arXiv subjects

Hiroaki Miura

Publications and source records attributed to Hiroaki Miura.

6 recordsLinked to original sources

arXiv2026

Quantum algorithm for the collision-coalescence of cloud droplets

Quantum computing may help reduce computational costs of simulating large and nonlinear systems, but research into the use of quantum computers in atmospheric and oceanic sciences is still at an early stage. This study explores the use of quantum computing for calculating the collision-coalescence process that dominates the size growth of liquid particles in cloud microphysics. Inspired by the quantum algorithms developed in the field of financial engineering, we propose a new algorithm based on a master equation that describes the time evolution of the droplet mass distribution. Our algorithm uses the quantum amplitudes to encode the probability distribution of droplet mass and calculates the expected number of droplets via the quantum amplitude estimation. The key contribution is an encoding strategy that maps the multivariate collision-coalescence problem onto a quantum computation by recording only the transition history rather than the full mass distributions at every time step. This approach reduces the per-step qubit cost from $O(\sqrt{N})$ to $O(\log N)$, where $N$ is the number of mass bins. Our resource analysis shows that the number of T gates (T-count) scales as $\widetilde{O}(N^2)$ per mass bin in $N$, for a fixed number of time steps and a given target accuracy, where the $\widetilde{O}$ notation suppresses polylogarithmic factors in $N$. This is a substantial improvement over classical master-equation methods, whose cost grows super-polynomially with $N$ as the number of possible states increases. Our results suggest that the collision-coalescence process is one of the promising targets of quantum computing in the field of atmospheric science.

quant-ph↗

arXiv2025

Propagation of the Madden-Julian oscillation as a deterministic chaotic phenomenon

The Madden-Julian oscillation (MJO), a gigantic tropical weather system, is marked by eastward travel of cumulus cloud clusters over the Indo-Pacific region and often causes severe weather and climate events worldwide. The physics and predictability of MJO propagation remain elusive, partly because of little attention to untangling roles of multi-scale processes relevant to the MJO. Here, we reveal the chaotic nature of MJO propagation arising from cross-scale nonlinear interactions, based on 4,000-member ensemble global cloud-system-resolving simulations of two MJO events. Against conventional linearized thinking, multiple regimes with distinct timings of MJO propagation emerge under a single atmosphere-ocean background. The bifurcation emergence depends critically on the equatorial asymmetry of climatological sea surface temperature. Selection of the bifurcated regimes is probabilistic, influenced by whether tropical-extratropical interplay promotes moistening associated with westward-propagating tropical waves over the western Pacific. These aspects help build a comprehensive MJO model and foresee when the MJO propagates.

physics.ao-ph↗

arXiv2025

A Stochastic Lattice Model for Convective Self-aggregation Incorporating Longwave Radiative Effect

Self-aggregation of tropical convection is a universal feature observed in a diverse range of atmospheric environments. Several preceding models conceptualized the self-aggregation of convection as a phase transition driven by collisions between cold pool gust fronts. However, self-aggregation may also be influenced by various physical processes, such as surface fluxes, radiation, and moisture perturbations in the planetary boundary layer, and it remains unclear which process plays a dominant role. In this study, we develop a simple stochastic lattice model for the pattern formation of deep convection, inspired by the two-dimensional Ising model. Here, in addition to the process of cold pool collisions, which have an effect of triggering new convection, we incorporate the process of clear-sky radiative cooling that has an effect of suppressing deep convection as an interaction between clouds. Our results show that by amplifying the intensity of the clear-sky radiative cooling effect, the transition from a quasi-uniform to an inhomogeneous cloud field can be reproduced. The model also successfully explains the dependence of self-aggregation on several model parameters, such as the experimental domain size and the characteristic size of cold pools. Furthermore, by varying the distance over which the subsidence induced by radiative cooling extends, we succeed in capturing a pattern formation that closely resembles the convective clusters observed in the real atmosphere and three-dimensional numerical model simulations.

physics.ao-ph↗

arXiv2024

Quantum Algorithm for a Stochastic Multicloud Model

Quantum computers have attracted much attention in recent years. This is because the development of the actual quantum machine is accelerating. Research on how to use quantum computers is active in the fields such as quantum chemistry and machine learning, where vast amounts of computation are required. However, in weather and climate simulations, less research has been done despite similar computational demands. In this study, a quantum computing algorithm is applied to a problem of the atmospheric science. The effectiveness of the proposed algorithm is evaluated using a quantum simulator. The results show that it can achieve the same simulations as a conventional algorithm designed for classical computers. More specifically, the stochastically fluctuating behavior of a multi-cloud model was obtained using classical Monte Carlo method, and comparable results are also achieved by utilizing probabilistic outputs of computed quantum states. Our results show that quantum computers have a potential to be useful for the atmospheric and oceanic science, in which stochasticity is widely inherent.

quant-ph↗

arXiv2022

Machine learning prediction of the MJO extends beyond one month

The prediction of the Madden-Julian Oscillation (MJO), a massive tropical weather event with vast global socio-economic impacts, has been infamously difficult with physics-based weather prediction models. Here we construct a machine learning model using reservoir computing technique that forecasts the real-time multivariate MJO index (RMM), a macroscopic variable that represents the state of the MJO. The training data was refined by developing a novel filter that extracts the recurrency of MJO signals from the raw atmospheric data and selecting a suitable time-delay coordinate of the RMM. The model demonstrated the skill to forecast the state of MJO events for a month from the pre-developmental stages. Best-performing cases predicted the RMM sequence over two months, which exceeds the expected inherent predictability limit of the MJO.

physics.ao-ph↗

arXiv2018

New Immersed Boundary Method with Irrotational Discrete Delta Vector for Droplet Simulations with Large Density ratio

The Immersed Boundary Method (IBM) is one of the popular one-fluid mixed Eulerian-Lagrangian methods to simulate motion of droplets. While the treatment of a moving complex boundary is an extremely time consuming and formidable task in a traditional boundary-fitted fluid solver, the one-fluid methods provide a relatively easier way to track moving interfaces on a fixed Cartesian grid since the regeneration of a mesh system that conforms to the interface at every time step can be avoided. In the IBM, a series of connected Lagrangian markers are used to represent a fluid-fluid interface and the boundary condition is enforced by adding a forcing term to the Navier-Stokes equations. It is known that the IBM suffers two problems. One is spontaneous generation of unphysical kinetic energy, which is known as parasitic currents, and the other is spurious reconstruction of interface. These two problems need to be solved for useful long-time-scale simulations of droplets with high density ratio and large surface tension. This work detects that the discrete delta function is the cause of unphysical parasitic currents. Specifically, the irrotational condition is not preserved when the common discrete delta function is used to spread the surface tension from Lagrangian markers to Cartesian grid cells. To solve this problem, a new scheme that preserves the irrotational condition is proposed to remove the spurious currents. Furthermore, for a smooth reconstruction of an interface, a B-spline fitting by least squares is adopted to relocate the Lagrangian markers. The conventional and new interpolation schemes are implemented in a multigrid finite volume Direct Numerical Simulation (DNS) solver and are subjected to standard test cases. It is confirmed that the unphysical parasitic currents are substantially reduced in the new scheme and droplet's surface fluctuation is eliminated.

physics.comp-ph↗