Thesis (BS Applied Mathematics) -- University of the Philippines Mindanao, 2010

The major determinant of crop production pattern is the climatic conditions while rainfall is debatably one of the major climatic indicators in the Philippines. Rainfall behavior indicates the success of failure of agricultural operations. In Mindanao, particularly in Davao Province, rainfall distribution is more or less uniformly distributed throughout the year. This study attempted to fit distribution models that significantly describe the probability distribution of rainfall data and determined the shift of rainfall pattern by comparing means during 10 and 20 year interval period of total rainfall. Rainfall observations that covered monthly rainfall total were taken from three different stations; Davao PAG-ASA station (SDC), USEP-PAGASA Agromet Station (USPTC), and PCA-PAGASA Agromet Station (PDC). Three sets of rainfall data (monthly, 10-year interval and 20-year interval) for each station were analyzed and described using the parameter estimation ? Maximum Likelihood Estimation (MLE). The distributions used were the Normal, Log Normal, Gamma, Weibull for minima, and Beta. MLE was implemented using VTFIT computer routine by Cooke (1993). To determine the goodness-of-fit of each fitted model, Chi-square and Kolmogorov-Smirnov tests were applied. The study found out that Gamma, Weibull for minima and Beta were the best-fitting models for the three sets of rainfall observations. Comparison of means showed significant change during 10-year monthly rainfall totals in PDC and USPTC stations. However, the data obtained from SDC station were not sufficient enough to show that there was a shift on rainfall pattern.