One Saddle Point and Two Types of Sensitivities within the Lorenz 1963 and 1969 Models
Original article: https://doi.org/10.3390/atmos13050753
Summary
This paper compares two sources of finite predictability in Lorenz's 1963 and 1969 models. The Lorenz 1963 model is nonlinear and limited-scale, with sensitive dependence on initial conditions emerging in the chaotic regime. The Lorenz 1969 model is physically multiscale but mathematically linear and ill-conditioned, so numerical sensitivity can appear through large condition numbers and unstable modes. The study cautions against treating either model, by itself, as proof of a universal two-week atmospheric predictability limit.
