Modeling Landscape Aesthetic Values Using Artificial Neural Network Method

Perception of aesthetic concept is one of the most important and required issues in designing recreational areas. Methods of aesthetic evaluation divide into two distinct parts: community perception evaluation and expert evaluation. In expert evaluation methode, valuation is based on visual concepts and objective beauty of landscape through scientific methods, but in community perception evaluation, evaluation is based on the sense experience, values and cognition of the individuals. The goal of this paper is objective assessment of landscape aesthetic values in Ziarat watershed basin using artificial neural network through multi-layer perceptron method. Mapping and standardization of criteria were done after reviewing the available documents, determining the effective criteria and generating the relevant layers for the study. Then, the suitability map of aesthetic values was obtained using the MLP model. The probability image created during the process was used to assess the accuracy through ROC. The ROC index was calculated at 0.875 which endorsed the accuracy of the model. Our experience shows that artificial neural network method is an appropriate tool for landscape aesthetic values modeling in the study area. The results of this model could eventually ease site selection process for recreational areas in more beautiful places.