Studies of the probability of rainfall and its spatial and temporal variation are important in the planning of agricultural activities and water resources engineering. However, statistical analyzes related to rainfall find limitations regarding the size of the available historical series, which are mostly insufficient or have a large number of faults. A good alternative to overcome these limitations is the generation of pluviometric series through the use of stochastic models. In this sense, the objective was to elaborate a methodology to determine the probability of occurrence of dry and rainy days and to estimate daily average rainfall. Thus, the determination of the occurrences was done using first order Markov Chains and two states and, for the quantities, the cumulative probability distributions Gamma and Weibull were used, whose parameters were estimated by both the Maximum Likelihood Method and By the Method of Moments. The developed model was applied to 80 rainfall stations distributed in the Tapajos River Basin (TRB). The results of the probabilities of occurrence of dry and rainy periods defined for the TRB the dry season from May to September and the rainy season from October to April. The elements of the probability transition matrix and the alpha and beta parameters showed variability in relation to time and, in addition, the influence of the geographic position of the rainfall station on the determination of dry and rainy periods in specific localities. The validation of the model was performed using the Kolgomorov-Smirnov adhesion test, which demonstrated that the average daily rainfall can be estimated with good performance through the first order Markov Chain and two states with the Gamma and Weibull distribution to two Parameters. However, the Gama distribution stood out in the estimation of average daily rainfall for most of the months of the year, except for the months of March, July and December, for which the Weibull distribution proved to be efficient.
|
