›› 2010, Vol. 18 ›› Issue (3): 327-332.DOI: 10.11733/j.issn.1007-0435.2010.03.004

Previous Articles     Next Articles

Accuracy Improvement in Yield Estimation of Large-scale Grassland by Stratified Sampling—-A Case Study of Lowland Meadow

BAO Hai-ming1, YUN Xu-jiang1,2, ZHANG De-gang1, LI Xin-yi2, LIU Xiao-dong 1   

  1. 1. Pratacultural College, Gansu Agricultural University, Lanzhou, Gansu Province, 730070, China;
    2. National Animal Husbandry Service, Ministry of Agriculture, Beijing, 100025, China
  • Received:2009-11-20 Revised:2010-04-28 Online:2010-06-15 Published:2010-06-15

地理空间分层抽样对大尺度草地估产精度的改善—-以低地草甸为例

包海明1, 贠旭疆1,2, 张德罡1, 李新一2, 刘晓东1   

  1. 1. 甘肃农业大学草业学院, 甘肃, 兰州, 730070;
    2. 农业部全国畜牧总站, 北京, 100025
  • 通讯作者: 贠旭疆,E-mail:yunxj@cav.net.cn
  • 作者简介:包海明(1982- ),男,内蒙古呼伦贝尔人,硕士研究生,研究方向为草地生态、管理与遥感监测,E-mail:bhm999@126.com
  • 基金资助:
    内蒙古自治区自然科学资金项目(2009MS0414)资助

Abstract: There are various types of grassland in China and the geographical spatial span of a given grassland type is very large.Due to different climate and soil condition in different areas,the situation of vegetative development is different from each other,thus these elements affect the precision of yield estimation by conventional methods.In order to optimize the monitoring strategy of natural grasslands,the aboveground biomass yield of lowland meadow was evaluated according to the principles of stratified sampling method and grassland zoning in China using stratification efficiency test,overall average error comparing analysis,and model analysis based on the ground measured data and MODIS-NDVI data.The results show that: the stratification efficiency of each strata area was more than 1;in comparison with random sampling method,the stratified sampling method improved the overall average accuracy by 9.672% and the regression mode accuracy by 13.827%.These results verify that stratified sampling method could be more recommendable than random sampling method for the biomass yield estimation of a given grassland type.

Key words: Large-scale, Stratified sampling, Accuracy, Stratification efficiency, Optimization strategy

摘要: 为优化天然草地监测策略,提高样本点的地域空间代表性和分层估产模型的稳定性,改善全国天然草地地上生物量的估测精度,以低地草甸草原为研究对象,结合地面采样数据和MODIS-NDVI数据,根据统计学的分层抽样方法和全国草地资源区划原理为理论依据,通过分层效率检验、总体平均值误差对比验证和模型分析等方法,以低地草甸为例进行分析。结果表明:各分层区域的分层效率都大于1;采用分层抽样方法相对于简单随机抽样方法地面采样样本的总体平均值精度提高9.672%,模型的估产精度提高13.827%。从而,验证了某一类型草地的区域分层抽样方法比简单随机抽样方法更可取。

关键词: 大尺度, 分层抽样, 精度, 分层效率, 优化决策

CLC Number: