19. Neural Network Modeling of Relative Humidity and Temperature Distribution Over Nigeria

by 1Ibeh G. F, 2Ibeh L. M., 3Ogbonna R. C, 4Ofoma J. N and 5Akande S
1 Department of Physics with Electronics, Evangel University Akaeze Ebonyi State, Nigeria;
2 Department of Geography and Environmental Sciences, Ludwig-Maximilians, University of Munich, Germany
3,4 Department of Computer Science and Mathematics, Evangel University Akaeze, Ebonyi State, Nigeria;
5 Centre for Space Research and Application, Federal University of Technology Akure, Nigeria

Abstract
This paper used neural network model to study the distributions of relative humidity and temperature over Nigeria. The theoretical explorations of the relationship between the parameters were reviewed. This study was carried out on thirty-six point stations over Nigeria. Temporal variations of estimation and prediction of relative humidity and temperature were carried out in this study. The results revealed that temperature and relative humidity distributions over Nigeria are in variant. They were inversely proportional to each other as affirmed by other researched. Spatio-temporal variations revealed that relative humidity is higher in wet seasons compared to dry seasons in Nigeria. It is also higher within the Southern part of Nigeria as a result of coastal nature, moisture content in the atmosphere of the region and low temperature gradient. The influences of temperature on relative humidity were study. The results shows inversion with the rates of relative humidity and the rates of temperature both in wet and dry seasons, and within the Southern and Northern part of Nigeria. This could be because increase in temperature raises saturated vapour pressure, which leads to reduction in the relative humidity. The variation of temperature may be due to diabatic heating and adiabatic effects in the atmosphere. Based on the similar signatures of the estimated and observed temporal distributions of the parameters, forecast of two years ahead of the years of the study were successfully achieved. The result showed strong negative relationship between temperature and relative humidity. The correlation coefficient is calculated to be -0.94. This strong negative correlation signifies that as the temperature decreases, the relative humidity increase (and vice versa). The performance of neural network model in the distributions shows the ability of the model in studying atmospheric parameters as confirmed by other researchers.

Keywords: Relative Humidity, Temperature, Neural Network, Diabatic, Adiabatic and Nigeria