农业工程学报Issue(2)：146-152,7.DOI:10.3969/j.issn.1002-6819.2014.02.019
基于遥感与多变量概率抽样调查的作物种植面积测量
Mapping crops acreages based on remote sensing and sampling investigation by multivariate probability proportional to size
摘要
Abstract
MPPS is a method widely used in crop area statistics in the Chinese crop area statistical investigation business. However, this method has two drawbacks. One is the outdated basic data. The other is the large workload of a field survey. The second land use survey data used as the basic data in the Chinese crop area statistical investigation is only updated every 10 years. The longer update cycle makes it difficult to react to the inter-annual change of crop areas. The artificial field survey is used in the Chinese crop area statistical investigations to survey the area of crops of every sampling village. Because of the large number of sample villages, the workload of field investigation is huge, and time-consuming and laborious. In order to solve those problems in a conditional sampling survey, a novel crop area extraction method was proposed in this paper using remote sensing and MPPS sampling technology. The sampling frame was prepared using the village-level administrative unit data of the second land use survey data. Crops were extracted using multi-temporal HJ-1 satellite data with a Spectral Angle Mapper method. Three HJ-1 satellite data sets acquired in April, May, and August were selected according to the Phonological data. In April, rice and winter wheat were in the seedling stage, and corn was not planted. In May, rice was in irrigation period. In August, winter wheat had been harvested, while rice and corn were in their maturity periods. So using images in those months, it was easy to differentiate rice from winter wheat and corn since the paddy land contains water, while the wheat and corn land were dry in May. It was also easy to differentiate winter wheat from corn because the growing period of winter wheat was 20 days earlier than the growing period of corn. Then the crop areas of each village were updated by the moderate resolution crop classification map. Combining the updated sampling frame data and MPPS sampling method, sampling villages were selected. Crops in the sampling villages were mapped using ZY-1 02c satellite data with an object-oriented classification method. The ZY-1 02C satellite is a new Chinese civil remote sensing satellite launched on December 22, 2011. It was the highest resolution civil remote sensing satellite in China which carried a panchromatic/multispectral sensor and a high-resolution sensor. The spatial resolution of the HR sensor was 2.36 m, and the spatial resolution of the panchromatic/multispectral sensor was 5m in a panchromatic band and 10m in three multispectral bands. Finally, according to the MPPS method, the total area of each crop in the study area and CV were calculated. The algorithm had been tested over a study area in Beizhen Country, Liaoning Province, China. The results showed that this method could effectively determine the rice and corn areas. A high mapping precision of 92%was obtained.关键词
遥感/农作物/测量/MPPS抽样/面向对象分类/种植面积Key words
remote sensing/crops/measurements/MPPS/object-oriented classification method/crop acreage分类
农业科技引用本文复制引用
邬明权,杨良闯,于博,王玉,赵昕,牛铮,王长耀..基于遥感与多变量概率抽样调查的作物种植面积测量[J].农业工程学报,2014,(2):146-152,7.基金项目
国家自然科学基金（41301390）；国家科技重大专项；国家重点基础研究发展计划项目（2013CB733405，2010CB950603）。 （）
