Relationship between optimal precursors for Indian Ocean Dipole events and optimally growing initial errors in its prediction

Abstract Using the Geophysical Fluid Dynamics Laboratory Climate Model version 2p1, we explored the precursory disturbances that are most likely to develop into a positive Indian Ocean Dipole (IOD). The dominant spatial patterns of these precursors are defined as the optimal precursors (OPRs) of positive IOD as they are more inclined to cause a positive IOD than other superimposed initial perturbations in the experiments. Specifically, there are two types of OPRs with opposite patterns; the surface component of OPR‐1 (OPR‐2) is an indistinctive west‐east dipole pattern, with a small area of negative (positive) perturbations to the coast of Sumatra and Java. Correspondingly, there is a significant west‐east dipole pattern in the subsurface component of the OPRs, with the largest values located in the eastern equatorial Indian Ocean. The dominant mode of the time‐dependent evolutions of the precursors features rapid development of positive IOD. Furthermore, the OPRs are similar to the optimally growing initial errors (OGEs) associated with IOD predictions that have been presented in previous studies. The shortwave radiation, latent heat flux, and westward Rossby waves play an important role in the time‐dependent evolution of OGEs. Moreover, the large values of the OPRs are located in the same areas as the sensitive areas of targeted observations identified by the OGEs. This infers that intensive observations over these areas would not only reduce initial errors, improve the accuracy of initial fields and decrease the prediction errors but would also detect the precursory signals in advance, which substantially improves the forecast skill of IOD.