National Academies Press: OpenBook
« Previous: APPENDICES
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

A
Present Status of Short-Term Climate Prediction

The 10-year (1985–1994) TOGA program was launched to study the predictable cycle of ENSO events in the tropics and their correlation with quasi-stationary features of atmospheric circulation at higher latitudes. One remarkable success of TOGA has been the rapid development of a theoretical understanding of ENSO as a dynamical phenomenon of the coupled ocean–atmosphere system. The TOGA research community now understands several basic mechanisms of ocean–atmosphere interactions that contribute to interannual variabilities; some insight into the interplay between these mechanisms has been developed. A hierarchy of coupled models (simple, intermediate, hybrid, and full GCMs) has simulated many aspects of tropical interannual climate variations. In addition, predictions are being made of the future evolution of aspects of ENSO using coupled models along with the wind and thermal data collected by the TOGA observing system.

The basic concept exploited in ENSO prediction is that ocean–atmosphere instabilities occur in the tropics at large spatial scales and low temporal frequencies. As first shown by Philander et al. (1984) in a model consisting of a simplified ''shallow-water'' atmosphere coupled to a simplified shallow-water ocean, unstable modes could arise as a result of the coupling of otherwise stable systems. In much the same way that ordinary weather disturbances go through a characteristic cycle of rapid growth and decay, ENSO disturbances have a life cycle with some repeating characteristics. Although the cycle is often irregular, many of ENSO's slowly evolving features can

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

be simulated and predicted by models that include only the tropical Pacific Ocean and the atmosphere above it. Some of the predictability comes from the fact that low-frequency atmosphere—ocean coupling is governed by physical laws that cause certain parts of the ENSO cycle to follow slowly and inevitably after others.

Predictability is a property of a dynamical system, whether natural, mathematical, or mechanical. It can be estimated from a single or multivariate time series produced by the system in question. Ideally, we would like predictability estimates of the coupled ocean–atmosphere–land system on seasonal-to-interannual time scales from direct measurements, but observed time series are much too short for making such estimates reliably. Therefore, direct estimates from oceanic, atmospheric, and land data have to be supplemented by predictability estimates from model systems. The accuracy of these estimates increases with the length of record available. The best predictability estimates can be obtained for the simplest systems. However, the reliability of model estimates as approximations of the coupled climate system's real predictability increases with the number of other features of the real system that are accurately simulated by the model in question. Thus, predictability estimates should be pursued by observational studies across a complete hierarchy of models, from the simplest mechanistic ones to the most highly resolved GCMs. One way of estimating predictability is by making ensembles of predictions.

It is now well understood that the ENSO phenomenon is a coupled phenomenon arising from the interaction of the atmosphere with the ocean, so that dynamical simulation and prediction must be based on coupled atmosphere—ocean models. While simple analog models of coupled atmosphere–ocean interactions (Battisti and Hirst, 1988; Schopf and Suarez, 1989; Cane et al., 1990; Neelin, 1991; Wakata and Sarachik, 1992; Jin and Neelin, 1993) have proven illuminating in interpreting and explaining the grosser aspects of ENSO, more complex coupled dynamical models are needed for more detailed and accurate simulations and predictions.

At present, there are five documented and ongoing routine prediction systems for ENSO, each of which demonstrated skill at least a season in advance. Two of the systems are statistical: one by Barnett and collaborators (Graham et al., 1987a,b), the other by investigators at the Max Planck Institute for Meteorology (Xu and von Storch, 1990). A third method developed at Florida State University (FSU) is a statistical-dynamical technique using only an ocean model (Inoue and O'Brien, 1984, 1986). The fourth method is based on a dynamical coupled ocean–atmosphere model developed at Lamont-Doherty Earth

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

Observatory by Zebiak and Cane that prescribes the mean climatology and predicts the anomalies (Cane et al., 1986; Zebiak and Cane, 1987). The fifth method is based on coupling complex models of the atmosphere and ocean (Latif et al., 1992; Ji et al., 1994). There are additional statistical climate forecasting schemes currently being developed based on neural nets and singular spectrum analysis (Keppenne and Ghil, 1992). Other dynamical forecasting schemes are being developed with coupled GCMs using data ingestion and assimilation methods. These and other forecasts are reported monthly by the NMC in the "Forecast Forum" section of the Climate Diagnostics Bulletin and in the Experimental Long-Lead Forecast Bulletin.

