Abstract

Understanding the meteorological characteristics helps in predicting the weather conditions; for example, the open burning by local farmers in the South East Asia caused adverse weather conditions in which hazardous haze affect the health conditions of the population. By looking at the multivariate weather variables such as wind speed, relative humidity, pressure, temperature at dew, temperature at dry, geo-potential meter, height above mean sea level and location, the dimensionality of the data is reduced to give a simpler understanding of the data. A Matlab program is written to perform the principal component analysis. Using diagrammatical outputs from scree plot, biplot, three dimensional scatter plot and loading plot, it is found that six components are needed to represent about 83% of the total variance of all components in the multivariate datasets obtained at the Kuala Lumpur International Airport and Bayan Lepas Airport at three different pressures. For the Bayan Lepas Airport Station, we found some modest negative correlation between the geo-potential variables. The components can be described as the variation of geo-potential at all levels, relative humidity at all levels and variables at 1000 level.

Key words: Principle component analysis, wind data, MATLAB, multivariate variables.