Targeted adaptive chaos control of regimes and eddy strength in two Lorenz models

We develop an energy-efficient adaptive control framework in two low-order atmospheric dynamical systems, the Lorenz 63 (L63) and Lorenz 84 (L84) models, with multiplicative noise explicitly incorporated. Control is triggered using estimates of the local Lyapunov exponent (LLE), identifying dynamically sensitive states where perturbations are most effective. Once triggered, control amplitudes are determined by solving a constrained optimization problem over a finite forecast horizon that minimizes total control energy while keeping trajectories within prescribed bounds. In L63, regime transitions represent analogs of persistent circulation states, while in L84, large eddy amplitudes serve as surrogates for synoptic-scale moisture transport events such as atmospheric rivers. To account for model uncertainty, we explicitly introduce multiplicative noise and apply control to randomly selected ensemble realizations rather than deterministic trajectories. Despite this stochasticity, effective control is achieved with total energy inputs of only 10−3 to 10−4 of the system energy. Although highly idealized, these results demonstrate how instability-aware, minimal-energy control strategies can limit extreme states in chaotic systems. In the context of increasing extreme weather under climate change, these results provide a conceptual foundation for the Weather Jiu-Jitsu idea: exploiting the sensitivity of atmospheric dynamics to redirect or defuse high-impact events using small, well-timed perturbations.