The coupled ocean-atmosphere model of Zebiak and Cane (1987) (hereafter referred to as the ZC model) has produced what are considered to be the longest-range successful numerical forecasts of ENSO events yet produced. The warm phases of the 1986–1987 and 1991–1992 events were predicted more than one year in advance, and skillful hindcasts of previous events support the validity of the approach. The ZC model is an anomaly model in which the annual climatology is specified and the interannual variability is calculated. Zebiak and Cane showed that coupling a diagnostic atmosphere to a shallow-water ocean (with the annual cycle of the coupled system specified) leads to anomalies that resemble ENSO closely, both spatially and temporally. The model has been investigated and analyzed extensively, and the mechanisms for the model ENSO (the "delayed oscillator mechanism") has been described in detail (Battisti, 1988; Battisti and Hirst, 1988; Battisti, et al., 1989). This same model was used to predict the onset of the 1986–1987 warm phase of ENSO a year in advance (Cane et al., 1986). That the ZC model is intrinsically predictable was demonstrated by Goswami and Shukla (1991), who showed that errors in initial conditions grow rather slowly (initial doubling times on the order of 6 months), so that knowledge of the initial state of the coupled system seems to contain foreknowledge of the state of the system up to a year or so in the future. This slow error-growth rate is a direct consequence of the inertia of the ocean in the ENSO cycle. The predictability of the future evolution of the coupled atmosphere–ocean system is embedded in the initial state of the ocean and is realized in the slow evolution of the ENSO cycle.

To initialize the ZC model, the observed monthly FSU-analyzed ship wind field for the Pacific is converted to surface stress specified over the model Pacific. Then, the atmosphere is coupled to the ocean, the system generates its own surface wind and SST fields, and the coupled model is subsequently allowed to run freely. There is an initial error in that the model-generated winds disagree with the im-

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

posed FSU winds at the initial time. Predictions of area-averaged SSTs in the Pacific Ocean are then made for up to 18 months in advance. Predictive skill is highly dependent on the time of year from which the prediction is made, with predictions from the northern spring having the lowest skill (Cane and Zebiak, 1987; Cane, 1991).

The work by Latif and collaborators (1993a) and by Leetmaa and Collaborators (Ji et al., 1994) are the only published examples to date of predictions made by fully-coupled ocean and atmosphere GCMs. Although the mean climate of the coupled model differs considerably from the observed climate, the skill of predicting interannual variations is comparable to that of other models.

It is known that the skill for predicting the phases of ENSO is seasonally varying. The Southern Oscillation is weakly phase-locked with the annual cycle (Rasmusson and Carpenter, 1982; Trenberth and Shea, 1987), with extreme values occurring preferentially at certain times of the year. Transitions between Southern Oscillation Index (SOI) values from one sign to the other occur most often in the April–June period. The structure of the lagged autocorrelation of the SOI shows relatively small persistence until April–May, when it increases substantially with significant values out to about 8 months. The immediate conclusion is that something often occurs in the late northern spring that is independent of ENSO but that can influence its development. Interestingly, Barnett (1983; 1984a,b) showed a strong relationship between the interannual fluctuations of the Indian Ocean monsoon and the Walker circulation, with the coupling of the monsoon and the Pacific trade winds depending on the phase of the annual cycle.

