ATMOSPHERIC SCIENCE LETTERS Atmos. Sci. Let. 9: 237–244 (2008) Published online 21 November 2008 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/asl.196 Regional scale prediction of the onset phase of the Indian southwest monsoon with a high-resolution atmospheric model Dodla Venkata Bhaskar Rao,* Desamsetti Srinivas and Satyaban Bishoyi Ratna Department of Meteorology and Oceanography, Andhra University, Visakhapatnam, India *Correspondence to: Dodla Venkata Bhaskar Rao, Department of Meteorology and Oceanography, Andhra University, Visakhapatnam-530 003, India. E-mail: dvb 1949@yahoo.com Received: 6 June 2008 Revised: 27 August 2008 Accepted: 29 August 2008 Abstract A nonhydrostatic atmospheric model with a resolution of 30 km is used to make predictions of the rainfall during the onset phase of the southwest monsoon (SWM) of 2003. Model predictions of the pentad rainfall time series indicate good predictions up to lead time of 5 days. The correlation coefficients (CCs) between the model-predicted and observed rainfall at different locations, representative of the five homogeneous regions of SWM rainfall, over the Indian subcontinent show correlations significant at 90% level up to 5 days lead time with values above 0.32. The spatial distribution of the model-predicted pentad rainfall show an advancement of the Arabian Sea and the Bay of Bengal branches of SWM over the Indian subcontinent up to 5 days lead time. Copyright  2008 Royal Meteorological Society Keywords: monsoon; onset phase; regional prediction; mesoscale models 1. Introduction Prediction of the monsoon rainfall is important due to its impact on agrarian economies of countries like India. About 70% of the Indian subcontinent receives 60–80% of its annual rainfall during the southwest monsoon (SWM) season i.e. June–September. Of the different phases of SWM, the onset phase is charac- terized by sharp increase in pentad (5-day accumu- lated) rainfall (Rao, 1976). The onset of the SWM is associated with a sudden, substantial, and sustained increase in rainfall over a large scale with spatial coherency and with rainfall amounts increasing from below 5 to over 15 mm per day (Soman and Kumar, 1993). However, the prediction of rainfall associated with the onset phase is of utmost importance due to its need for agriculture and other socioeconomic applications. It is established that the onset of SWM occurs through its Arabian Sea and Bay of Bengal branches striking the Kerala coast (southern tip of India) and Northeast India (Assam and neighborhood) around 1st of June. According to climate normals, the Arabian Sea branch touches the southern tip of the Indian peninsula by 30 May and progresses north- ward to cover Gujarat by 30 June, whereas the Bay of Bengal branch touches the Northeast India around 2 June and progresses westward covering western Uttar Pradesh by 30 June, and with the monsoon establish- ing over the Northwest India around 15 July (Rao, 1976; IMD, 1990; Soman and Kumar, 1993; Hasten- rath, 1994; Webster et al., 1998). Though the mean onset dates of the SWM are available, due to their large variability, precise prediction of the onset of the SWM is important for agricultural planning. A vari- ety of synoptic and statistical methods were proposed to predict the onset of SWM over the Kerala coast (Ananthakrishnan and Soman, 1988; Fasullo and Webster, 2003 and Joseph et al., 2006). Global atmo- spheric general circulation models provide predic- tion of the atmospheric circulation associated with the Indian SWM but have a limitation of predicting the regional characteristics due to their coarse res- olution. Attempts were made to simulate the Indian SWM and the seasonal rainfall characteristics using regional climate models which have a horizontal reso- lution of 30–50 km (Bhaskaran et al., 1996; Bhaskar Rao et al., 2004). The models use parameteriza- tion schemes for the different subgrid scale phys- ical processes. For the prediction of the monsoon rainfall associated with stratiform cumulus convec- tion, the representation of convective precipitation is important. Bhaskar Rao (2008) evaluated the per- formance of five different convection parameteri- zation schemes (Grell, Anthes–Kuo, Betts–Miller, Kain–Fritsch, Kain–Fritsch2) in the short range pre- diction of SWM rainfall during its onset phase over the Indian subcontinent. The results indicate that the Grell scheme of convection parameterization shows best performance followed by Betts–Miller and Kain–Fritsch2 and that of Anthes–Kuo is the lowest. In this study, an attempt is made to predict the progress of the Indian SWM during its onset phase (i.e.) of June-2003 at different lead times using National Center for Atmospheric Research Mesoscale Mode Version 5 (NCAR MM5) model. The model is integrated for a limited area covering the South Asian region and so requires the specification of the bound- ary conditions. In the present study, the time varying lateral and surface boundary conditions are provided from National Centers for Environmental Prediction Copyright  2008 Royal Meteorological Society