The IDT System serves to identify an optimum flood frequency model with time dependent parameters from a class of competing models and then to design high flow structures in a changing environment. The maximum likelihood method is used to estimate the parameters of each model and the Akaike Information Criterion for the identification of an optimum model. In all, six probability distribution functions are included, namely, normal, two and three parameter lognormal, Fisher-Tippett I, Gamma, and Pearson III. A time trend can be assumed in the two first cumulants, i.e., (a) in the mean value, (b) in the standard deviation, and (c) in both the mean and standard deviation, keeping the variation coefficient constant in time. It can be either of a linear form, which adds one parameter to the number of parameters to be estimated in a stationary case (S), or a square trinomial form, which adds two parameters to the stationary model. Altogether, it makes up 42 competing models. The system can be used for a time series with random observation gaps without the necessity of filling them in. The IDT System contains the procedure for estimation of a probability distribution together with the standard error not only with respect to one year but to a period of any length as well.The IDT System can be used for any environmental time series as long as its elements can be considered independent and the skew of the distribution being non-negative. Emphasis is put on data requirements, and both model and parameter uncertainties are discussed with the focus on hydrological design under nonstationarity.
Weitere Kapitel dieses Buchs durch Wischen aufrufen
- System of Identification of an Optimum Flood Frequency Model with Time Dependent Parameters (IDT)
W. G. Strupczewski
W. W. Feluch
- Springer Netherlands