Thermal comfort estimation in relation to different orientation in mountainous regions in Greece by using artificial neural networks

Abstract

This study focuses on the estimation of thermal comfort conditions in a high alt. site (1455 m) in Apodotia (MA), Greece, by using the MLP neural network model. This estimation is based on the air temperature (t) and relative humidity (f) data of the middle (1078-1163 m) and of the low altitude levels (816-862 m) in relation to different orientations in MA. Also, the MLP model was applied on the top of the Mt. Koromilies (alt. 1350 m) for the estimation of thermal comfort conditions based on the t and f of a site (alt. 797 m) on the northern slope of Mt. Parnassos. In both MA and Parnassos regions the "Cold" and "Comfortable" classes of thermal comfort prevailed. The MLP model provided more satisfactory estimations of the THI values from the t and f of the middle alt. level compared to the respective estimations from the low alt. level. The model provides less accurate estimations of the THI values at 1455 m alt., in the case of sites located near watery surfaces in MA. The application of the MLP model in the case of the Mt. Koromilies, in Parnassos indicated more accurate estimations of the THI values compared to the respective estimations in Apodotia. © 2012 Global NEST.