基于时间序列叶面积指数傅里叶变换的作物种植区域提取  

Extraction of planting areas of main crops based on Fourier transformed characteristics of time series leaf area index products

作　　者：王鹏新[1,2] 荀兰 李俐 解毅[1,2] 王蕾 

机构地区：[1]中国农业大学信息与电气工程学院,北京100083 [2]农业部农业灾害遥感重点实验室,北京100083

出　　处：《农业工程学报》2017年第21期207-215,共9页Transactions of the Chinese Society of Agricultural Engineering

基　　金：国家重点研发计划重点专项资助项目(2016YFD0300603-3)

摘　　要：为了获取不同农作物的空间分布信息,以华北平原黄河以北地区为研究区域,利用Savitzky-Golay滤波对2014—2016年的时间序列叶面积指数(leaf area index,LAI)进行重构,进而应用一阶差分法和重构LAI的傅里叶变换的谐波特征对研究区域主要农作物冬小麦、玉米和棉花种植区域进行识别和提取,并对不同作物的识别精度进行验证。结果表明,基于Savitzky-Golay滤波重构的LAI能够去除由云、大气等因素造成的LAI骤降的影响,重构LAI曲线平滑且符合作物的生长规律特征。研究区域2014—2016年作物识别的总体精度均大于80.00%,2015年达到87.08%,冬小麦-夏玉米、春玉米、棉花和单季夏玉米的识别精度分别为92.50%、80.00%、85.00%和82.50%,表明利用一阶差分法能够准确提取研究区域一年一季和一年两季作物种植区域。结合傅里叶变换方法和作物物候信息能够有效地识别不同作物的种植区域,进而获取研究区域主要农作物的分布信息。该研究可为研究区域主要作物的长势监测及产量估测预测提供参考。Crop identification is a basis of crop monitoring using remote sense data. Crop acreage is essential information for food security, land management and trade decisions. In this paper, the north of the Yellow River in North China Plain was chosen as the study area, where the main crops are winter wheat, maize, and cotton. The 4-day composite MODIS time series leaf area index （LAI） data with a spatial resolution of 1 km were collected yearly for the identification of different crops in the study area. In order to obtain the distribution of crops, we used 92 phases of MODIS LAI images yearly from 2014 to 2016. To avoid the edge effect of the time series LAI caused by data processing, including Savitzky-Golay filter and the second-order differential methods, the last 2 phases of LAI images in last year and the first 2 phases of LAI images in next year were added to build the time series LAI of a year. And time series LAI of each year was analyzed respectively. Then Savitzky-Golay filter was used to denoise and reconstruct time series LAI curves. The results showed that the Savitzky-Golay filter can remove the influence of factors such as cloud, atmosphere, and so on, and the reconstructed time series LAI curves were smooth and consistent with the crop growth and development characteristics. The Fourier transform provides a new representation of the time series images, which allows analysis of the crop phenology using the amplitudes and phases of the most important periodic components. In this study, the first-order differential method was employed to study the crop planting patterns in the study area by acquiring the number of peaks of the LAI curves, and the fields of one crop a year and two crops a year were extracted. The Fourier transform method based on Savitzky-Golay filtered LAI was further employed to extract the key parameters, such as the amplitudes and phases of the time series LAI. The 11 parameters, including the amplitudes of 0-5 terms and the phases of 1-5 order were used to build a multiband image

关 键 词：作物 遥感 提取 面积 叶面积指数 傅里叶变换 Savitzky-Golay滤波 华北平原 

分 类 号：S127[农业科学—农业基础科学]