Differential Role of External and Internal Forcing in the Prediction of Boreal Summer and Fall Season Indian Ocean SST Variability in ECMWF SEAS5.

The present study analyses the boreal summer (JJA) and fall (SON) season tropical Indian Ocean (TIO) Sea Surface Temperature (SST) prediction skill of ECMWF SEAS5 using 4‐month and 1‐month lead hindcasts. The primary mode of TIO SST during JJA (EOF1), basin mode (IOBM), is captured well with a 4‐month lead time. However, the 4‐month (Feb IC) hindcast overestimates the IOBM variability (70% of the seasonal variability), reducing variability of the second mode (EOF2), dipole mode (IOD), to 8%. This leads to low potential skill (0.45) and insignificant actual skill for EOF2. However, both the JJA IO SST modes are better captured by I‐month (May IC) hindcasts and IOD has a potential skill of 0.84 and an actual skill of 0.62. The co‐occurring El Nino Southern Oscillation (ENSO) induces basin‐wide anomalies in the TIO for Feb IC without a dipole pattern. It leads to the false alarm for IOBM suppressing the IOD‐related pattern resulting in reduced skill. Meanwhile, IOD skill is significant for the boreal fall (SON) season for the 4‐month IC (May IC) hindcast, even with a low ENSO‐IOD relationship. These hindcasts have stronger internal variability inducing dipole structure in IO with warming in the west and cooling in east. However, the ensembles with dominance of internal variability oppose the ENSO‐induced SST anomalies in the northwest Pacific, reducing the external forcing and having low IOD skill for both ICs. Thus, even though ECMWF SEAS5 hindcasts have a skill of more than a season for SON IOD, the over‐predicted influence of internal variability in seasonal mean and weak teleconnection of ENSO limits the skill of IO SST during the boreal summer. [ABSTRACT FROM AUTHOR]