Journal of Applied and Natural Science 2(1):42-47(2010)
Statistical modeling of wet and dry spell frequencies over North-East India
S. Deka1*, M. Borah1 and S.C. Kakaty2
1 Department of Mathematical Sciences, Tezpur University, Napaam,Tezpur- 784028 ( Assam), INDIA
2 Department of Statistics, Dibrugarh University, Dibrugarh-786004 ( Assam), INDIA
*Corresponding author. E-mail: email@example.com
Abstract : In this paper an attempt has been made to develop a discrete precipitation model for the daily series of precipitation occurrences over North East India. The point of approach is to model the duration of consecutive dry and wet days i.e. spell, instead of individual wet and dry days. Various distributions viz. uniform, geometric, logarithmic, negative binomial, Poisson and Markov chain of order one and two, Eggenberger-Polya distribution have been fitted to describe the wet and dry spell frequencies of occurrences. The models are fitted to the observed data of seven stations namely Imphal, Mohanbari, Guwahati, Cherrapunji, Silcoorie, North Bank and Tocklai (Jorhat) of North-East India with pronounced attention to summer monsoon season. The goodness of fit of the proposed model has been tested using Kolmogorov-Smirnov test. It is observed that Eggenberger-Polya distribution fairly fits wet and dry spell frequencies and can be used in the future for an estimation of the wet and dry spells in the area under study.
Keywords : Geometric Distribution, Logarithmic Series, Negative binomial distribution, Poisson distribution, Markov Chain, Kolmogorov-Smirnov test