A marked seasonal reduction in skill in predicting SST anomalies in the eastern equatorial Pacific (Cane et al., 1986; Barnett et al., 1988; Latif and Graham, 1992) is almost certainly related to this seasonal structure of the Southern Oscillation, which has been attributed to the somewhat variable and rather rapid onset of the Asian summer monsoon by Webster (1987) and Webster and Yang (1991). If the intensity of the monsoon is predictable (for example, Shukla, 1987) and the influences of the Asian monsoon on the Pacific Ocean and the overlying atmosphere are included in our prediction models, would we see a significant increase in prediction skill for ENSO onset?1

1  

Most ENSO models presently do a poor job of simulating the termination of warm events. Observations show that the termination process begins with a reversal of westerly winds to easterly winds in the far western Pacific, a reversal not included in the Pacific-only models of ENSO. This reversal may be associated with the monsoons.

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

The answer depends on the degree to which air–sea interactions in the maritime continent and western Pacific are well-simulated by the prediction model. Relatively small changes in the way that atmosphere–ocean coupling is parameterized can result in what may be called the SST instability mode (see, for example, Barnett et al., 1991; Latif et al., 1991; Neelin, 1991). This eastward propagating mode has been seen previously in observations (Gill and Rasmusson, 1983) and in some intermediate models (Anderson and McCreary, 1985a,b; Yamagata and Masumoto, 1989). Although relatively slow ocean dynamics dominates, as in the "delayed oscillator" mechanism, this mode appears to be more sensitive to local atmospheric forcing in the western Pacific, where SST is high and coupling is strongest. The relative forcing of these two modes of interannual variation may account for some of the differences between ENSO events and may be important in predicting the strength of the Asian monsoon.

Additional skill in ENSO prediction could arise from the recently made distinction between two modes of variability of the tropical ocean–atmosphere system, quasi-biennial (Trenberth, 1975, 1980; Trenberth and Shin, 1984) and lower-frequency (Rasmusson et al., 1990). The separation of the broad 42-month peak identified by Rasmusson and Carpenter (1982) into these two sharper peaks holds promise for the better physical understanding of possible distinct oscillation mechanisms, as well as for ENSO prediction beyond a year. Barnett et al. (1991) suggested nonlinear interactions between these two modes, Jiang et al. (1992) described the standing and traveling patterns of SST and surface zonal wind associated with them, and Keppenne and Ghil (1992) exploited the greater regularity of the two separate modes to demonstrate very high hindcast skill of the SOI for up to 30 months. Both diagnostic evidence and hindcasting exercises show that large warm or cold events seem to be associated with the two separate oscillations being in phase.

On the basis of the skill and predictability demonstrated by various models, the TOGA Program on Prediction (T-POP) was formed to institute a routine interannual prediction effort. This effort uses the best available model and data assimilation procedure to extend the skill and spatial and temporal range of the predictions, and to do the research and development necessary to accomplish these objectives (see Cane and Sarachik, 1991). T-POP was also formed to take advantage of the skill already indicated by statistical models and by the ZC model (see Barnett et al., 1988), and to exploit to the extent possible the skill inherent in interannual predictions. It was recognized that the ENSO signal offers the best initial hope for predictions, since the existence of a nearly periodic cycle in the tropical

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

Pacific implies a certain predictability once the cycle is ongoing (Sarachik, 1990). In particular, the great potential of subsurface data for the tropical Pacific provided by the TOGA observing system needs to be exploited.

ENSO is not the only important source of seasonal-to-interannual variations. The annual cycle and some important interannual variations of climate are due to processes and phenomena other than those associated with ENSO. For example, variations of rainfall in Brazil and the Sahel are subject to interannual fluctuations that are linked to Atlantic SST variations (Hastenrath, 1990). To achieve the objective of predicting non-ENSO signals in climate, other major climate signals must be identified and understood. The continuum, or background spectrum, of natural climate variability must also be recognized and it must be determined whether these fluctuations are predictable, especially if they influence ENSO predictability.

