Generalized inversion of Tropical Atmosphere-Ocean (TAO) data and a coupled model of the tropical Pacific. Part II: The 1995-96 La Nina and 1997-98 El Nino

The investigation of the consequences of trying to blend tropical Pacific o bservations from the Tropical Atmosphere-Ocean (TAO) array into the dynamic al framework of an intermediate coupled ocean-atmosphere model is continued . In a previous study it was found that the model dynamics, the prior estim ates of uncertainty in the observations, and the estimates of the errors in the dynamical equations of the model could not be reconciled with data fro m the 1994-95 period. The error estimates and the data forced the rejection of the model physics as being unacceptably in error. In this work, data fr om two periods (1995-96 and 1997-98) were used when the tropical Pacific wa s in states very different from the previous study. The consequences of inc reasing the prior error estimates were explored in an effort to find out ii it is possible at least to use the intermediate model physics to assist in mapping the observations into fields in a statistically consistent way. It was found that such a result is possible for the new data periods, with larger prior error assumptions. However. examination of the behavior of the mapped fields indicates that they have no dynamical utility. The model dyn amical residuals, that is, the size of the quantity that is left after eval uating all of the terms in each intermediate model equation, dominate the t erms themselves. Evidently the intermediate model is not able to add insigh t into the processes that caused the tropical Pacific to behave as it was o bserved to, during these time intervals. The inverse techniques described here together with the relatively dense TA O dataset have made it possible for the unambiguous rejection of the nonlin ear intermediate model dynamical system. This if the first time that for th e unambiguous rejection of the nonlinear intermediate model dynamical syste m.