Prediction and predictability of tropical intraseasonal convection: seasonal dependence and the Maritime Continent prediction barrier
Prediction and predictability of tropical intraseasonal convection in the WMO subseasonal to seasonal (S2S) forecast database is assessed using the real-time OLR based MJO (ROMI) index. ROMI prediction skill in the S2S models, as measured by the maximum lead time at which the bivariate correlation coefficient between forecasts and observations exceeds 0.6, ranges from ~ 15 to ~ 36 days in boreal winter, which is 5–10 days higher than the MJO circulation prediction skill based on the MJO RMM index. ROMI prediction skill is systematically lower by 5–10 days in summer than in winter. Predictability measures show similar seasonal contrast in the two seasons. These results indicate that intraseasonal convection is inherently less predictable in summer than in winter. Further evaluation of correlation skill assuming either perfect amplitude or perfect phase forecasts indicates that phase bias is the main contributor to skill degradation at longer forecast lead times. Nearly all the S2S models have lesser skill for target dates in which the MJO convection is centered over the Maritime Continent (MC) in boreal winter, and phase bias contributes to this MC prediction barrier. This issue is less prevalent in boreal summer. Many S2S models significantly underestimate ROMI amplitudes at longer forecast leads. Probabilistic evaluation of the S2S model skills in forecasting ROMI amplitude is further assessed using the ranked probability skill score (RPSS). RPSS varies significantly across models, from no skill to more than 30 days, which is partly due to model configuration and partly due to amplitude bias. Accounting for the systematic underestimates of the amplitude improves RPSS.
This research has been conducted as part of the NOAA MAPP S2S Prediction Task Force and supported by NOAA Grant NA16OAR4310076. SW and AHS also acknowledge support from NSF AGS-1543932 and ONR N00014-16-1-3073. We are grateful for the insightful comments by three anomalous reviewers. We thank Haibo Liu for obtaining and organizing the S2S data set from the ECMWF data portal.
- Gottschalk J, Wheeler M, Weickmann K, Vitart F, Savage N, Lin H, Hendon H, Waliser D, Sperber K, Prestrelo C, Nakagawa M, Flatau M, Higgins W (2010) A framework for assessing operational model MJO forecasts: a project of the CLIVAR Madden–Julian oscillation Working Group. Bull Am Meteorol Soc 91:1247–1258CrossRefGoogle Scholar
- Liebmann B, Smith CA (1996) Description of a complete (interpolated) outgoing long-wave radiation dataset. Bull Am Meteor Soc 77:1275–1277Google Scholar
- Wilks DS (2011) Statistical methods in the atmospheric sciences, 3rd edn. Academic, San DiegoGoogle Scholar
- Xie YB, Chen SJ, Zhang IL, Hung YL (1963) A preliminarily statistic and synoptic study about the basic currents over southeastern Asia and the initiation of typhoon. Acta Meteorol Sin 33(2):206–217Google Scholar