Much remains to be done to assess the potential for predicting tropical disturbances and their influence on the extratropical circulation. The skill of ENSO forecasts varies markedly with season. The relationship of ENSO to monsoon rainfall over the tropical and subtropical land masses is not well established. Furthermore, the understanding and successful simulation of the extratropical response to the tropical sea-surface anomalies is proving to be a more difficult problem than anticipated at the outset of TOGA. Coupled ocean–atmosphere–land models are proving to be unusually sensitive to errors in component systems, leading to problems in correctly simulating the climatology and the spectrum of variations. In the tropical Pacific, it appears as if the annual cycle is harder to simulate than interannual variations without annual forcing. The annual cycle of SST in the eastern Pacific, for example, seems to be in phase with the heat flux into the ocean, rather than out of phase as in ENSO variations. These heat fluxes also seem to be sensitive to stratus clouds, which induce a positive feedback to SST at low temperatures but a negative feedback at high temperatures. It is becoming even more clear that a complex mix of processes is at work in the annual cycle.

As the complexities of fully coupled GCMs are being explored, considerable effort is being devoted to the problem of data assimilation and initialization. Both atmospheric and upper-ocean data need to be inserted, in a dynamically consistent manner, into a coupled model to gain the best possible estimate of the initial state from which predictions are made. Such a scheme should be multivariate—able to handle data of a variety of types, such as velocity, temperature, and pressure. It should produce estimates of the full three-dimensional structure of the atmosphere and ocean temperature and

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

Figure A-1 Observed (upper) and predicted (lower) 500-mb height-anomaly fields for the warm ENSO event of the northern winter of 1982–1983. The observations are from analysis by the ECMWF; the prediction is an average of three forecasts, made at a lead time of 6 to 8 months, by Bengtsson et al. (1993). Solid contours are associated with positive height anomalies and dashed lines show negative height anomalies.

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×

velocity fields. Finally, it should provide some indication of the accuracy of the estimate of that state.

No complete prediction system using coupled GCMs and a fully coupled data ingestion and assimilation procedure is yet in regular operation, although major pieces of such an operational system have been developed at the National Meteorological Center (Ji et al., 1994). Considerable progress is being made in constructing and evaluating such systems. While this process continues, various hybrid-prediction schemes are showing considerable promise.

A recently developed scheme by Bengtsson et al. (1993) uses a hybrid coupled model (a statistical atmosphere coupled to an ocean GCM) to predict SST in the tropical Pacific 6 to 8 months in advance of the winter season. The SST is then used as a lower boundary condition for a relatively high-resolution atmospheric GCM. Good predictions of the 500-mb height field over the northern Pacific and west coast of the United States result, with potentially useful skill for predicting rainfall over the western part of the United States (see Figure A-1). A similar scheme is under development at the National Meteorological Center.

Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 75
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 76
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 77
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 78
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 79
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 80
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 81
Suggested Citation:"APPENDIX A: PRESENT STATUS OF SHORT TERM CLIMATE PREDICTION." National Research Council. 1994. GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis. Washington, DC: The National Academies Press. doi: 10.17226/4811.
×
Page 82
Next: APPENDIX B: THE 1993 GOALS STUDY CONFERENCE »
GOALS (Global Ocean-Atmosphere-Land System) for Predicting Seasonal-to-Interannual Climate: A Program of Observation, Modeling, and Analysis Get This Book
×
Buy Paperback | $40.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

This book lays out a science plan for a major, international, 15-year research program. The past 10 years have seen significant progress in studies of short-term climate variations, in particular for the region of the tropical Pacific Ocean and the El Nino/Southern Oscillation phenomenon. Some forecast skill with lead times as long as a year in advance has already been developed and put to use. The GOALS program plans to capitalize on this progress by expanding efforts on observations and seasonal-to-interannual predictions to the remainder of the tropics and to higher latitudes.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!