作物类型遥感识别研究进展

A Review of Remote Sensing Application in Crop Type Discrimination

及时获取作物种植面积是研究粮食区域平衡,预测农业综合生产力和人口承载力的基础。遥感技术已经成为提取作物种植面积的重要手段,而前提是识别作物。为了理清当前该领域的国内外研究现状,以遥感在作物类型识别中的应用为主线,归纳了国内外作物类型识别研究中常用的各类遥感数据,如资源遥感影像、气象遥感影像、高分辨率影像、高光谱影像和微波影像等,分析其优缺点和适用性;同时总结了利用遥感进行作物类型识别的3类研究方法,包括基于光谱的识别方法、基于物候差异的识别方法以及光谱与物候相结合的方法,分析了各种方法的特点;指出目前作物类型遥感识别中存在的主要问题,如影像空间精度与价格的平衡问题,多分辨率遥感数据的综合应用问题,物候差异对作物识别的影响问题等;认为不同分辨率遥感数据的结合可以弥补各自不足,遥感影像的时相选择是提高精度的关键,另外需要应用除光谱和物候以外的更多解译标志;建议进一步深入研究作物识别机理和多尺度数据融合方法。以期为遥感技术在作物类型识别中的深入研究提供参考和借鉴。

Timely crop planting acreage and yield information are important basis for researching regionalbalance of food supplies, predicting overall agricultural productivity and population carrying capacity. Remotesensing technology has become an important means to extract crop acreage, while the prerequisite is crop typediscrimination. In order to illuminate the current status of research in this field at home and abroad, based onthe remote sensing application in crop type discrimination, this paper firstly introduced the remote sensing datacommonly used in crop type discrimination researches, including Landsat satellite imagery, meteorologicalsatellite imagery, high resolution imagery, hyperspectral remote sensing imagery, and microwave remotesensing imagery, then further analyzed the advantages, disadvantages and practicability of each type of remotesensing data. Moreover, this paper summarized three types of crop type discrimination methods based onremote sensing, namely, the methods based on spectral characteristic, those based on phenology differencesand the combination of both methods, and analyzed the characteristics of various methods. Finally this paperinstance, the balance of image spatial resolution and price, comprehensive application of multi- resolutionremote sensing data, the impact of phenological differences on crop type discrimination, etc. The authorconsidered that the combination of different resolution remote sensing data could compensate for their lack;temporal selection of remote sensing images was the key to improve the discrimination accuracy; in addition,more interpretation signs need to be applied besides spectrum and phenology. Therefore, the authorrecommended further development crop discrimination mechanism and multi-scale image data fusion method,making research results more accurate and practical. The study could provide the references for the remotesensing crop type discrimination in-depth research.