| 摘要: |
| 在广东省 2012 年布设的 459 个 2 km×2 km 大样地基础上，以大样地为一个群，每个群等间距布
设 25 个 25.82 m×25.82 m（1/15 hm2
）方形样地组成群团样地，对每个方形小样地的地类进行遥感判读。采
用群团抽样方法分别产出 2012 年和 2016 年全省森林资源面积数据，对两期森林资源面积做动态分析，同
时与 2016 年连清固定样地地类复查结果进行对比分析。结果表明：全省 2016 年森林覆盖率为 53.18%，森
林面积为 940.06 万 hm2
，较 2012 年分别增长 1.58 个百分点和 27.93 万 hm2
，两期森林覆盖率抽样精度分别
为 94.43% 和 94.53%；对比连清体系，两期群团抽样森林覆盖率较 51.26% 和 53.00% 相差 0.34 和 0.18 个百分
点，两者较接近，而森林覆盖率抽样精度较连清体系低 2.43 个百分点；经实地验证纠正后的群团抽样结果
与连清体系总体差异增大，除林地和乔木林面积成数差异分别缩小 0.58% 和 1.17%，森林、竹林、特殊灌木
林地和其他林地面积成数差异分别扩大 2.46%、3.3%、2.3% 和 1.89%。结论显示，基于遥感判读的大样地群
团抽样能够快速产出森林覆盖率，监测时效性和成本均优于连清体系，但对森林覆盖率变化量监测结果波动
大、无精度保证。 |
| 关键词: 大样地 群团抽样 动态监测 森林面积 |
| DOI： |
| 分类号: |
| 基金项目:其它 |
|
| Dynamic Monitoring of Forest Area in Guangdong Province Based on Cluster Sampling and Large Plot |
|
Wang Qiulai1, Xue Chunquan2, Chen Chuanguo2, Yang Zhigang2, Yu Songbai2, She Guanghui1
|
|
1.Nanjing Forestry University,Nanjing;2.Forestry Surveying and Designing Institute of Guangdong Province,Guangzhou
|
| Abstract: |
| A cluster sampling was designed based on 459 large plots of 2 km by 2 km built in Guangdong
Province in 2012.Each large plot as a cluster was arranged 25 square plots of 25.82 m by 25.82 m (1/15 hm2
) of
equal distance, and the land type of each small square plot was identified by remote sensing image. Forest resource
area of the province in 2012 and 2016 were output, while analyses were conducted on the dynamic changes of
forest resource and contrasts with the National Forest Inventory (NFI) scheme in 2016. The results showed that the
province forest cover rate was 53.18% and forest area was 9 400.6 thousand hectares in 2016, had increased 1.58
percentage points and 279.3 thousand hectares than 2012, while the sampling accuracy of cluster sampling in 2012
and 2016 were 94.43% and 94.53%. Compared with NFI scheme, the two installment forest cover rates output
by cluster sampling scheme differed by 0.34 and 0.18 percentage points contrasted to 51.26% and 53.00%, close
to each other, while the sampling accuracy of it was lower 2.43% than NFI. Compare with the results of clustersampling by remote sensing identification which were revised by site invention and NFI scheme, the differences
were overall increased, area rates of forest, bamboo forest, special shrub forest and other forest land had increased
2.46%,3.3%,2.3% and 1.89% ,while area rates of forest land and arbor forest had decreased 0.58% and 1.17%. In
general, the forest cover rate could by output fast by the cluster sampling scheme of remote sensing identification and
large plot, and the timeless, cost were superior to NFI scheme, but the change in forest cover rate had high fluctuation
and low sampling accuracy. |
| Key words: large plot cluster sampling dynamic monitoring forest area |
