基于多时相NDVI数据的复种指数提取方法研究

A Methodology for Multiple Cropping Index Extraction Based on NDVI Time-Series

多熟种植是中国重要的种植制度,在农业生产中具有重要的地位。复种指数反映了耕地实际利用强度,及时获取其空间分布信息是国家农业决策的基础。论文在重新理解和定义复种指数的基础上,提出了一种基于NDVI曲线迭代修正的复种指数遥感提取方法,并以SPOT/VGT多时相NDVI数据为基础,提取了1999-2004年中国北方17省市区农用地的复种指数。结果表明：提取的复种指数空间分布符合我国耕作制度区划,基于目视解译的样点检验总体精度为95.24%,Kappa系数为0.905 7,与根据统计数据得出的复种指数结果的线性回归斜率为0.928 8（R2=0.915 9,P〈0.001）,表明该方法能够准确快捷地提取复种指数,及时地为农业和土地管理部门提供耕作制度的空间信息。

Multiple cropping system characterized by cropping index is crucial to Chinese food security.Multiple cropping index（MCI）refers to the times of sequential crop planting in the same arable land in one year,reflecting utility degree of arable land to be used at a certain period.It is desired to extract the MCI and its spatial distribution information by remotely sensed data for sustainable agricultural development.Traditionally,the MCI is calculated by statistical data at local administration unit,which is time-lagged,labor-consuming and poor in creditability as well as lack of spatial distribution.Meanwhile,remote sensing technology has been widely applied to agriculture and crop monitoring.The advancement in remote sensing technology provides potenial to obtain the actual MCI information efficiently and reliably.Many researches indicate that crop growth dynamic can be monitored by the time series of Normal Difference Vegetation Index（NDVI） data.It is found that the peak of the time series of the NDVI curve indicates that the ground biomass of crops reaches the maximum,and fluctuates with the crops growing processes such as sowing,seeding,heading,ripeness,and harvesting within one year.Thus,the MCI is defined as the number of peaks of the time series of the NDVI curve.However,since NDVI data are affected by cloud and poor atmospheric conditions,the curve of time series of NDVI turns out to be noisy with a lot of small peaks and valleys in one cycle,which makes it more difficult to extract the actual MCI.This study develops a new method for extracting cropping index based on NDVI time-series by which cloud and other contaminations can be corrected effectively.The details about the method include：（1） smoothing the NDVI time-series by Savitzky-Golay filter to get the long-term growing trend;（2） correcting the growing trend curve by iteration process to get rid of the false peaks;and（3） obtaining the MCI by counting the number of true peaks.Using 1km 10-day Maximum Value Composite SPOT/VGT NDVI time-series,the MCI of 17 provinces of northern China from 1999 to 2004 was extracted by this new method.The results revealed that： the cropping index of Huang-Huai-Hai region was higher than that of the other regions in northern China.The spatial distribution of the extracted MCI was consistent with the actual Chinese cropping system.The total precision of sample validation based on visual identification was 95.24%,the coefficient of Kappa 0.9057,and the slope of linear regression of the MCI between remotely sensed data and statistical data was 0.9288（R2=0.9159,P〈0.001）,suggesting that this method could provide an effective way to extracting spatial information of the MCI for agriculture and land management.