基于多源影像时间序列的盐沼植被演替与扰动监测方法研究

Salt marsh is one of the most productive ecosystem and a sensitive area for global change, which becomes an important proxy for Future Earth Coasts (FE-Coasts) program. Recently, a significant number of salt marsh vegetation have experienced successions and disturbances due to the impacts of global climate change and human activities, which profoundly hampered stability of coastal ecosystems. Remote sensing has increasingly become an important way for coastal investigation, but remote-sensing-based salt marsh vegetation monitoring still confronts two key shortcomings: coordinating spatial scale with temporal scale and identifying subtle disturbance-response processes, as a result of the frequent cloudy weather in the coastal zone and complex community species and spectral similarity of salt marsh vegetation. Using Jiangsu middle coast as the study area, this project will construct time-series model at the pixel level to characterize salt marsh vegetation types, based on multi-source remote sensing images with harmonized spectrum. At long-term scale, the process of salt marsh vegetation succession will be revealed by alternations of the time-series model; at short-term scale, the process of salt marsh vegetation disturbance will be detected by fluctuations of the time-series model. Since the time-series models of each pixel run through the whole observation period, the proposed method can predict the distribution of salt marsh vegetation at any given time, thus achieves continuous change detection of salt marsh vegetation. The project is expected to provide the approach for salt marsh vegetation monitoring at a fine time scale, which can enrich remotely sensed technology and method framework in coastal zones. Meanwhile, all the findings are expected to support for coastal resource development and ecological conservation.

盐沼是最富生产力的生态系统，是全球变化的敏感区域，也是未来地球海岸带计划(FE-Coasts)的重要研究对象。气候变化和人类活动加剧了盐沼植被的演替和扰动，深刻影响着海岸带系统的稳定性。遥感日益成为海岸带监测的重要手段，但由于海岸带云雨天气频发且盐沼植被群落组成复杂、光谱相似度高，盐沼植被遥感监测尚存在时空观测尺度难以兼顾与扰动响应过程难以辨识等瓶颈。本项目拟以江苏中部沿海为研究区，协同多源遥感影像光谱差异，构建像素级影像时间序列模型表征盐沼植被类型；以长时间尺度下时间序列模型的更替表征盐沼植被物种演替，以短时间尺度下时间序列模型的波动表征盐沼植被扰动响应；各像元的时间序列模型贯穿观测时期的始终，可预测任意时刻的盐沼植被分布，进而监测盐沼植被的连续变化。本项目有望发展精细时间尺度下盐沼植被监测方法，扩充海岸带遥感观测理论与技术体系，同时为海岸带资源开发和环境保护提供基础数据支撑。

针对海岸带云雨天气频发且盐沼植被群落组成复杂、光谱相似度高等问题，项目研发了像元级时间序列模型，监测了盐沼植被的演替和扰动过程，预测了任意时刻的盐沼植被空间分布。具体内容包括：（1）开展了江苏中部沿海野外调研，采集海岸带湿地植被信息，作为验证滨海湿地演替监测结果。（2）收集、整理了多源光学遥感影像数据，依据“逐步协同”思路，依次协同两影像数据源，达到多源影像数据的统一。（3）构建了影像特征指数，逐像元建立时间序列模型，充分刻画滨海湿地植被演替特征。（4）提取了植被物候特征参数，建立时间序列模型与滨海植被类型的对应关系，实现滨海湿地植被精细识别。目前，项目发表SCI论文7篇，其中4篇论文发表在Remote Sensing of Environment、ISPRS Journal of Photogrammetry and Remote Sensing等遥感领域顶级期刊上。此外，项目申请发明专利3项，获批计算机软件著作权2项，参编学术专著1部，负责人孙超成功入选宁波市泛“3315”创新人才。项目研发的方法有望扩充海岸带遥感观测理论体系，相关结果可为江苏沿海资源开发利用和生态修复工作提供科学依据。