Subseasonal to multidecadal variability of northeast monsoon daily rainfall over Peninsular Malaysia using a hidden Markov model

Abstract

This study involves the modelling of a homogeneous hidden Markov model (HMM) on the northeast rainfall monsoon using 40 rainfall stations in Peninsular Malaysia for the period of 1975 to 2008. A six hidden states HMM was selected based on Bayesian information criterion (BIC), and every hidden state has distinct rainfall characteristics. Three of the states were found to correspond by wet conditions; while the remaining three states were found to correspond to dry conditions. The six hidden states were found to correspond with the associated atmospheric composites. The relationships between El Nino–Southern Oscillation (ENSO) and the sea surface temperatures (SST) in the Pacific Ocean are found regarding interannual variability. The wet (dry) states were found to be well correlated with a Nino 3.4 index which was used to characterize the intensity of an ENSO event. This model is able to assess the behaviour of the rainfall characteristics with the large scale atmospheric circulation; the monsoon rainfall is well correlated with the El Nino–Southern Oscillation in Peninsular Malaysia.