ORIGINAL PAPER Probabilistic seasonal streamflow forecasts of the Citarum River, Indonesia, based on general circulation models Netrananda Sahu 1,5 • Andrew W. Robertson 2 • Rizaldi Boer 3 • Swadhin Behera 4 • David G. DeWitt 2 • Kaoru Takara 5 • Manish Kumar 6 • R. B. Singh 1 Ó Springer-Verlag Berlin Heidelberg 2016 Abstract In this study we discuss probabilistic forecasts of Citarum River streamflow, which supplies 80 % of the water demands in Jakarta, Indonesia, based on general circulation model (GCM) output, for the September– November (SON) season. Retrospective forecasts of pre- cipitation made over the period 1982–2010 with two cou- pled-ocean atmosphere GCMs, initialized in August, are used in conjunction historical streamflow records, with a cross-validated regression model. Pearson’s product moment correlation skill values of 0.58–0.67 are obtained, with relative operating characteristic scores of 0.67–0.84 and 0.74–0.92 for the lower and upper tercile categories of flows respectively. Both GCMs thus demonstrate promis- ing ability to forecast below/above normal streamflow for the Citarum River flow during the SON season. Keywords Probabilistic forecast Á Streamflow Á Exceedance probability Á Skill scores Á Pearson’s correlation Á Citarum River 1 Introduction The condition of water resources in Indonesia has come to a stage where integrated action is needed to reduce over- consumption, losses, and the increasing threat of drought (e.g., Sutardi 2003). Moreover, the demand on water resources has been rapidly increasing because of the national implementations of development programs like irrigation, safe drinking water and industrial water uses in Indonesia. The Citarum River (Fig. 1) provides important sources of water to the two big cities of Bandung and Jakarta and supplies 80 % of the water demands in Jakarta alone; it is an essential resource for urban and industrial development including the economic development and livelihood of the people by supporting agriculture, fish- eries, hydroelectric power generation, public water supply of West Java Province and Jakarta City (e.g., Fares and Yudianto 2004; Boer et al. 2012), and accommodates the need of nearly 10 million people (e.g., Mashudi 2001). The Citarum is the largest river basin in West Java with a catchment area of 12,000 km 2 , stretching 269 km in length. Skillful seasonal streamflow forecasts have not yet been developed for the Citarum, but could potentially provide a powerful tool to water resources managers and planners, who must allocate water supplies in response to the demands of competing water users. Previous studies have shown that accurate and reliable forecasts of streamflow several months in advance can be potentially very important to water resources management systems. For example, Zhao et al. (2015) have evaluated the changing trends and regime shifts of streamflow using long term historical records of the Yellow river basin. Piechota et al. (1998) and Chiew et al. (2003) developed seasonal streamflow forecasts based on ENSO conditions for the eastern Australian river catchments; Fliho and Lall & Netrananda Sahu nsahu@geography.du.ac.in 1 Department of Geography, Delhi School of Economics, University of Delhi, Delhi 110007, India 2 International Research Institute for Climate and Society (IRI), The Earth Institute at Columbia University, Palisades, NY, USA 3 Bogor Agricultural University, Bogor, Java, Indonesia 4 Research Institute for Global Change, JAMSTEC, Yokohama, Japan 5 Disaster Prevention Research Institute (DPRI), Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan 6 Department of Geography, Kalindi College, University of Delhi, Delhi 110008, India 123 Stoch Environ Res Risk Assess DOI 10.1007/s00477-016-1297-4