Scaling Up Forest Ecosystem Carbon Budget from Stand to Landscape: Impacts of Forest Structures

该项目的总体目标是通过综合利用遥感、生态模型和地面观测，提高我们对多维森林结构对扩大森林生态系统碳循环的影响的理解。虽然只为北卡罗来纳大学教堂山申请资金，但这是北卡罗来纳大学教堂山地理系和杜克大学环境学院之间的合作研究。这两个校区的专业知识和研究设施是互补的。拟议的研究将扩大杜克森林现有项目的范围和深度。该项目将采取必要步骤，将杜克森林森林生态系统和大气之间的质量和能量交换的理解转变为地区性的理解。该项目将使用Duke森林中的ameriFlux和Face站点作为锚点，通过综合使用遥感和生态系统模型来扩大规模。目前，全球的FLUXNET在全球拥有200多个通量塔，本项目中研究的扩大规模战略将使我们深入了解如何根据FLUXNET的测量结果扩大规模，以提高我们对全球碳循环的理解。该项目建议使用多传感器遥感数据绘制多维森林结构图，包括树木大小和密度、林龄、叶面积指数和亚像素树木覆盖率。遥感数据包括来自Ikonos/QuickBird的高分辨率(=1m)数字正射影像象限和星载图像、中分辨率(30m)Landsat图像和高分辨率(250m)MODIS/MISR图像。该项目将开发利用遥感数据的空间、光谱/时间和方向域的信息的算法。该项目可以大幅提高利用遥感提取详细空间植被信息的能力。该项目将使用一系列成熟的生态模型，每个模型都将以适当规模的森林结构来模拟陆地生态系统的碳循环。除了量化在模拟碳循环中忽略森林结构所造成的误差外，该项目还将导致对这些模型的重大改进。

The overall goal of this project is to improve our understanding of the impacts of the multi-dimensional forest structures on scaling up forest ecosystem carbon cycle from stand to landscape through the integrative use of remote sensing, ecological models and ground observations. Though funding is requested for UNC Chapel Hill only, this is a collaborative research between the Department of Geography at UNC Chapel Hill and the School of Environment at Duke University. The expertise and research facilities from the two campuses are complementary. The proposed research will expands the scope and depth of the existing projects in Duke Forest. The project will take the necessary steps to transform the understanding of mass and energy exchange between forest ecosystems and the atmosphere at Duke Forest to a regional understanding. The project will use the AmeriFlux and FACE sites in the Duke Forest as the anchor points for scaling up through the integrative use of remote sensing and ecosystem models. Currently the global FLUXNET has over 200-flux towers world wide, and the up-scaling strategy investigated in this project will provide in-depth understanding of how to scale up from FLUXNET measurements to improve our understanding in global carbon cycle. The project proposes to use multisensor remotely sensed data to map multi-dimensional forest structures, including tree size and density, stand ages, leaf area index, and subpixel tree cover. The remotely sensed data include high-resolution (=1m) digital orthophoto quads and space-born images from Ikonos/QuickBird, medium resolution (30 m) Landsat images, and coarse-resolution (250m) MODIS/MISR images. The project will develop algorithms to use information from spatial, spectral/temporal and directional domains of remotely sensed data. The project can substantially enhance the use of remote sensing to extract detailed spatial vegetation information. A series of well-established ecological models will be used in the project, each of which will take forest structure at the appropriate scale to simulate terrestrial ecosystem carbon cycle. In addition to quantifying the errors caused by omitting forest structures in simulating carbon cycle, the project will lead to major improvements to the these models.