Utilizing the AMeDAS weather data a mathematical model to estimate atmospheric temperature variation of Kinki district is proposed for the purpose of designing thermal performance of buildings and urban area. In the process of the model building temperature record is divided into a deterministic and a stochastic component, further the deterministic component is decomposed into year average, yearly and diurnal periodic subcomponents. In this study the modeling of deterministic component is proposed. The model is built based on multiple regression approach with the predictors ; latitude, height, solar radiation and land use of a weather station. The model was validated on the basis of air-conditioning load.