A Seasonal Multivariate Multilag Auto-Regressive Model in Hydrology 'Source Title' Modeling Hydrologic Processes, Proceedings Of 3rd International Hydrology Symposium
Bibliographic details
Summary
The multivariate lag one auto regressive model suggested by Matalas (1967) is extended to be applicable for seasonal hydrologic sequences. The model has periodic matrix parameters to account for the mean, standard deviation, lag one up to lag m serial correlation and lag zero up to lag m cross correlation, with m the order of the auto regressive model. General equations are developed for estimating the periodic matrix parameters of the model from the data at n sites. The estimation procedure is presented in two ways: first estimating the parameters by the method of moments and second by maximizing the likelihood function of the errors given the moment estimators as a starting set. Since the first and the second order models are the most commonly used in hydrology, some of the parameter estimators are given in explicit form. Also, restrictions on the matrix parameters of these models are indicated for cases that may be of practical relevance.