外来植物入侵遥感监测预警研究进展

文章紧密围绕近年来国内外关于外来植物入侵遥感监测、预警新技术、新方法,总结了当前监测预警技术方法的问题和不足,阐明了未来针对外来入侵植物的遥感监测、预警技术发展趋势和方向。目前在监测方法上,主要有图像识别法、高光谱监测法、雷达数据辅助识别法和中低分辨率时序序列数据分析法等。其中,图像识别法和高光谱监测法应用较多,但其数据来源获取困难,后期数据处理工作量大,缺少普适性强的自动解译算法;雷达数据辅助识别法由于雷达数据自身空间分辨率粗糙,在植被监测中往往配合其他数据源同时使用,因此应用上有一定的局限性;中低分辨率时序序列数据分析法适用于较大空间尺度上的植被外来物种入侵监测,监测结果空间定位能力差,很难在小尺度区域上进行应用。在预警方法上,主要有生态机理模型预测和数学模型机理预测两种方法,基于GIS和遥感技术的生态学模型预测机理性强、易推广,预警精度较好,具备一定的空间定位能力,适宜于在宏观区域尺度上应用,但存在部分关键植被生理遥感参数获取困难等问题。数学统计模型预测以理论统计为主,只是对外来植物入侵发生的概率进行了估测,对物种或区域环境依赖程度高,普适性较差,遥感参数较少参与模拟过程,空间定位能力不足。总体上,利用遥感技术快速、准确地进行外来入侵植物定位监测和预警已成为防控外来植物入侵,维护区域生态安全的必然趋势。在数据源集成上,传统可见光遥感和新型激光雷达遥感、高光谱遥感、多角度遥感等多源数据集成、融合应用更为多见,数据获取范围进一步拓宽。特别是诸多新型国产高分卫星发射升空后,使得监测时效性和成本进一步改善,实用化、规模化、业务化动态监测成为可能。在技术方法上,受信息技术、自动化技术和传感器制造技术的影响,数据获取范围不断拓宽,空间定位能力不断增强,监测、预警也将不再局限于单一技术和模型的应用,而是多模型、多平台(星-机-地)综合监测、预警模式成为趋势,外来植物入侵的遥感监测技术正逐步发展成为植被生态遥感学科领域的重要分支学科。

PROGRESS ON MONITORING AND EARLY WARNING OF INVASIVE ALIEN PLANTS BY REMOTE SENSING

This study focuses on the new methods and technologies of monitoring and early warning based on remote sensing of invasive alien plants, summarizes the problems, and illuminates the future technology development trends and directions.At present, there are four mainly monitoring methods: image classification, high spectral data analysis, radar data aided identification method, and middle-low resolution sequence data analysis. Among them, image classification and high spectral data analysis are more applied, but the datum is difficult to obtain, the late data processing is very big, and universal automatic interpretation algorithm is lack.The radar data aided identification method also has some limitations such as rough spatial resolution radar data, and other types remote sensing datum. The middle-low resolution temporal sequence data analysis method is usually suitable to large spatial scale of invasive species of vegetation monitoring, but the monitoring results of spatial orientation ability are poor. On the early warning methods, there are two main methods: ecological mechanism model and mathematical mechanism model. Ecological mechanism model is based on geographical information system and remote sensing technology and has higher prediction accuracy.It is usually applied to large spatial scale, but some of model parameters is difficult to obtain. Mathematical statistics model prediction is given priority to with theory of statistics.It estimates the probability of alien plant invasion and has high degree of dependence on species or regional environment, but lack of fixes positions ability.In summary, monitoring and early warning of invasive alien plants by remote sensing quickly and accurately has become an inexorable trend of controlling of exotic plant invasion and maintenance of regional ecological security. Especially new remote sensing data sources are spring up in the world, spatial analysis methods sustaining innovations and rapid improvement of spectral resolution, all of these provide a new opportunity to invasive species of vegetation remote sensing monitoring and early warning technology. In data sources integration, multi-sources data integration and fusion of traditional optical remote sensing,new type of radar remote sensing, hyper-spectral remote sensing, multi angular remote sensing are applied more frequently,the data acquisition range is further widened. HJ, ZY-02C, ZY-3,GF-1/2 and other new domestic satellite remote sensing data are already used, and the remote sensing monitoring cost is further decreased. So the aquatic invasive plants can be regularly monitored in large scale with low cost. With the development of information technologies automatics, transducer making techniques, data access scope has been widened and its capability of spatial orientation is constantly enhanced. The monitoring and early warning of invasive alien plants methods are no longer limited to single technology and model but multi-models and multi-platforms (Aircraft-Satellite-Surface system). At the same time, the monitoring and early warning of Invasive alien plants by remote sensing has been gradually developed as an important branch of ecological remote sensing.