Air temperature and dew point temperature are two of the important atmospheric variables that affect the growth rate of plants as well as many other processes in agricultural and ecological systems. Extremely low air temperature and dew point temperature are harmful to the crops and might cause severe economic losses. Therefore, accurate predictions of air temperature and dew point temperature are necessary in order to prevent crops from being damaged by severe frost. Previous studies developed artificial neural network (ANN) models to predict air temperature and dew point temperature from one to twelve hours in advance. The goal of the research herein was to develop more accurate air temperature and dew point temperature prediction models. This research incorporated evolutionary approaches in the development of ANNs to refine the selection of input prior data for each applicable atmospheric variable.