首页 >  2015, Vol. 19, Issue (4) : 568-577

摘要

全文摘要次数: 5583 全文下载次数: 190
引用本文:

DOI:

10.11834/jrs.20154146

收稿日期:

2014-05-29

修改日期:

2014-09-11

PDF Free   HTML   EndNote   BibTeX
MODIS植被指数的美国玉米单产遥感估测
1.中国农业科学院农业资源与农业区划研究所, 北京 100081;2.农业部农业信息技术重点实验室, 北京 100081
摘要:

针对中国开展的国外农作物产量遥感估测大多依靠中低分辨率耕地信息、省级(州级)或国家级作物产量统计数据的现状,本文以美国玉米为例,探讨利用多年中高分辨率作物分布信息、时序遥感植被指数和县级作物产量统计数据开展国外重点地区作物单产遥感估测技术研究,以期进一步提高中国对国外农作物产量监测精度和精细化水平.首先,利用美国农业部国家农业统计局(NASS/USDA)生产的作物分布数据(CDL)获得多个年份玉米空间分布图,并对相应年份250 m分辨率16天合成的MODIS-NDVI时序数据进行掩膜处理,统计获得每年各县域内玉米主要生育期NDVI均值;其次,以各州为估产区,以多年县级玉米统计单产和县域内玉米主要生育期NDVI均值为基础,建立各州玉米主要生育期NDVI与玉米单产间关系模型;然后,通过主要生育期玉米单产和玉米植被指数间拟合程度,筛选确定各州玉米最佳估产期和最佳估产模型.最终,利用最佳估产模型实现美国各州玉米单产估测和全国玉米单产推算.其中,建模数据覆盖时间为2007年—2010年,验证数据为2011年.结果表明,应用最佳估产模型的2011年美国各州玉米单产估测相对误差在-4.16%—4.92%,均方根误差在148.75—820.93 kg/ha,各州估测结果计算获得全国玉米单产的相对误差仅为2.12%,均方根误差为285.57 kg/ha.可见,本研究的作物单产遥感估测技术方法具有一定可行性,可准确估测全球重点地区作物单产信息.

MODIS vegetation index data used for estimating corn yield in USA
Abstract:

The coarse resolution arable land and crop yield statistics at state level or country level were often used in most of former research of crop yield estimation for other overseas countries and regions with remote sensing in China. In view of this above research status, taken corn yield estimation with remote sensing in USA as an example, the method of crop yield estimation based on higher resolution crop-specific maps, county-level statistic data and MODIS NDVI time series data was explored in this paper in order to further improve the accuracy and refinement level of crop yield estimation research results. Firstly, corn maps with higher spatial resolution of multiple years were acquired from Cropland Data Layer (CDL) produced annually by the National Agricultural Statistics Service (NASS) of United States Department of Agriculture (USDA). Taken the above crop-specific maps as mask image respectively, mean NDVI value of corn in each main growth stage of every year was calculated in each county. Then, taken each state as a corn yield estimation region respectively, relationship between mean NDVI of corn in each critical growth stage and statistic county-level crop yield data was built up in each state. Thirdly, according to the fitting degree between NDVI and crop yield in each growth stage of corn, the best time and best model of corn yield estimation in each state was selected out. Finally, corn yield of each state was calculated depending on the best crop yield estimation model and the mean NDVI of corn in each county and corn yield at national level was gotten through aggregating the corn yield of each state. Accuracy validation of each state and whole country were carried out. Among them, data in the year from 2007 to 2010 were used to build the models and data of the year of 2011 was used as validation data. It was shown that Relative Error (RE) of corn yield estimation in each state in the year of 2011 was between -4.16% and 4.92% and that Root Mean Square Error (RMSE) was between 148.75 kg/ha and 820.93 kg/ha and that the RE of the corn yield estimation at national level was only 2.12% and RMSE was only 285.57 kg/ha. We could draw the conclusion that method of crop yield estimation based on crop-specific maps, county-level statistic data and MODIS NDVI time series data was reasonable and feasible and could estimate crop yield more accurately in larger overseas region.

本文暂时没有被引用!

欢迎关注学报微信

遥感学报交流群