Forecast of NINO3 using a low order coupled ocean-atmosphere model

contributed by R. Kleeman

Bureau of Meteorology Research Centre, Melbourne, Australia

A simple coupled ocean/atmosphere model has been developed at BMRC (Kleeman 1993) in order to explore the physical basis of ENSO predictability. In particular a variety of very simple ocean models with varying hermodynamical equations governing SST have been coupled to a simple atmospheric model which performs well when forced by a full range of ENSO SST anomalies (Kleeman 1991).

The coupled models are somewhat similar to that of Cane and Zebiak (see entry in this issue; Cane and Zebiak 1987) but differ in aspects of the coupling, atmospheric convection and heating, and ocean thermodynamics.

The hindcast skill of these coupled models was tested using the ocean models initialized at regular 3 month intervals between January 1972 and July 1986 using FSU winds and it was determined that optimal skill was obtained when the ocean model SST was determined purely by equatorial thermocline perturbations. Later (Kleeman et al., 1995) the initialization of the coupled model was improved by using a space-time variational (adjoint) technique to assimilate sub-surface thermal data, as well as the usual wind data, into the ocean model. The forecasts made here are with an unaltered version of this 1995 model.

The model skill was assessed in two different ways: First, predictions made 1 month apart from Jan 1981 to Dec 1994 were compared with the observations. Figure 1 shows a scatter plot of predicted and observed six and twelve month forecasts. This diagram should allow users to obtain some idea of both the skill and uncertainties of the real time prediction below. Secondly, historical forecast skill was compared with that obtained by other prediction schemes using the recent intercomparison of Kirtman and Shukla (1998). In this study, skill was computed for a common subset of forecasts for the years 1982-84, 1986-89 and 1991. This was repeated for the system here and anomaly correlations of .83, .82 and .80 were obtained for NINO3 forecasts of duration 6, 9 and 12 months respectively. These levels of skill compare well with those obtained from the other systems in the intercomparison.

The model in exactly its present form has been run since early 1994 so validation of real-time forecasts is appropriate since that date. Displayed in Figure 2 is a comparison of these forecasts with observations at six month lag.

Displayed in Figure 3 is the most recent forecast of NINO3 which uses the FSU winds up to January 2000 and sub-surface thermal data up to February 2000 to initialize the model. The forecast shows that weak cold conditions are expected to occur towards the end of the current year.

References

Kirtman B.P. and J. Shukla, 1998, ``Current Status of ENSO Forecast Skill'' A report to the Climate Variability and Predictability (CLIVAR) Numerical Experimentation Group (NEG).

Kleeman R., 1991, ``A simple model of the atmospheric response to ENSO sea surface temperature anomalies'', J.Atmos. Sci. , 48 , pp3-18.

Kleeman R., 1993, ``On the dependence of hindcast skill on ocean thermodynamics in a coupled ocean-atmosphere model'', J. Clim., 6, pp2012-2033.

Kleeman R., A.M. Moore and N.R. Smith, 1995, ``Assimilation of sub-surface thermal data into an intermediate tropical coupled ocean-atmosphere model'', Mon. Weath. Rev. 123, pp3103-3113.