Deterministic Skill of Subseasonal Precipitation Forecasts for the East Africa‐West Asia Sector from September to May

• Published in: Journal of geophysical research. Atmospheres 2019-11, Vol.124 (22), p.11887-11896
• Main Authors: Vigaud, N., Tippett, M. K., Robertson, A.W.
• Format: Article
• Language: English
• Subjects: Atmospheric precipitations; El Nino; El Nino phenomena; El Nino-Southern Oscillation event; Ensemble forecasting; Extreme weather; Forecast skill of dynamical models exceeds that of statistical models based on ENSO and the MJO; Geophysics; Precipitation; Precipitation forecasting; Rain; Rainfall; Rainfall forecasting; S2S rainfall forecasts show statistically significant skill in the East Africa‐West Asia sector out to Week 4; Seasonal variability; Seasonal variation; Seasonal variations; Southern Oscillation; Statistical analysis; Statistical significance; Tropical climate; Tropical‐temperate interactions may represent a new source of S2S predictability; Weather forecasting
• ISSN: 2169-897X; 2169-8996
• DOI: 10.1029/2019JD030747

The East Africa‐West Asia (EA‐WA) sector is a region where the skill in forecasting rainfall is unusually high in the subseasonal‐to‐seasonal (S2S) time range and where ENSO and MJO signals have significant impacts. Much of regional rainfall intra‐seasonal variability is related to tropical‐temperate interactions on synoptic time scales, and we examine whether the skill of dynamical S2S forecasts for the region exceeds the predictability associated with ENSO and the MJO. Deterministic skill (Spearman's rank correlation) of multimodel ensemble forecasts of weekly and Week 3–4 precipitation averages is examined for starts in September–April, based on three ensemble prediction systems (EPS) from the S2S prediction project over the common 1999–2010 reforecasts period. The skill of weekly forecasts decreases with increasing lead but remains statistically significant out to Week 3 and Week 4, a level of skill that has not been previously reported at such leads. The skill of Week 3–4 forecasts is higher than that of Week 3 or Week 4 forecasts. The prediction skill of weekly and 2‐week precipitation averages from ENSO and the MJO is estimated using a multiple linear regression and found to be less than that achieved by the dynamical forecasts for Week 3, Week 4, and Week 3–4 targets. This is evidence for achievable forecast skill that extends beyond typical weather‐scale lead times and that is not simply related to ENSO or MJO signals. The opportunity for skillful predictions is illustrated through successful forecasts up to 4 weeks in advance of extreme rainfall over the Arabian peninsula related to strong tropical‐temperate interactions during weak ENSO and MJO conditions. Key Points S2S rainfall forecasts show statistically significant skill in the East Africa‐West Asia sector out to week 4 Forecast skill of dynamical models exceeds that of statistical models based on ENSO and the MJO Tropical‐temperate interactions may represent a new source of S2S predictability