Seamless Precipitation Prediction Skill in the Tropics and Extratropics

The skill with which two coupled ocean–atmosphere models – the Australian Bureau of Meteorology's POAMA model, and the European Centre for Medium Range Forecasts' monthly forecasting system (ECMWF) are able to predict precipitation over a range of time scales (days to months) is analyzed. The goal is to understand the relative predictability of tropical and extratropical precipitation across a broad range of time scales, from those relevant to short-term weather forecasts to those relevant to seasonal forecasts. This work is thus relevant to the S2S THORPEX follow-on project. For a fair comparison across the seamless range of scales, the verification is performed using data averaged over time windows equal in length to the lead time. In addition to actual skill, potential skill is also computed by taking one ensemble member as “reality” and computing the skill with which the other members can forecast that member.

In POAMA at a lead time of 1 day, actual skill is greatest in the extratropics around 40–60 latitude and lowest around 20 latitude, and has a secondary local maximum close to the equator. The extratropical skill at this short range is highest in the winter hemisphere, presumably due to the higher predictability of winter baroclinic systems. In ECMWF, skill is higher overall compared to POAMA, most markedly so at short lead times, and the subtropical minimum is less pronounced at 1-day lead time, with tropics and extratropics showing similar skill. The local equatorial maximum comes mostly from the Pacific Ocean in both models, and thus appears to be mostly from El Ni~no–Southern Oscillation (ENSO), even at short lead times.

As both the lead time and averaging window are simultaneously increased, the extratropical skill drops rapidly in both models, while the equatorial maximum remains approximately constant, causing the equatorial skill to exceed the extratropical at leads of greater than a few days. At leads longer than 2 weeks, the extratropical skill flattens out or increases, but remains below the equatorial values. Comparisons with persistence confirm that the models beat persistence for most leads and latitudes, including for the equatorial Pacific where persistence is high. The results are consistent with the view that extratropical predictability is mostly derived from synoptic-scale atmospheric dynamics, while tropical predictability is primarily derived from the response of moist convection to slowly varying forcing such as from ENSO. In POAMA, the biggest difference between actual and potential skill is in the tropics at short time scales. The potential skill in the tropics is always greater than that in the extratropics; there is no "cross-over" time, as there was for the actual skill in this model. In contrast, for the short lead times and windows, the ECMWF model's potential skill is greater in midlatitudes than in the tropics. The potential skill in midlatitudes decreases with increasing time lead and window, dropping below that of the tropics beyond 4 days lead time.