Distribution mapping of forest types in Ziarat Forestey Plan using parametric and nonparametric algorithm

• Main Authors: Mojtaba Amiri, Mohsen Mostafa, Mohammad Rahimi
• Format: Article
• Language: fas
• Published: Iranian Society of Forestry 2019-08-01
• Series: مجله جنگل ایران
• Subjects: logistic regression; artificial neural network; physiographic factors; climate factor
• Online Access: http://www.ijf-isaforestry.ir/article_93394_0ac34c3c8463206f5098c358d5898e7f.pdf

Due to the interaction of Tree species and its environment, descriptions and analysis of forest types are necessary. The aim of present study was to evaluate modeling distribution of forest types using parametric and nonparametric algorithm. Current research was carried out in Ziarat forestry plan, Golestan province, Iran. 556 samples were taken to measure the quantitative parameters of trees including tree height, diameter at the breast height and type of species via Systematic- Randomize pattern with 150×200 m. After that, the forest types have been determined according to frequency of species. Subsequently, the map of forest types have been produced using Physiographic factors (elevation, slope and aspect), Climate factor (rain fall, evaporating and temperature) via Parametric algorithm (Logistic Regression (LR)), Nonparametric algorithm (Artificial Neural Network (ANN)). The results showed that based LR and ANN, the largest area of forest type was observed in Fageto - Carpinetum with Parrotia persica (23.32%) followed by Fageto –Carpinetum (24.69%). In both methods, the elevation and rainfall events have been recognized as impotent factors. Regarding the limitation of input data and complexity of forest ecosystem, the result of LR and ANN are acceptable. Generally, ANN was more effective compared to LR. However, both algorithms are recommended in distribution mapping of forest type.