Chain-dependent Markov correlation pulse model for daily streamflow generation

A stochastic model for the temporal and spatial description of daily streamflow at multiple sites is presented. It is an extension of a chain-dependent process commonly fitted to daily precipitation series. The extension involves allowing daily pulses to be auto- and cross-correlated at temporal and spatial scales. The chain-dependent Markov correlation pulse (CDMCP) model, consisting of a two-state, first-order Markov chain for occurrences and a Markov correlation model for nonzero pulse height, is extended to simultaneous pulse generation at multiple sites. The pulse model coupled with a variable response function was applied to the daily streamflow generation for Wuper River in Germany, and was satisfactory in preserving the short-term structure of the occurrence process, as well as part of the statistics of the seasonal streamflow. The use of Markov model yields good agreement of autocorrelation in the synthetic pulse and streamflow series.