Standard software of Automatic Weather Stations allows for measured parameters recording with a high frequency. It leads to the registration of superfluous ballast of unnecessary data, gross of which probably will not be used. This data collection unnecessary burdens the memory of the data logger or other similar devices. Parameters of chosen functions are sufficient for the data recording in case of model functions, which allow for reconstruction - with a set up accuracy - variability of a given parameter in accepted time series, are found. Results of daily variability modelling (on the base of hourly values) of air temperature and soil temperature for a few depths were presented at the paper. The results of measurements from the years 2000-2002, obtained from the registration of CR23X Automatic Weather Station of Campbell Sci. Ltd, installed at the AU Agro- and Hydrometeorological Observatory in Wroclaw, were used. In case of stabilized weather the good approximation result may be obtained with the use of polynomials functions of higher orders. For example in the case of very typical soil and air temperature variability, their course is described enough good by the polynomial of the 3rd order (4 values are sufficient for the data registration). All the calculations were made with the use of Statistica (ver.5) computer program.