关键词: 种植制度, NDVI, 二次差分法, 傅里叶变换法

Abstract: 【Objective】 The objective of this paper was to map the cropping systems of China using ten day NOAA- AVHRR- NDVI data of 1986 and 1996 years, and SPOT-VEGETATION-NDVI data of 2000. In order to extract the cropping systems information, the accuracies of the twice difference algorithm and Fourier transform method were compared. The performed better method was applied to map the cropping systems distributions for each year. On the basis of the cropping systems distributions, the changes of cropping systems in recent years would be known. Some meteorological data were used to reveal the possible reason of the change of cropping systems. The results were expected to provide information about the crop yield changes and their reasons.【Method】First, the cropping area was subset from the global land cover map provided by the European Space Agency (ESA GlobCover). Second, the cropping system distribution of 1986 was mapped using twice difference algorithm and Fourier transform method by ENVI and ArcGIS software. The accuracies of the two methods were compared using the cropping system information in the “Vegetation Map of China” at the scale of 1:4 000 000 as a reference. The cropping systems distributions of 1996 and 2006 were mapped by the twice difference algorithm. The distributions were compared to extract the change information of cropping systems. At last, the accumulated temperature indicators of the zero-grade zone were used to analyze the possible impact of the climate change on the cropping systems change.【Result】The twice difference algorithm and the Fourier transform both provided reasonable accuracy for mapping cropping system of China. The area of one crop a year was mainly in the northeast area, the two crops a year was distributed in Henan, Hebei, Shandong, and Anhui provinces, and the three crops a year was covered by the area south to the Hubei Province. The accuracy of twice difference algorithm (overall accuracy was 76.5% and Kappa coefficient was 0.64) was a little higher than the Fourier transform (overall accuracy was 69.4% and Kappa coefficient was 0.51) in this research. The northern limit of the two crops a year moved from Huludao, Liaoning Province in 1986 to Benxi and Shenyang, Liaoning Province in 2006. The western limit of it also moved from Pingliang, Gansu Province, Baoji, Shanxi Province in 1986 to Linxia, Ningxia Hui Autonomous Region and Haidong, Qinghai Province in 2006. The cross transformation was so complicated that the area of the one crop a year decreased by 330 thousand square kilometers, two crops a year increased by 360 thousand square kilometers, and three crops a year increased by 40 thousand square kilometers. The contour line of the accumulated temperature of the 4 200ºC and 6 100 ºC also moved to north and west in the studied years. The results may imply that the temperature is a main impact factor of changes of cropping systems.【Conclusion】Remote sensing data based on time sequence could be used to monitor real change of cropping system in China. The twice difference algorithm is a valid method to extract the crop systems information in China. The limit of the cropping system, especially for the two crops a year moved northwards and westwards. This movements resulted in the large scale decrease of one crop a year and significant increase of two crop a year, while the three crops a year did not change much. The global warming had an effect on the cropping system changes in China. Analyzing the distribution and change of cropping system in China by remote sensing data can not only help us understand and evaluate the crop production and production potential timely and accurately, but also benefit the agriculture-related departments for drawing up effective agricultural policies.

Key words: cropping system, NDVI, twice difference algorithm, Fourier transform