چکیده:
Meteorological stations usually contain some missing data for different reasons.There are several traditional methods for completing data, among them bivariate and multivariate linear and non-linear correlation analysis, double mass curve, ratio and difference methods, moving average and probability density functions are commonly used.
In this paper a blended model comprising the bivariate exponential distribution and the first-order Markov chain is introduced for estmating of missing precipitation data. In this method, the day having the missing precipitation record is marked as either wet or dry using the first-order Markov chain and randomly generated numbers. If the Markov chain model marks the day as wet, then a bivariate exponential distribution is used for estimating the magnitute of the missing precipitation datum. Application of the model to the precipitation data from Tehran Mehrabad station shows a good correlation between the statistics of the predicted precipitation data with observed ones.
خلاصه ماشینی:
"In this paper a blended model comprising the bivariate exponential distribution and the first-order Markov chain is introduced for estmating of missing precipitation data.
Karl and Knight (1998) and Brunetti et al.
In the present study, we introduce a blended model based on the bivariate exponential distribution and the Markov chain with randomly generated numbers for estimating of missing data and evaluate its results for the Tehran Mehrabad station.
The magnitude of precipitation during day t at the station k is: :historical precipitation datasets with missing data from a station, Todorovic and (view the image of this page) Woolheiser (1975), Katz (1977) and The conditional probability in a first-order Markov chain is:Woolheiser (1992) used a first order Pr{X t ( k ) = 1 X t −1 ( k ) = 0} = P01 ( k ) (4)Markov chain and randomly generated numbers to define the wet days and the dry days.
The scatter diagram shows estimated daily precipitation using the blended model for days 2 to 6 of April during the period 1958- 1986 versus those observed at the Tehran Mehrabad station are presented in Figure 2.
5. The results show that the blended model is quite successful in estimating the statistical characteristics of the missing daily precipitation in Tehran, It is not, however, successful in predicting the wetness condition of exact dates.
49-55 (view the image of this page) Conclusion Based on the first-order Markov chain and the bivariate exponential distribution, a blended model was introduced for estimating missing precipitation data."
(
, 
, 
)
