Abstract: Relying on forest resources planning and design for Yunnan Yunjing Forestry Development Co., LTD, using spatial stratified sampling technique, based on the small class vector data from the fourth investigation of forest resources planning and design and "one map" of forest resources management in 2018, with land types, dominant tree species, and age group as the stratification variables, the 62 total volume control plots in the fourth forest resources planning and design survey were stratified, and the number of sample units required under the same reliability(95%) and different design precision(95%, 90%, 85%) were calculated, and compared with the system sampling. The calculation results show that when the reliability is 95%, the proportion of sample units required by the system sampling with 95%, 90% and 85% design accuracy is 3.8:1.0:0.4. Stratified sampling requires fewer sample units than systematic sampling. The number of sample units required by stratified sampling calculated according to the stratified variables of land class, dominant tree species and age group was 8.79%, 12.38% and 18.35% less than that of systematic sampling when the design accuracy was 95%. When the design accuracy is 90%, it is 12.50%, 15.83% and 21.67% less than the system sampling. When the design accuracy is 85%, it is 8.82%, 11.76% and 17.65% less than the system sampling. Under the same sampling design accuracy, the number of sampling units showed the general rule of land species > dominant tree species > age group. The results of 247 total volume control plots in the forest resource planning and design survey of Yunnan Yunjing Forestry Development Co., LTD were stratified by using spatial stratification technique, and the actual sampling accuracy was calculated, so as to carry out the comparative study of pre-stratified sampling, post-stratified sampling and systematic sampling.From the actual sampling accuracy calculation results, the stratified sampling accuracy is generally higher than the system sampling accuracy. In stratified sampling, the accuracy of post-stratified sampling is 1-2 percentage points higher than that of pre-stratified sampling. In prior stratified sampling, with the same stratified variables, the sampling accuracy of stratified sampling based on the small-class vector data of "one Map" of forest resource management in 2018 is slightly higher than that based on the small-class vector data of the fourth forest resource planning and design survey. When the reliability and sampling design accuracy are the same, the actual sampling accuracy is always the highest when the dominant tree species are used as stratified variables, indicating that the stratified sampling effect is the best when the dominant tree species are used.