is then defined as the sum of the time evolution increments of the initial conditions so defined. The theory is compared to the fit parameters that define the model error in certain approximations of the average forecast error growth. Parameters are interpreted in this context, and the hypotheses are used to estimate the errors described in the theory. The results of the annual averages of the prediction error growth (geopotential height 500 hPa) of the ECMWF system from 1987 to 2011 are discussed. 

How to cite: Bednar, H. and Kantz, H.: Analysis of initial and model error in forecast errors of extended atmospheric Lorenz' 05 systems and the ECMWF system, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-2014, https://doi.org/10.5194/egusphere-egu25-2014, 2025.

The prediction and mitigation of extreme weather events are important challenges in science and society. Recently, Miyoshi and colleagues introduced the control simulation experiment framework to examine the controllability of chaotic systems under observational uncertainty. Using this framework, they developed a method to reduce extreme events in the Lorenz 96 model by exploiting the system’s sensitivity to initial conditions, guiding trajectories toward desired outcomes with small control inputs (Sun et al., Nonlin. Processes Geophys., 30, 117-128, 2023). Their method is primarily designed to apply control inputs to all grid variables, reducing the success rate of extreme event mitigation to approximately 60% when the control input is applied to only one site. In this study, we propose a new approach to mitigate extreme events in the Lorenz 96 model through local interventions based on multi-scenario ensemble forecasts. Specifically, we explore effective intervention scenarios by computing the system’s responses to a limited number of feasible interventions. Our method achieves a success rate of 94%, even when interventions are applied to only one site per step. This represents a significant improvement over the ~60% success rate of the previous study, albeit at a moderately higher intervention cost. Furthermore, the success rate increases to 99.4% when interventions are applied to two sites.

How to cite: Mitsui, T., Kotsuki, S., Fujiwara, N., Okazaki, A., and Tokuda, K.: Mitigating extreme events through multi-scenario ensemble forecasts and local interventions in the Lorenz 96 model, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-8594, https://doi.org/10.5194/egusphere-egu25-8594, 2025.

Tropical cyclones are highly destructive natural disasters that pose a grave threat to society. As a result, the Moonshot Project of the Typhoon Control Research Aiming for a Safe and Prosperous Society is working towards finding artificial means to reduce an approaching typhoon. Therefore, making it less destructive. To achieve this less destructive storm, we initiate a cold pool, a natural feature of convective storms, to suppress convection by cooling the sub-cloud layer of an approaching typhoon thereby reducing the amount of heat energy being fed to the storm. To test if this approach is feasible, we conduct a series of experiments using the stretched version of the non-hydrostatic icosahedral atmospheric model (NICAM) with a minimum grid spacing of 1.4km. The artificial cold pool is generated by simulated rain which we produce by pumping seawater continuously through 1km high cylindrical stacks and then ejecting it at the top. The stacks have a radius of 5km and 50km and are on a moving platform that is positioned at the centre of the Typhoon. Typhoon Jebi we use as our test case. Our generated cold pool has an intensity that is a constant cooling source of 1K/hr and 10K/hr each for each radius making a total of 4 experiments. The experiments run for 48 hours prior to landfall in Japan. Overall, the tracks of the four experiments are not affected, only minor shifts in the location of the centre. The time evolution of the sea level pressure (slp) shows that the presence of the cold pool affects the slp for all experiments where the experiment with an intensity of 10K/hr at a 50 km radius experiences the greatest increase in minimum slp, signaling a weakening of the cyclone. Snapshots at 6, 12, 24 and 48hr time intervals of the slp, 10m-windspeed, the 2m-temperature, and the total precipitation reveal that the presence of the cold pool varies in the degree it affects these parameters for each experiment. A pattern where areas of weakening and areas of strengthening encircle the cooling source emerged in the differences in the 10m-windspeed and slp snapshots. The experiment 10K/hr at 50km radius tends to show the only discernible cold pool in the 2m-temperature difference snapshot. However, this experiment was proven to be difficult to reproduce in reality so our focus is on the 10K/hr at 5km radius experiment. After zooming in on the 2m-temperature difference between the experiment and the control, we notice that the drop in temperature (evidence of the cold pool) for the 10K/hr at 5km radius, is very small therefore indicating that the cold pool is very weak. This initiated a new group of experiments where we are testing different locations to find the most suitable area in a tropical cyclone to position our moving platform with the cooling source. The preliminary results suggest that the best location lies somewhere in the inner eyewall region where the winds are no more than 20m/s. Further testing is being conducted.

How to cite: Lee, M. and Satoh, M.: Understanding the Impact an Artificial Cold Pool has on an Approaching Typhoon using The Nonhydrostatic Icosahedral Atmospheric Model (NICAM), EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14252, https://doi.org/10.5194/egusphere-egu25-14252, 2025.

We report that in several chaotic, high-dimensional nonlinear systems, the evolution of multiple ensembles starting from nearby initial conditions exhibits a transient low-dimensional distribution in phase space. This low-dimensional distribution of the ensembles is primarily achieved by stretching the ensemble distribution along the unstable directions of the system's trajectories. Furthermore, we discuss the potential of using this transient low-dimensional distribution to significantly reduce the search space when controlling the system's future states. As a concrete example of a high-dimensional nonlinear system, we use the Lorenz 96 model under a parameter setting that produces chaotic behaviors. We generate ensembles by adding small random perturbations to the initial conditions and compute the trajectories starting from each initial condition. By applying principal component analysis (PCA) to the ensemble distributions at each time step, we evaluate the dimension of the ensemble spread using a statistics that we call PCA dimension. Our results demonstrate that the PCA dimension initially decreases to values much smaller than the number of ensembles or the system's dimension, before increasing and converging to a value approximately equal to the Kaplan-Yorke dimension of the attractor. This phenomenon is considered to correspond to the asymmetry of the local Lyapunov exponents. Moreover, we show that at times when the PCA dimension of the ensemble distribution transiently decreases, it is possible to accurately regress the system's state at the time of a future extreme event using the scores of the first two principal components of the ensemble distribution. Additionally, using a regression model trained in this low-dimensional latent space, we succeed in identifying an optimal perturbation to the initial conditions to demonstrate the possibility of avoiding extreme events. Since meteorological phenomena are ultra-high-dimensional systems, attempts to reduce the dimensionality of the control search space may contribute to the feasibility of implementing such control measures. This presentation includes the most recent progress, such as using a weather prediction model, at the time of the conference.

How to cite: Tokuda, K., Mitsui, T., Kotsuki, S., and Fujiwara, N.: Transient low dimensional distribution of ensemble prediction in high dimensional chaos and control using the low dimensional latent representation, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14879, https://doi.org/10.5194/egusphere-egu25-14879, 2025.