نوع مقاله : علمی - پژوهشی
نویسندگان
1 دانشجوی دوره دکتری، رشته جنگلداری دانشگاه علوم کشاورزی و منابع طبیعی ساری
2 دانشیار دانشکده منابع طبیعی، دانشگاه علوم کشاورزی و منابع طبیعی ساری
3 دانشگاه علوم کشاورزی و منابع طبیعی ساری
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
In this study, the effect of different cluster sampling schemes based on nonparametric algorithms of the artificial neural network model, random forest, support vector machine, and nearest neighbor to estimate the characteristics per hectare and canopy cover of customary Olad Ghobad forests using data Ground and satellite images of the Sentinel 2 were modeled. To estimate the characteristics, 150 clusters in the form of six designs (triangular, square, star 1, linear, L-shaped, star 2) were implemented in a regular-random manner in the region. Then, six cluster sampling designs were collected inside the subplots, density characteristics, and canopy. Each cluster consisted of four sub-plots with an area of 700 square meters (radius of circular sub-plot 15 m and distance between sub-plots 60 m). After pre-processing the images and appropriate processing, the numerical values corresponding to the ground sample parts were extracted from the spectral bands and considered as independent variables. Modeling was performed with 75% of the data and the results were evaluated with the remaining 25%. For both density characteristics (number per hectare) and canopy, artificial neural network method (double star 2 design and linear design with multilayer perceptron algorithm, respectively) with the mean squared values of mean squared error and skew, respectively (10.53, 2.48) and (9.38, 0.33) have a more suitable situation in modeling than other methods used. considering the issue of cost and optimal time, the use of different cluster sampling schemes, nonparametric modeling methods, and Sentinel 2 satellite images yielded good results in estimating .
کلیدواژهها [English]