遥感技术因其高时效、宽范围和低成本等优点正被广泛应用于对地观测活动中,为大区域尺度掌握农作物空间格局提供了新的科学技术手段。本文系统总结了近10年来国内外农作物空间格局遥感监测在理论、方法、实践应用等方面取得的新进展,指出了亟待解决的问题,并对今后的发展方向进行了展望。研究认为,农作物种植面积遥感监测主要根据遥感传感器记录的不同农作物光谱特征的差异,进行不同农作物种植面积的识别,方法主要包括:基于光谱特征、基于作物物候特征和基于多源数据的农作物遥感识别方法。遥感技术应用于农作物复种模式监测主要根据时间序列植被指数描述的作物季节活动过程,利用不同的拟合方法得到作物生长曲线,实现作物复种模式有效监测。农作物种植方式遥感监测是更高层次的遥感应用,主要利用时间序列遥感数据,根据作物植被指数的变化规律区分不同作物生育周期,判断不同复种模式下作物的种植顺序和方式。在未来相当长的一段时间内,建立农作物空间格局遥感监测的理论和技术体系、发展和改进遥感影像分类方法、优化时间序列遥感数据平滑技术和提高信息提取的自动化与流程化将是农作物空间格局遥感监测需要重点解决的几个关键问题。
As a new high-technology with an advantage of high temporal resolution,wide coverage and low cost,remote sensing is currently used in a wide arrange of earth observation activities and thus provides a useful tool to detect and monitor the spatial patterns of crop cultivation.Based on the systematic summary of the progress of studies in remote-sensing-based monitoring of spatial patterns of agricultural crops in the latest decade,including its theories,methods and applications,a series of problems that should be urgently resolved in the study are put forward,and some important study directions and priorities for future are viewed.Studies show that crop acreage can be monitored according to the differences in spectral characteristics of different crops,which are normally recorded by the satellite sensors.There are three major approaches used for crop acreage monitoring:spectral-based identification,phenology-based identification and multiple data-fusion-based identification methods.Mapping multiple cropping systems using remote sensing is mainly based on the crop growth curves,which can be derived from the smoothed time-series vegetation index(VI) data.Furthermore,cropping patterns can be also examined through discriminating the crop growth period from variations in time-series VI data and characteristics of different cropping patterns.How to construct the theoretical and technological systems,to develop and verify the image classification methods,to optimize the smoothing methods for time-series data and to improve the capability of automatic extraction of information could be the major development trends of this field in the future.