علی علیزاده

To plan in order to increase the efficiency, speed of action and reduce the cost of statistics, it is necessary to examine and use different sampling methods with variable area such as linear or transect. This study was chosen in order to investigate the effect of forest classification and compare the accuracy of estimators in the transect sampling method in Safaroud forests of Mazandaran with an area of 176 hectares. First, using the statistical method of 100% of all the trees in the field, then the linear method (Transect, the length of the sample line was 30 meters) was done in two stages. The first stage was in an area of 176 hectares without classification with a grid size of 100x100. meter was measured. The second stage of the forest area under investigation using the "Neyman" classification method into three categories in terms of volume inventory including less than 200 cubic meters, between 200-250 cubic meters and more than 250 cubic meters per hectare and according to the classification method A linear (transect) was made using the new grid dimensions for each stratum. In this study, the effect of segmentation on accuracy and time was investigated using the criterion (E%2×T). The results of this study in terms of the accuracy and cost index show that by segmenting the forest and implementing linear sampling, the estimated average of the investigated characteristics is 22.8% closer to the real average and the error rate is higher than the standard. Statistics have decreased. From the results of this study, it is concluded that the estimation of forest characteristics such as the number and volume per hectare is possible with the sampling method with linear variable area (Transect) for heterogeneous and dense forests in the north. However, due to various reasons such as the bushiness and density of trees and the interference of their crowns, the results of linear sampling, such as the sample line, which is based on the collision of the trunks or the image of the crowns of the trees, is associated with sampling error, and the average of the estimation results is with the actual amount of that distance has it.

The continuous need to improve statistical methods has led to the use of methods with minimum bias and minimum variance in forest statistics. to plan in order to increase the efficiency, speed of action and reduce the cost of statistics, it is necessary to examine and use different sampling methods with variable area such as linear or transect. This method is one of the specialized methods that ecologists can use to estimate the density or the total number of plants in a studied area. In the heterogeneous forest, to increase the accuracy of sampling, the forest can be classified in terms of stock. Classified estimation is a statistical technique that can reduce the variance of estimates without increasing the sample size.

This study was chosen in order to investigate the effect of forest classification and compare the accuracy of estimators in the transect sampling method in Safaroud forests of Mazandaran with an area of 176 hectares. The selected arenas were measured and recorded using the statistical method of 100% of all the trees in the areas (counting limit, diameter equal to chest more than seven and a half centimeters). In order to avoid mistakes and make it easier to perform 100% statistics, the studied area was divided into square pieces with dimensions of 50x50 meters and 704 plotes were placed in the field. Then the trees in each of these parts were measured separately. The linear transect method was carried out in two stages with a sample line length of 30 meters. The first stage was measured in an unclassified area of 176 hectares with a grid size of 100x100 meters. The second stage of the forest area under investigation using the method of "Neyman" classification into three categories in terms of volume inventory including less than 200 cubic meters, between 200-250 cubic meters and more than 250 cubic meters per hectare and based on The linear transect method was performed using the new grid dimensions for each stratum. In this study, the effect of segmentation on accuracy and time was investigated using the criterion (E%2×T). For this purpose, the time required to unload the sample plotes for the necessary measurements inside the sample plot and the time required to travel the distance between the adjacent sample plotes were also calculated and recorded using a stopwatch. In the present study, in addition to comparing the accuracy of statistical estimators, time studies were also conducted. In order to perform calculations related to time studies, the index value (E%2×T) was used. The effect of segmentation on accuracy and time was investigated using the criterion (E%2×T).

The results of the comparison of the statistical test between 100% statistical methods and Transect linear sampling without classification and with the implementation of classification showed that there is a significant difference at the level of 0.05%.With the implementation of stratification in the linear sampling method, the amount of statistical error has decreased and the estimated average of the examined characteristics has become closer to the real average. But the difference between the results and the actual amount is still significant. In other words, the reduction of statistical errors after classification for the characteristic of number per hectare by 3.37% and the reduction of 1.14% of statistical errors for the characteristic of cross-section width and 2.96% reduction of statistical errors for the characteristic of volume per hectare has not been able to improve the calculation results. put in the range of the real average. The results of this study in terms of the accuracy and cost index show that by segmenting the forest and implementing linear sampling, the estimated average of the investigated characteristics is 22.8% closer to the real average and the error rate is higher than the standard. Statistics have decreased. From the results of this study, it is concluded that the estimation of forest characteristics such as the number and volume per hectare is possible with the sampling method with linear variable area (Transect) for heterogeneous and dense forests in the north. However, due to various reasons such as the bushiness and density of trees and the interference of their crowns, the results of linear sampling, such as the sample line, which is based on the collision of the trunks or the image of the crowns of the trees, is associated with sampling error, and the average of the estimation results is with the actual amount of that distance. has it.

This study was conducted to determine the optimal systematic sampling dimensions in forest using Variogram analysis in geostatistics in Catchment Area 30, Safarood forest of Ramsar, with a total area of ​​100 ha. Sampling was done within a systematic grid (200m×150m) with randomly starting point. To determine the correct spatial correlation between specimens at short distances, from the intersection of the sides of the grid, two other samples were taken at a distance of 50m in two main directions from the central sample plot. Each piece of the sample consisted of two common-centered circular area of ​​300 and 700m2. The analysis of the selected model for fitting the experimental variogram of the total reference sample plots for the variables of growing stock and basal area/ha indicates the existence of a strong spatial structure and a significant relationship between the measured samples in the 700m2 sample plots. The existence of a weak spatial relationship between samples for basal area/ha in 300m2 sample plots indicate a weak spatial structure in these plots.