Abstract: The purpose of this research is to assess the reliability of hydrological models, two-parameter monthly water balance model for example, in changing environments. First of all the consistency of precipitation and runoff data in 104 typical catchments in the United States is analyzed by Mann-Kendall method; in line with variable fuzzy theory, the impact of the consistency of precipitation and runoff data on the simulation accuracy of the two-parameter monthly water balance model is examined; furthermore the influences of the consistency of precipitation and runoff data and the climatic characteristics of catchment on the model parameters are also investigated. Results show that the consistency of precipitation or runoff data is damaged in 92.31% of the study catchment. Evaluation on the simulation effect of the model with variable fuzzy sets implies that the variation of precipitation and runoff trend weakens the simulation ability of hydrological model. The deterioration of precipitation consistency is the main reason for the weakening of hydrological model simulation ability. Moreover, the variable fuzzy method could accurately identify the secondary factors affecting the simulation ability. In addition, model parameters C and SC both increase with the augment of the annual average runoff coefficient. Parameter C represents the wetting degree of the basin, and parameter SC represents the storage capacity of the basin. The research findings offer technical support for flood control, draught relief, water resources planning and management.

Key words: precipitation and runoff variation, two-parameter monthly water balance model, model parameter, simulation accuracy, variable fuzzy set, consistency