POAMA 2 (m24abc) MJO Skill

For a climatological forecast of the bivariate RMM anomaly index the RMSE=√2, and thus forecasts are typically deemed to be skilful for RMSE < √2. However, the approach to √2 at long lead time is slow, so we can more confidently say that POAMA-2 provides about a 1-week improvement in skill compared to POAMA-1.5 for lead times beyond about 2 weeks.

Root-mean-square error of the predicted bivariate RMM index for the ensemble mean POAMA-2

Root-mean-square error of the predicted bivariate RMM index for the ensemble mean POAMA-2 (solid) and POAMA-1.5 (dashed) hindcast as a function of lead time (days), for hindcasts initialized 1980-2006.

Bivariate RMM correlation skill for the ensemble mean POAMA-2

Bivariate RMM correlation skill for the ensemble mean POAMA-2 (solid) and POAMA-1.5 (dashed) hindcast as a function of lead time (days), for hindcasts initialized 1980-2006.