Houghton Lecture Series – Dale Durran (University of Washington)

Lorenz revisited: The two- to four- day predictability of mid-latitude cyclones

Ed Lorenz published two seminal papers on predictability. Lorenz (1963) demonstrated that deterministic systems for which all solutions are bounded can exhibit solutions with sensitive (chaotic) dependence on the initial conditions such that an accurate long-range forecast of the system behavior would require initial conditions to be defined to within exceedingly small error. In contrast, Lorenz (1969) demonstrated that fluid systems with many interacting scales of motion can have a finite predictability limit for all initial conditions with errors larger than zero owing to a rapid upscale cascade of unobservable small-magnitude initial errors. The Lorenz (1963) system is nonlinear but does not involve the interaction of motions on different scales. The Lorenz (1969) system is linear, except for a nonlinear saturation hypothesis; the scale interactions in this system come from the product of the perturbations and the spatially variable ensemble mean state. Being a linear system approximating homogeneous turbulence, the Lorenz (1969) has questionable relevance to the problem of determining the initial-condition sensitivity of forecast busts involving mid-latitude cyclones. A forecast bust for a coherent system like the cyclone inherently involves a significant nonlinear modification of the fluid state.
We conduct idealized simulations of moist baroclinically unstable flow to examine the
importance of nonlinearity, moist convection, and the horizontal scale of initial perturbations on error growth during rapid cyclogenesis. Our results suggest that on forecast lead times of 2 to 4 days, there is no need to sweat the small stuff. The forecast challenge is closer to that suggested in Lorenz (1963).

About this Series:
Supported by the Houghton Fund, Houghton Lecturers are distinguished visitors from outside MIT invited by the EAPS Program in Atmospheres, Oceans and Climate to spend a period of time, ranging from a week to several months, as scientists-in-residence within our Program. For more information and Zoom password please contact Kayla Bauer: kbauer@mit.edu