Systematic biases over India in the Met Office Unified Model and their relationship with BSISO index

The long-standing Indian summer monsoon low-precipitation bias in the Met Office Unified Model is here shown to be closely related to the phase of the boreal summer intraseasonal oscillation (BSISO). In particular, in operational weather forecasts, when the observed phase implies increasing monsoon activity there is a much greater bias than when it implies decreasing activity. A moisture budget analysis shows evidence of a link to upstream effects in the form of a delayed reduction in the moisture flux entering India from the west, and also shows that an increase in the net flow of moisture out of India to the east is strongly linked to the low-precipitation bias; the identification of an anticyclonic bias suggests this could be related to a lack of low-pressure systems over India. Some of the aforementioned issues have also been identified in previous studies looking at longer-timescale simulations. To extend our analysis to longer timescales, a similar study on seasonal forecasts is carried out. The seasonal forecasts are shown to have a similar relationship with BSISO phase, during their initial period, to the weather forecasts, confirming that they are suitable for this purpose. Both setups show a reduction in precipitation with forecast lead time, that is worse for some BSISO phases but does occur for all phases. It will be investigated whether this reduction continues with longer lead times for all phases or is only a problem for some of the phases. This study helps to focus efforts in reducing precipitation biases in atmospheric models and works towards a systematic analysis of the problem across timescales. It could also be used to identify periods when precipitation biases will be worse in operational forecasts, as the BSISO phase is generally more predictable than the precipitation itself.