Effect of pre-stratification on increasing the precision of cluster sampling method to estimate forest attributes (Case study: Bayangan, Kermanshah)

In this study a pre-stratification was applied to estimate quantitative attributes of oak forests by cluster sampling method in Bayangan district, Kermanshah province. Firstly stratification was conducted according to NDVI on landsat-8 images and four strata were determined. Sample size was computed according to allowable error and samples were allocated to three strata with regard to their heterogeneity. Totally 34 five-plot clusters include 170 samples with 90×90 meter dimensions were measured in 30000 ha area of study area. Crown diameter and diameter at breast height (DBH) were measured and finally density, crown cover and basal area per unit area were computed. Results of stratified sampling showed 149.5 tree, 1367.6 m2.ha-1 (13.67 percent) and 3.21 m2.ha-1 for density, crown cover and basal area respectively. While this measures for cluster sampling without stratification were 221.8, 2013.5 and 4.77 respectively. This was due to high acreage (weight) but low density and crown cover in the first strata with huge effect on forest mean. Error percent was 1.74, 2.65 and 3.14 for pre-stratification status and 9.11, 9.14 and 9.33 for cluster sampling without stratification. It can be said that pre-stratification can decrease sample size from 66 to 80 percent. We conclude that with regards to heterogeneity of Zagros forests in type, density and crown cover, pre-stratification can improve precision of cluster sampling and is recommended for general inventories.