Stepwise Disaggregation Scheme for Synthetic Hydrology

Disaggregation models are widely accepted tools for generating synthetic hydrologic time series. Attributes and limitations of such models have been reviewed and discussed by several researchers and practitioners. One shortcoming has been related with the size of the matrices involved and the corresponding number of parameters. The main objective of this paper is to analyze and exploit the potential of disaggregation schemes by steps (stages or cascades). It has been shown that nonsymmetric (NSYM) and a mix of symmetric/nonsymmetric schemes are better than other schemes with regard to the size of matrices and the number of parameters. More specifically, the NSYM stepwise lag-one and lag-two models for single-site, and the lag-one model for multisite, disaggregation have been described and applied to modeling and generation of bimonthly flows. The results of the case studies indicate that the proposed parsimonious stepwise disaggregation schemes reproduce quite well the intended variance-covariance properties of the seasonal flows.