Department of Atmospheric Science, University of California, Los Angeles, California

A hybrid coupled model (HCM), similar to the one used in Syu et al. (1995), Waliser et el. (1994) and Blanke et al. (1997), is used to predict the NINO3 SST anomaly (SSTA). The atmospheric model is estimated from observations using a singular value decomposition (SVD) technique. The model contains the first seven SVD modes of the covariance matrix calculated from the time series of pairs of observed monthly mean Reynolds SST anomalies and Florida State University (FSU) subjective pseudo-stress anomalies over a 19-year period from January, 1970 through December, 1988. Atmospheric adjustment is parameterized by a simple 60-day spin-up time scale (Syu and Neelin 2000a). Heat flux is parameterized according to Oberhuber's (1988) formulation using climatological data, with the negative feedback on SST estimated following Seager et al. (1988). The OGCM is a version of the GFDL Modular Ocean Model (Pacanowski, Dixon and Rosati, 1991, personal communication) for the Pacific basin. The vertical resolution is 27 levels, with 10 levels in the upper 100 meters. A Richardson-number-dependent vertical mixing scheme is combined with a surface mixed layer parameterization (see Syu and Neelin 2000a for model details).

The HCM has a reasonable simulation of ENSO in spatial and temporal features, with ENSO periods of 3 to 4 years. Model performance in "retroactive real-time forecasts" (hindcasts hereafter) from 1980-1992 has been shown in the September 1997 issue of the Experimental Long-Lead Forecast Bulletin, with further analysis in Syu and Neelin (2000b). The ocean climatology used in all hindcast/forecast experiments is specified to be the averaged model SST, forced by the FSU subjective wind stress product over 1978 to 1993 without modification by the data assimilation scheme. The climatological wind stress used in the hindcast/forecast experiments is also specified to be the average of the FSU subjective wind stress over the same period (1978-1993). The forecast results after 1994 are verified against the observations from Reynolds' (1988) SST data set after applying optimum interpolation method as described in Reynolds and Smith (1994).

The initialization scheme makes use of both the wind information (FSU converted wind stress) and the ocean model data assimilation product from the Climate Prediction Center (CPC) (Ji et al., 1995). In addition to the specified FSU wind stress forcing, the CPC reanalyzed anomalous ocean temperature field is "injected" into the ocean model (27 layers) every month since 1980 up to the start of the hindcast (injection scheme hereafter). Because our ocean model (GFDL MOM) is in a version reasonably close to that used by CPC, approximate consistency is assumed in injecting the CPC reanalyzed data. To make distinction between this procedure and raw-data injection, we refer to it as a "piggy-back" data assimilation scheme, because it makes use of the effort from an CPC data assimilation product. The "piggy-back" data assimilation scheme gives a substantial improvement in hindcast skill (see the Sep. 1997 issue and Syu and Neelin 2000b), and thus appears to be a viable, economical forecast method.

As of March 2002 Florida State University ceased output of their subjective pseudo- stress product in favor of an objectively analyzed product. For continuity of the forecast system, we continue to use the same coupled model climatology which makes use of the FSU subjective pseudo-stress climatology from 1978-1993. The coupled model itself thus does not change. In initializing the ocean, anomalies of the new FSU objective wind product (relative to its 1978-1993 mean) are used. Subsurface and SST data are employed as before.

For this issue of ELLFB we present results of the past and present initialization schemes along with the routine forecast. Figure 1 presents the NINO3 index for forecasts from 1993 to May 2002 using the new FSU objective wind product and 1993 to February 2002 for the previous run (FSU subjective product not available past February). Also shown are observed NINO3 SSTAs through May 2002 (3-month running average, thick black curves) and forecasts (gray curves) at 3-, 6- and 9-month lead for the present experiment along with the previous experiment (March 2002, dashed gray curve) for comparison of the two initializations. Averages of each lead month based on forecast verification over the 1980-1992 time span are removed before plotting the curves. These corrections are modest: 0.051, 0.21, and 0.34 C for 3, 6 and 9 month lead respectively. For continuity, these values were not changed. Vertical bars represent plus and minus one RMS error, over the same forecast verification time span. The two initialization systems behave similarly overall, especially for longer leads. For 9-month lead there are very few months in the interval where the error bars of the two forecasts do not overlap in some range and usually one forecast is within the error bars of the other. For 3-month lead, there are a number of exceptions to this: in late boreal summer of 1998 the initialization with the FSU objective product overpredicts the onset of cold conditions, while in boreal winter of 1999-2000, this system predicts a spurious mild warming, both of which compare worse to observations than the published forecasts using the FSU subjective product. At six month lead, while the small spurious warming in 99-00 occurs with the FSU objective product, this initialization fares slightly better at capturing the 1997-98 warming and subsequent return to normal than the subjective product.

The latter part of the forecast track record in Figure 1 indicates that the two initializations differ for the current forecast period. The previous forecast (Mar. 02 ELLFB) initialized with the FSU subjective wind product showed a mild warming trend for each of the three leads, but the forecast initialized with the objective winds tends to be slightly cooler.

