Weather typing or categorization continues to be popular and numerous methods have been developed over the past century for this investigation. This paper attempts to classify the types of weather for Karachi on the basis of diurnal data according to seasonal classification. The idea of using weather classification in climate change research is inspired by both uncertainties accompanying climate simulation on a regional scale, and the conflicting results of examining long-term instrumental series by traditional statistical methods. Multivariate techniques of Principal component analysis and Cluster analysis are used to obtain different types of weather for each season separately. Calculations show that we have 14, 5, 8, 7 and 8 different weather types for Monsoon, winter, summer, spring and autumn seasons respectively. Noticeable greater variations came into view for monsoon season and winter appears as least varied season. The aim of this research is to investigate the properties of the adopted clustering procedure, the consequences of the modifications introduced, and the physical interpretation of the weather types in terms of meteorological variables.