基于主被动遥感的沼泽植被群丛时空分布与水文情势耦合研究

Freshwater marsh plant community spatially present annular or zonal distribution with the change of hydrological situation, called micro-zones differentiation of vegetation association, of which formation causes and mechanics has not yet been conducted a systematic study. The study on coupling relationship between temporal and spatial distribution of vegetation association and water level, flooding frequency and duration is able to provide a breakthrough to explain the micro-zones differentiation. This project utilized high spatial resolution UAV (unmanned aerial vehicle) and space-borne multispectral images and polarimetric SAR data to build identification and classification models for mapping and analyzing temporal-spatial evolution of vegetation association in Honghe National Nature Reserve. The UAV-based SFM and CR/PSInSAR technique both are used to time-series monitor dynamic change of hydrological situation, and to explore the change law of water level, flooding frequency and duration. Finally, the essential hydrology conditions of formation and stable distribution of vegetation association is quantified. The driving mechanism between water level, flooding frequency, flooding duration and temporal-spatial evolution of vegetation association is discussed to reveal formation causes of micro-zones differentiation of vegetation association. This project will provide scientific basis for developing and improving the theory of wetland boundary definition, wetland protection and restoration.

沼泽植被群丛响应水文情势的变化，在空间上呈现有规律的环状或带状分布现象，称为湿地群丛结构微域分异，但其成因及机理到目前仍未得到系统研究。从水文情势上研究沼泽植被群丛时空分布与湿地水位、淹水频率和淹水时长的耦合机制和定量关系为科学解释湿地群丛微域分异提供了突破口。本项目以三江平原腹地的洪河国家级自然保护区为研究区，在高分辨率无人机、星载多光谱影像和极化SAR影像的支持下，构建沼泽植被群丛遥感识别模型，解析植被群丛时空分布及其演变特征；利用无人机SFM技术和CR/PSInSAR技术时序监测湿地水文情势，探究沼泽湿地水位、淹水频率和淹水时长的动态变化规律；最终，确定和量化沼泽植被群丛形成及其稳定分布所需的基本水文情势条件，探明湿地水位、淹水频率和淹水时长对植被群丛时空演替的驱动机制，揭示湿地群丛结构微域分异的成因。本项目将为我国湿地边界界定理论的发展和完善，湿地恢复与保护等提供重要科学依据。

植被-水相互作用机制一直是湿地生态系统研究的热点和难点之一，由于目前缺乏定量地描述湿地植被-水时空演变特征及其耦合关系的方法，导致湿地植被群落微域分异成因及机理仍未得到系统研究和科学解释。在本基金项目的资助下，团队成员以沼泽植被-水的相互关系为研究核心，在沼泽植被迁移学习分类、湿地水位变化遥感监测、湿地植被和水耦合关系、湿地水文边界界定等方面取得了突破性进展，主要内容包括：（1）系统开展探究了多源遥感影像、浅层机器学习和深度学习算在沼泽植被多尺度分类中的精度差异和适用性；（2）在空间分辨率梯度和光谱维度上首次探究了深度学习算法对空间分布复杂的自然沼泽植被迁移学习分类能力；（3）创新性提出了基于无人机辅助采样的沼泽植被冠层叶绿素a浓度反演方法；（4）评估了卫星测高和DInSAR技术监测湿地水位变化的精度和适用性；（5）提出并完善了沼泽湿地的“淹埋深－历时－频率”阈值研究的理论和方法，创新性构建湿地水文边界的界定模型，并界定了沼泽湿地水文边界。（6）首次构建高精度的沼泽植被和水体遥感时空监测模型，创新地构建沼泽植被-水体耦合协调度模型和传递熵，系统揭示了沼泽植物群落空间分异特征与水文系统之间的响应关系。