摘要
野外采样设计是数字土壤制图的首要步骤,基于不同设计样点推理的土壤图具有不同的精度。另一方面,采样方案应该充分整合研究区内已有样点资源以减少采样成本。而目前经常用到的三种样点设计方案:基于经典统计理论的空间采样方案,基于地统计的空间采样方案和目的性采样方案都很难整合已有样点资源。在推理不确定性图的指导下,利用模拟退火的方法提出了一套既能充分整合已有样点资源又能提高采样效率的补充样点设计方案。在黑龙江省嫩江县鹤山农场老莱河左岸,利用模拟退火算法和规则采样方法分别设计了1个、4个、5个、10个、15个、23个样点,结果证明在相同样点数量情况下(尤其是样点数量较少时),模拟退火算法设计的样点具有更好的代表性,基于这些样点推理的土壤属性图具有更高的可靠性。
Soil field samples are mainly data source for predicting spatial variation distribution of soil properties, and sampling designing is the first step for digital soil mapping. Soil maps produced from samples designed by different schemes have different accuracy. On the other hand, there exist some soil field samples which were accumulated through several historical soil surveys and/or specific field studies. Therefore, the follow-up sampling scheme should integrate the existing samples for reducing sampling cost. However, integrating existing soil field samples is very difficult for the main three kinds of traditional sampling schemes including sampling scheme based on classic statistic theory, spatial sampling scheme based on geo-statistic theory and purposive sampling scheme. Under the predicting uncertainty distribution map, this paper presented a new method which not only can integrate the existing samples but also can improve sampling efficiency by simulated annealing method. In the Laolaihe watershed, Neijiang county, Heilongjiang Province, China, we designed 1, 4, 5, 10, 15, 23 samples by simulated annealing method and regular method. The predicting uncertainty showed that under the same number of soil samples condition, especially the number is small, the soil map produced by samples designed by simulated annealing method was more reliable than the ones designed by regular method.
出处
《土壤通报》
CAS
CSCD
北大核心
2013年第4期820-825,共6页
Chinese Journal of Soil Science
基金
国家自然科学基金项目(40971236
41023010
41001298)
科技部国际科技合作项目(2010DFB24140)资助
关键词
已有土壤样点
推理不确定性
模拟退火算法
Existing soil field samples
Predicting uncertainty
Simulated annealing method
作者简介
张淑杰（1983-），女，山东德州人，博士研究生，主要从事数字土壤制图及样点设计研究。E-mail：zhangsj@lreis．cn
