Impact of the observed SST frequency in the model initialization on the BSISO prediction

Abstract The impact of the observed sea surface temperature (SST) frequency in model initialization on prediction boreal summer intraseasonal oscillation (BSISO) over Western North Pacific (WNP) is investigated using Beijing Climate Center System Model. Three sets hindcast experiments initialized by monthly, weekly and daily SST data (referred to as Exp_MSST, Exp_WSST Exp_DSST, respectively) are conducted with 3-month integration starting from 1st, 11th, 21st day each month June–August during 2000–2014, respectively. results show that useful skill BSISO index reaches out about 10 days further increases 1–2 Exp_DSST. differences among various especially apparent forecast time 6–20 days. Focusing strong cases this period, activity its related moist static energy (MSE) characteristics WNP diagnosed. It found Exp_MSST enhanced associated more realistic variations MSE tendency. Among budget terms dominate tendency, latent heat flux advection evidently improved, reduction mean biases than 21% 10%, Therefore, better reproduced variation may contribute skillful through improving evaporation well atmospheric convergence divergence activity. Our findings suggest necessity increasing (i.e., monthly or daily) process coupled models enhance actual predictability, since some current subseasonal operations researches still use relatively low-frequency observations for initialization.