A modelling framework for a better understanding of the tropically-forced component of the Indian monsoon variability

Abstract

An experimental technique is introduced for controlling atmospheric diabatic heating in a fully-coupled climate model (NCEP Climate Forecast System version 2), with the motivation of better understanding the role of tropical forcing in modulating the Indian summer monsoon and contributing to global seasonal prediction skill. The ‘added heating’ mechanistic approach has the benefit of not interfering with any internal model feedbacks. The approach is applied in an iterative fashion to correct bias in the three-dimensional tropical heating over the Indo-Pacific, defined as the difference between the model ensemble mean heating and that estimated from ERA-Interim. The June through September seasonal mean, trend and parabolic fit in diabatic heating are corrected for each year separately over a 20-year set of seasonal re-forecasts, resulting in a 60–90% reduction in the mean-squared error of Indo-Pacific heating from an initial set of control re-forecasts, two-thirds of which is associated with the climatological bias. This results in higher skill in the climatological mean and inter-annual variability of local low-level winds and the underlying sea surface temperature (SST). Improvements are most significant over the equatorial Indian Ocean with a removal of the climatological low-level westerly bias and SST gradient bias along with substantially more skill in inter-annual variability. Although there is very little improvement in the model rainfall over India, there is improvement in the low-level circulation. The lack of significant improvement in rainfall prediction may be partly related to the representation of convection and/or the coarse resolution grid of the model.