E. Aldrian Æ L. Du¨menil-Gates Æ D. Jacob Æ R. Podzun D. Gunawan Long-term simulation of Indonesian rainfall with the MPI regional model Received: 17 June 2003 / Accepted: 16 February 2004 / Published online: 13 May 2004 Ó Springer-Verlag 2004 Abstract Simulations of the Indonesian rainfall variability using the Max Planck Institute regional climate model REMO have been performed using three different lateral boundary forcings: Reanalyses from the European Centre for Medium-Range Weather Forecasts (ERA15), the National Centers for Envi- ronmental Prediction and National Center for Atmo- spheric Research (NRA) as well as from ECHAM4 climate model simulation. The result of those simula- tions are compared to station data. REMO simulations were performed at 0.5° horizontal resolution for the whole archipelago and at 1/6° for Sulawesi Island. In general the REMO model, reproduces the spatial pattern of monthly and seasonal rainfall well over land, but overestimates the rainfall over sea. Superiority of REMO performance over land is due to a high-reso- lution orography, while over sea, REMO suffers from erroneously low surface fluxes. REMO reproduces variability during El Nin˜o-Southern Oscillations years well but fails to show a good (wet and dry) monsoon contrast. Despite strong influences of the lateral boundary fields, REMO shows a realistic improvement of a local phenomenon over Molucca. Significant improvement for the step from the relatively high glo- bal 1.125° to 0.5° resolution is noticeable, but not from 0.5° into 1/6°. The REMO simulation driven by ERA15 has the best quality, followed by NRA and ECHAM4 driven simulations. The quality of ERA15 is the main factor determining the quality of REMO simulations. A predictability study shows small internal variability among ensemble members. However, there are systematic intrinsic climatological errors as shown in the predictability analysis. These intrinsic errors have monthly, seasonal and regional dependencies and the one over Java is significantly large. The intrinsic error study suggests the presence of the spring predictability barrier and a high level of predictability in summer. 1 Introduction The Indonesian part of the Asian monsoon has re- ceived less attention in comparison to other Asian monsoon areas like India and China. Studies of the monsoon over the region using global and regional climate models are rare. This study is motivated by the inadequate results of the Max Planck Institute’s high- resolution ECHAM4 global circulation model (GCM) over Indonesia at T106 (1.125°) resolution (Aldrian et al. 2003). Due to the presence of complex topogra- phy and insufficient model resolution in the region, current GCM shows a mediocre performance in simu- lating the precipitation and its variability in the region. The complexity of the region, especially the land sea representation, requires a high-resolution model. Fur- thermore, previous studies showed significant improvements of rainfall representation when a mod- erate resolution was replaced by a higher resolution GCM (Stendel and Roeckner 1998; Jha et al. 2000). Regional modeling (RCM) may allow the high resolu- tion required, but there is no study yet using a regional climate model in the region. The meteorology of Indonesian rainfall has been described recently by Haylock and McBride (2001); Hamada et al. (2002); Hendon (2003) and Aldrian and E. Aldrian (&) Æ L. Du¨menil-Gates Æ D. Jacob Æ R. Podzun Max Planck Institut fu¨r Meteorologie, Bundesstraße 55, 20146 Hamburg, Germany E-mail: aldrian@eudoramail.com Present address: E. Aldrian The Agency for the Assessment and Application of Technology, BPPT, Jakarta, Indonesia Present address: L. Du¨menil-Gates National Science Foundation, 4201 Wilson Blvd, Arlington, VA 22230, USA D. Gunawan Institute of Bioclimatology, Georg–August–Universita¨t, Go¨ttingen, Germany Climate Dynamics (2004) 22: 795–814 DOI 10.1007/s00382-004-0418-9