To provide a comparison in the format of the actual forecast, as opposed to the track record sorted by lead, Figure 2 shows examples of published past forecasts with the FSU subjective wind product with the objective initialization case added (starting from January and February, 2001 and 2002, respectively, for 12 months). Model initialization runs and forecasts are dashed and dotted, respectively with gray/black denoting those initialized with the FSU subjective/objective products. There are some substantial differences in the SSTA of the initialization runs, especially in mid-1998 when the FSU objective wind causes excessive cooling in NINO3. For the forecasts from January and February 2001 data (Figure 2a), there is a general agreement between the two of a moderate warming by the winter of 2001-02, although the short-term SSTA evolution of the two differs. Note that both forecasts were off, as observed NINO3 SST decreased to normal conditions by Nov. 2001. In Figure 2b, the forecast curves fr January and February 2002 for both initializations are also in relative agreement at the end of the 1 year forecast showing near normal conditions for early 2003, but the amplitude of oscillation about normal is much exaggerated in the new FSU objective wind initialization curve. It thus appears that, as of the last available FSU subjective wind data available, the current period is one in which the two products have substantially different impact on the forecast.

Figure 3 shows observations (thick, solid curve) and the model initialization run with the FSU objective product since 1993 (thick dashed curve) along with the previous initialization with subjective FSU wind product (gray curve) used in previous forecasts. The differences between the initialization runs depend only on the ocean model reaction to the different wind stress products and occur despite the inclusion of both SST and NCEP subsurface data. The differences are often larger than the differences in the forecasts with the two products seen in Figure 1.

Figure 3 also displays the latest two forecast results (starting from April and May, 2002, respectively, for 12 months), with the FSU objective wind used in the initialization. Both forecasts initiated with April and May observations show a cooling trend through the summer months rebounding to normal conditions by spring 2003. Given the differences in the forecasts from February data with the subjective wind product and the fact that observed NINO3 SSTA has shown a warming trend since January 2002, we place lower than normal confidence in this forecast.

References:

Blanke, B., J. D. Neelin, and D. Gutzler, 1997: Estimating the effect of stochastic wind stress forcing on ENSO irregularity. J. Climate, 10, 1473-1486.

Ji, M., A. Leetmaa, and J. Derber, 1995: An ocean analysis system for seasonal to interannual climate studies. Mon. Wea. Rev., 123, 460-481.

Oberhuber, J. M., 1988: An atlas based on the COADS data set: the budgets of heat, buoyancy and turbulent kinetic energy at the surface of the global ocean. Max-Planck-Institut f"ur Meteorologie Report No. 15, Bundesstrasse 55, D-2000, Hamburg 13, FRG.

Reynolds, R. W., 1988: A real-time global sea surface temperature analysis. J. Climate, 1, 75-86.

Reynolds, R. W. and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate, 7, 929-948.

Seager, R., S. E. Zebiak, and M. A. Cane, 1988: A model of the tropical Pacific sea surface temperature climatology. J. Geophys. Res., 93, 1265-1280.

Syu H.-H., J. D. Neelin, and D. Gutzler, 1995: Seasonal and interannual variability in a hybrid coupled GCM. J. Clim., 8, 2121-2143.

Syu, H.-H., and J. D. Neelin, 2000a: ENSO in a hybrid coupled model. Part I: sensitivity to physical parameterizations. Climate Dynamics, 16, 19-34.

Syu, H.-H., and J. D. Neelin, 2000b: ENSO in a hybrid coupled model: Part II: prediction with piggyback data assimilation. Climate Dynamics, 16, 35-48.

Waliser, D. E., B. Blanke, J. D. Neelin, and C. Gautier, 1994: Shortwave feedbacks and El Nino-Southern Oscillation: Forced ocean and coupled ocean-atmosphere experiments. J. Geophys, Res., 99, 25109-25125.

Figure Captions:

Fig. 1. The forecasts of NINO3 SST anomalies from 1993 to present using the new FSU objective wind product (solid gray curve) and 1993 to February 2002 using the old FSU subjective wind product (dashed gray curve). The latest forecast starts from May 2002. The mean for each lead month over the forecast verification time span (1980-92) is removed before plotting. Vertical bars represent plus and minus one RMS error over the same forecast verification time span. Shown for (a) 3-month, (b) 6-month and (c) 9-month lead.

Fig. 2. Comparison of past forecasts with the two FSU wind products (starting from January and February, 2001 and 2002, respectively, for 12 months). NINO3 observations are shown as the solid black curve; the dashed lines are the model initialization runs with FSU subjective (gray curve) and objective (black curve) wind products; forecasts are dotted lines with the same convention for gray and black. a.) January and February 2001 forecasts. b.) January and February 2002 forecasts.

Fig. 3. The latest two forecasts (dotted lines) of NINO3 SST anomalies up to 12 lead months starting from April and May 2002. Observations (black solid line) and model control run (black dashed line) from 1993 to present are also shown. The mean for each lead month is removed as in Fig. 1. Vertical bars indicate the same plus and minus one RMS error used in Fig. 1. The model control run from 1993 to February 2002 which employed the FSU subjective wind product is shown for comparison (gray dashed line).