遥感反演植被理化参数的光谱和空间尺度效应

Review of spectral and spatial scale effects of remotely sensed biophysical and biochemical vegetation parameters

植被理化参数是生态系统中碳和养分等物质循环与能量交换的重要指标,利用遥感技术反演是获取区域及全球植被理化参数的重要手段,但光谱和空间尺度效应的存在,限制了源自不同遥感传感器植被理化参数产品的统一应用。阐述了遥感反演植被理化参数光谱尺度效应的概念及其产生原因,主要从光谱波段位置和波段宽度两方面对国内外相关研究进行了介绍和评述。同时,从遥感反演植被理化参数的空间尺度效应产生原因、空间异质性描述方法和空间尺度转换方法等方面对其国内外研究现状进行了归纳和评述。最后,总结了遥感反演植被理化参数光谱和空间尺度效应研究的不足之处和发展趋势,并指出光谱和空间耦合效应的研究将是一大趋势,而在生态学等领域形成的尺度效应研究的理论和方法也值得借鉴参考。

Biophysical and biochemical vegetation parameters are significant indicators of carbon and nutrient cycling and energy exchange in ecosystems. Remote-sensing inversion is an important method of obtaining regional and global biophysical and biochemical vegetation parameters. However, spectral and spatial scale effects result in differences among remotely sensed bio-parameters from different sources. This limits their use in unified applications and affects the precision of ecological models that utilize them as input parameters. This review examines the concept of the spectral-scale effect of remotely sensed bio-parameters and its causes. The spectral-scale effect is the phenomenon in which differences exist in the remotely sensed bio-parameters obtained using different band resolutions and band positions. The spectral-scale effect has two main aspects：the response of different regions of the spectrum to different internal structures, and the response to the chemical composition of green vegetation. Moreover, the response intensities and bandwidths of different sensors are different. This paper reviews relevant research on the spectral-scale from the perspectives of spectral-band position and bandwidth. Studies on band position usually analyze finite and discontinuous bands through statistical methods, but the physical characteristics of spectrum itself are not accounted for. Studies have not reached a consensus on whether a wideband or narrow band is more suitable for vegetation physicochemical parameter estimation. The manner in which spectral bandwidth influences extraction of vegetation information, construction of vegetation indices, and bio-parameter estimation is not yet clear. Research on the influence of the spatial effect on remotely sensed bio-parameters was summarized and analyzed from the perspectives of understanding the causes of the spatial effect, spatial heterogeneity description methods, and spatial scaling methods. The main conclusion of existing research is that the spatial-scale effect of remotely sensed bio-parameter inversion is related to use of nonlinear inversion methods and spatial heterogeneity of the parameters. However, most studies focused on leaf area index （LAI） and are aimed at relatively simple estimation models based on vegetation indices such as the normalized differential vegetation index （NDVI）. Commonly used spatial-scale effect description methods are the variance method, the fractal method, the variation-function method, and the wavelet-transform method. Future research should seek to explain the physical meaning of the spatial-scale effect and to systematically and quantitatively simulate and analyze its characteristics. Although the spatial-scaling method has been adapted for use in spatial-scaling modeling from mathematical, statistical and physical methods, a great gap still exists between the theory of spatial-scaling models and their practical application. This review elucidates the need for more studies on spectral-scale effects in biophysical and biochemical vegetation parameters. In-depth spatial-scale effect studies aimed at the various vegetation parameters and inversion methods, using more complex physical models, are required. Studies on coupled spectral and spatial effects are a possible future direction, and theories and methods from ecology and other fields may provide valuable guidance.