Multiple initialization parameter ensemble as a new ensemble forecasting strategy for improving ENSO prediction in a coupled model

ENSO, one of the most dominant predictable interannual fluctuations in the tropical regions, can influence global climate. Accurate ENSO prediction is propitious to the more skillful forecast of other climate variables. Thus, if ENSO prediction is more skillful at a longer lead time, we can prepare for meteorological disasters in advance and reduce casualties and economic losses. Here the coupled general circulation model (CGCM) ICMv2 is used to investigate the enhancement of multi-initialization parameter ensemble forecasting (MIPE) on ENSO prediction skill. A particular negative feedback parameter that reflects nudging strength of the initialization process is selected. That is dqdt0 (the absolute value of dQ/dT), where Q and T respectively represent the surface heat flux and the model's surface temperature.

Different SST-nudging strengths of the initialization process can generate different initial values, which leads to different ENSO prediction skill. In the hindcast experiments for the period 1981-2010, with all initial values generated from the same initialization parameter, the best performance of ICMv2 occurs at dqdt0=2300, where the anomaly correlation coefficient (ACC) can reach 0.665 at 12 months lead. Increasing the ensemble size is inefficient in improving the skill of single initialization parameter ensemble (SIPE) forecasts. With ensemble members from different SST-nudging strength groups, MIPE forecasts are significantly more skillful than the SIPE forecasts at 1- to 10-month lead time. Our findings suggest that MIPE forecasting is another efficient strategy that can improve ENSO prediction skill besides multi-model and multi-member ensembles using different initial values with random disturbances.