This section reports primarily on the intercomparison of two statistical models (PEAC and CPC which generate predictions of seasonal rainfall for particular Pacific island locations. In the case of the PEAC model, these predictions are limited to the U.S. affiliated islands, while the CPC model includes predictions for several other Pacific island locations as well. In addition, other model results are now becoming available which generally apply to rainfall in tropical Pacific region. For this particular issue, the latest results from the CPC statistical model for all predicted locations is shown below. This map-like presentation applies only to two particular seasons (June - August 2001 and September - November 2001) of the 13 three-month periods out to a year in advance that are available from the model. The full time series in graphs and tables are updated monthly on the internet at: http://www.cpc.ncep.noaa.gov/pacdir/. At present the model covers the following sites, however work is underway to add more:
| 1.Hilo, Hawaii | 12. Wake Island | 23. Hihifo, Wallis & Futuna |
| 2. Kahului, Hawaii | 13. Yap WSO, FSM | 24. Niuatoputapu, Tonga |
| 3. Honolulu, Hawaii | 4. Chunk WSO, FSM | 125.Vavau,Tonga |
| 4. Lihue, Hawaii | 15. Henderson, Solomon Islands | 26. Rarotonga, Cook Islands |
| 5. Andersen AFB Guam | 16. Luganville, Vanuatu | 27. Atnona, French Polynesia |
| 6. Guam WSMO | 17. Koumac, New Caledonia | 28. Bora Bora, French Polynesia |
| 7. Johnston Island | 18. Noumea, New Caledonia | 29. Papeete, French Polynesia |
| 8. Koror WSO, Palau | 19. Funafuti, Tuvalu | 30. Takaroa, French Polynesia |
| 9. Kwajalein Atoll | 20. Rotuma, Fiji | 31. Rikitea, French Polynesia |
| 10. Majuro WSO, Marshall Is. | 21. Undu Point, Fiji | 32. Tubai, French Polynesia |
| 11. Pohnpei WSO, FSM | 22. Nadi, Fiji | 33. Rapa, French Polynesia |
Negative (bold) numbers show negative deviations from normal (less rainfall than normal) while positive (light) numbers show positive deviations from normal (more rainfall than normal). The size of the number (not the value of the number) indicates how accurate the forecast is expected to be, on the average, for the station at the given time of year and the forecast lead time. There are three sizes: the smallest size indicates low skill, the medium size indicates moderate skill, and the largest size indicates a relatively high skill. Forecasts shown in the smallest type size are considered to have little reliability, while those in the largest type size can be trusted more*. The value of the numbers tell how large a deviation from normal is expected. These values are not in millimeters, but are in standardized units that indicate how typical (or nontypical) the rainfall conditions are expected to be relative to the station's normal climate for the season in question. For example, numbers from 0 to 25 (or 0 to -25) are small deviations, indicating conditions that would usually be considered typical of the climate for the station and the time of year. Deviations from 25 to 60 (or -25 to -60) are moderate deviations, indicating somewhat wetter (or dryer) conditions than would be expected for the station and the time of year. Deviations of over 60 (or less than -60) are large deviations, indicating much wetter (or dryer) conditions than non-nal for that location and time of year.
*NOTE - Lower skill forecasts also tend to be for smaller deviations in standardized units - meaning the model tends toward "near-normal" predictions for seasons where skill is low - however, climatological or "near normal" conditions are not necessarily more likely for those seasons ... especially when higher skill forecasts indicate larger deviations for adjacent seasons, which may indicate a prevailing trend. ENSO is generally the dominant influence on tropical Pacific climate. This general pattern is well reflected in these latest results from the model.
- sources: PEAC and NCEP/CPC
