Abstract

The frequency factor approach was used for the prediction of the annual maximum rainfall based on the Log Pearson Type III (LP3) and Pearson Type III (P3) distributions while the sample moments and the reduced variate were utilized in the development of the Gumbel Type 1 (EVI) model for Calabar metropolis. The EVI, LP3 and P3 predicted and observed values were compared using regression analysis approach. Twenty-Three years annual maximum rainfall dataset was obtained and analyzed using the Weibull plotting position. The results of the Adjusted R square values, p value, F, t and Durbin–Watson (DW) statistics corresponding to the P3, EVI and LP3 distributions are as follows: Adjusted R2 of 0.999, F value of 4473.175, t=66.882, p<0.01, DW value of 1.033; Adjusted R2 of 0.998, F value of 3548.123, t=59.566, p<0.01, DW value of 1.199 and Adjusted R2 of 0.990, F value of 509.372, t=22.569, p<0.01, DW value of 0.867. The DW statistics value of 0.867 obtained for the LP3 predictions revealed the possibility of the presence of a positive autocorrelation between adjacent residuals. Therefore, the P3 and EVI distributions are more suitable for the fitting of the rainfall of Calabar metropolis compared to the LP3 distribution. The Breusch–Pagan and Koenker tests of homoscedasticity revealed that there is no heteroscedasticity (p<0.05) in the data, hence, the correctness of the explanatory power of our models.

Keywords

rainfall, distribution, frequency factor, comparison