Comment Comment on 'Wang W, Ding J. 2007. A multivariate non-parametric model for synthetic generation of daily streamflow. Hydrological Processes 21: 1764-1771'

摘要

Wang and Ding (2007) present a daily streamflow simulation model based on the Gaussian kernel function. The model is found powerful in preserving statistical characteristics of the daily streamflow sequences. The purpose of this comment is to improve the presentation of the proposed model. To my understanding, the following three points were missed by Wang and Ding (2007): 1. Four kinds of stochastic hydrological models are mentioned and three of them are listed by Wang and Ding (2007) as regressive models, disaggregation models, and shot noise (SN) models. Experiences on the daily streamflow simulation studies (Aksoy and Bayazit, 2000; Aksoy, 2003, 2004a,b) show that Markov chain (MC) models play an important role in the daily streamflow modelling. Therefore, they should be listed along with the models mentioned above and not be considered among the others. Numerous studies are available in the hydrological literature using MCs either as single- or multi-site schemes to simulate the daily streamflow sequences for either perennial or intermittent streams (Treiber and Plate, 1977; Sargent, 1979; Kottegoda and Horder, 1980; Kron et al., 1990; Xu et al., 2001 a,b, 2003; Szilagyi et al., 2006).