Copula-Based Method for Multisite Monthly and Daily Streamflow Simulation

Abstract

Multisite stochastic simulation of streamflow sequences is needed for water resources planning and management. In this study, a new copula-based method is proposed for generating long-term multisite monthly and daily streamflow data. A multivariate copula, which is established based on bivariate copulas and conditional probability distributions, is employed to describe temporal dependences (single site) and spatial dependences (between sites). Monthly or daily streamflows at multiple sites are then generated by sampling from the conditional copula. Three tributaries of the Colorado River and the upper Yangtze River are selected to evaluate the proposed methodology. Results show that the generated data at both longer and shorter time scales can capture the distribution properties of a single site and preserve the spatial correlation of streamflow at different locations. The main advantage of the method is that the trivariate copula can be established using three bivariate copulas and the model parameters can be easily estimated by the Kendall tau rank correlation coefficient. The method provides a new tool for multisite flow data stochastic simulation.