| description abstract | Earlier studies have used a series of tests based on linear inverse modeling (LIM) to investigate the proposition that the dynamics of observed tropical Indo-Pacific sea surface temperature anomalies may be described as a stable linear system driven by spatially coherent stochastic forcing. Previous verifications of these tests? dependability have been performed on highly idealized systems such as the chaotic Lorenz model. In this study, LIM and tests for validity of its underlying assumptions are performed on the output of an intermediate coupled ocean?atmosphere numerical model. Five cases are considered: 1) a coupled ocean?atmosphere, with the dynamics dominated by a single, slightly unstable mode (case 1); 2) a model equivalent to case 1, but with additional stochastic forcing (case 2); 3) a coupled ocean?atmosphere as in case 2, but with the dynamics adjusted so as to be in the stable regime (case 3); 4) yet another coupled case (case 4) but with a stable dynamical regime farther away from the bifurcation point than in case 3; and 5) an uncoupled ocean, driven by a stochastic atmosphere (case 5). Although none of the cases passed the tests for linearity as well as the data considered by Penland and Sardeshmukh, cases 1, 2, and 3 were identified as more strongly nonlinear systems than cases 4 and 5. The authors? results support the reliability of the tests performed by Penland and Sardeshmukh on Comprehensive Ocean?Atmosphere Data Set data, which suggested that ENSO is likely to reside in a stable dynamical regime maintained by stochastic forcing. | |
