Impact assessment of Indian Ocean Dipole on the North Indian Ocean tropical cyclone prediction using a Statistical model

This study examined the predictive skill of a statistical Generalised Additive Model (GAM) by considering the impact of the Indian Ocean Dipole (IOD). The proposed technique is powerful but simple and considers both the linear and non-linear relations hidden in the data. The model considers tropical cyclogenesis through kernel density estimation, trajectories by velocity field, and landfall through a country mask approach. A lead-lag analysis for TC forecast potential confirms that the IOD is a good predictor for 2-month lead forecast. Result shows that TC occurrence increase (decrease) during the negative (positive) IOD events and that the landfall probability vary for each IOD phase. Altering convection, steering flow, and low-level vorticity influence the NIO TC activity. Result also highlights the importance and potential of the GAM approach (approximately 72% skill) in matching predicted landfall with the observations.

Automated summary

This study examined the predictive skill of a statistical Generalised Additive Model (GAM) by considering the impact of the Indian Ocean Dipole (IOD). The results show that IOD is a good predictor for tropical cyclone activity and landfall probability. The GAM approach has a potential skill of approximately 72% in matching predicted landfall with observations.