Simulation of precipitation fields from probabilistic quantitative precipitation forecast D.-J. Seo * ,1 , S. Perica, E. Welles, J.C. Schaake Hydrology Laboratory, Office of Hydrologic Development, National Weather Service, 1325 East-West Highway, Silver Spring, MD 20910, USA Received 19 November 1999; revised 12 July 2000; accepted 21 September 2000 Abstract To assimilate Probabilistic Quantitative Precipitation Forecast (PQPF) into ensemble hydrologic forecasting, a procedure is needed that generates ensemble traces of future precipitation fields from PQPF. This paper describes one such procedure used in the National Weather Service’s (NWS) Ensemble Precipitation Processor (EPP). Given PQPF for the first 24 h valid at the forecast scale (approximately 64 × 64 km 2 ; the procedure generates ensemble traces of 24- or 6-h precipitation at the simula- tion scale (approximately 4 × 4 km 2  in space or in space and time, respectively. The steps involved are quasi-analytical downscaling of PQPF and conditional simulation of precipitation fields based on optimal linear estimation techniques that explicitly account for both intermittency and inner variability. To evaluate the procedure, radar-based precipitation data from the Ohio River Forecast Center (OHRFC) are used to develop a climatological PQPF, and to compare ensemble statistics between simulated and observed precipitation. Published by Elsevier Science B.V. Keywords: Probabilistic; Quantitative precipitation forecast; Ensemble; Simulation; Precipitation 1. Introduction Among the initial and boundary conditions necessary for hydrologic models used in opera- tional river stage forecasting, Quantitative Precipi- tation Forecast (QPF) is very often the biggest source of uncertainty. In an attempt to account for uncertainties in QPF, and subsequently those in river stage forecast, the National Weather Service (NWS) is experimenting with probabilistic river stage forecasting based on Probabilistic Quantitative Precipitation Forecast (PQPF) (Krzysztofowicz, 1998). As described in the Section 2, PQPF refers to a set of variables that collectively specify, fully or partially, the probability distribution of future precipitation in space and time at some scale of aggregation. To obtain Probabilistic River Stage Forecast (PRSF) from PQPF, two approaches are currently pursued in NWS; the derived distribution and the ensemble. The former is due to Krzysztofowicz (1998, 1999), and amounts to solving (quasi- analytically via the total probability law) for the conditional probability distribution of future river stage given the initial and boundary conditions (including future precipitation parameterized in the form of PQPF: see Section 2). The key to the solution is to decompose the above conditional Journal of Hydrology 239 (2000) 203–229 0022-1694/00/$ - see front matter Published by Elsevier Science B.V. PII: S0022-1694(00)00345-0 www.elsevier.com/locate/jhydrol * Corresponding author. Tel.: +1-301-713-0640; fax: +1-301- 713-0963. E-mail address: dongjun.seo@noaa.gov (D.-J. Seo). 1 Present address: Department of Civil and Environmental Engi- neering, University of Utah, USA.