基于优化植被指数的草本湿地植被净初级生产力遥感估算

There are very important significance for understanding regional, and even global, carbon budget and influences from global change that investigating the method of estimating marsh vegetation NPP (Net Primary Productivity) and evaluating quantitatively the spatial pattern of NPP. Estimation models based on light use efficiency are commonly used method for estimating vegetation NPP. The researches in relation to the method of estimating marsh NPP and pattern of NPP are limited. This project will select typical marshes in Northeast China to study the relations between different marsh vegetation indices and canopy chlorophyll content, compare and analyze the correlations between vegetation indices and marsh vegetation LUE (Light Use Efficiency), further to accomplish selecting the best vegetation index for simulation marsh vegetation LUE. Finally, we will develop a model based on multi-spectral vegetation index for marsh NPP estimation. And also, present study intend to accurately map marshes based on high-spatial resolution remote sensing images and linear spectral mixture analysis model, estimate NPP of typical marshes in Northeast China using the developed model, and discuss the spatial pattern of LUE and NPP and effects from temperature, solar radiation, and terrain and so on, through the transect analysis of typical marshes at various natural and environmental backgrounds. Implementation of this study project will be of benefit to advance the level of quantitative remote sensing and can provide scientific basis for regional wetland carbon sequestration research and wetland's conservation and management.

深入探讨湿地植被净初级生产力 (NPP)估算方法，定量评估湿地植被NPP空间格局，对于理解区域乃至全球的碳收支具有重要意义。基于光能利用率的估算模型是目前植被NPP估算的主要方法，但是针对湿地植被的相关研究还较少。本项目选择中国东北地区典型草本湿地为研究对象，通过野外观测、理论分析和计算机模拟等方法，研究不同植被指数与冠层叶绿素含量间的关系，比较分析不同波段组合的植被指数与光能利用率间的相关程度；选择估算草本湿地植被光能利用率的最优植被指数，建立基于植被指数的湿地植被NPP估算模型；基于高空间分辨率遥感影像和线性混合像元分解模型，实现草本湿地的高精度提取；进而完成典型样区草本湿地植被NPP估算；以湿地样带分析为途径，研究NPP格局和各环境要素及植被类型对草本湿地植被光能利用率和NPP的影响。本项目的实施有助于提高湿地植被定量遥感研究水平，并将为区域湿地固碳研究及湿地保护与管理提供科学依据。

基于面向对象的湿地信息提取方法和决策树分类方法，对六个湿地样区进行遥感解译，结果显示：双台河口、查干湖、七星河的芦苇面积分别是440、75、和122km2，分别占保护区面积的34%、13%、61%。红星、南瓮河、洪河各有苔草湿地155、931和220km2，分别占保护区总面积的40%，13%，88%。应用线性混合像元分解计算的光谱植被指数较好地表征了湿地植被生长的空间异质性。. 基于实测数据和遥感数据分析发现：东北典型芦苇湿地叶片叶绿素含量与MSAVI、NDVI、RVI之间呈现显著相关（P＜0.05），相关性最大的是MSAVI。芦苇湿地植被NPP与MSAVI之间相关性最显著；冠层叶绿素含量与NDVI之间相关性最强，相关系数为0.79；LUE与冠层叶绿素含量显著相关（R2=0.64）；芦苇冠层APAR与LAI之间具有显著相关性（R2=0.62），APAR与MSAVI之间显著相关（R2=0.56）。基于LUE模型结构和MSAVI与NDVI光谱指数所构建的优化模型是芦苇植被NPP估算的最优模型，估算精度为72%。优化模型估算结果显示：双台河口、查干湖、七星河湿地样区芦苇年NPP值为3620、3050和3001gC•m2•yr-1，区域NPP受气候因子和人类活动干扰影响明显。CASA和BIO-BGC模型均低估了芦苇NPP。. 苔草湿地植被NPP与LAI极显著正相关（r=0.70），LAI能解释51%的变化量。NPP与各光谱指数间也存在极显著相关关系，其中NPP与NDVI之间相关性最大，相关系数为0.718，NDVI可以解释71%的变化量。基于实测数据分析发现NPP与LUE和APAR之间均存在极显著正相关关系；LUE与各个指数间均存在极显著相关关系（P<0.01）；其中，LUE与NDVI之间相关性最显著，R2=0.65。APAR与各指数间均存在极显著相关关系（P<0.01）；其中，APAR与NDVI之间相关性最大，相关系数为0.74。然而RVI的一元二次多项式更好地表达了二者间的回归关系，决定系数最大（R2=0.58）。由MSAVI和NDVI组成的双变量模型估算精度最好，误差为77gC•m-2•yr-1，精度为83%。洪河、红星、南瓮河三个样区苔草群落植被NPP平均值分别为632、318和683gC•m2•yr-1；太阳辐射是影响湿地植被NPP的关键因素。