Pixel-based Minnaert Correction Method for Reducing Topographic Effects on a Landsat 7 ETM+ Image

The topographic effect on land surface reflectance is an important factor affecting quantitative analysis of remotely sensed data in mountainous regions. Different approaches have been developed to reduce topographical effects. Of the many methods, the Minnaert correction method is most frequently used for topographic correction, but a single global Minnaert value used in previous research cannot effectively reduce topographic effects on the remotely sensed data, especially in the areas with steep slopes. This paper explores the method to develop a pixel-based Minnaert coefficient image based on the established relationship between Minnaert coefficients and topographic slopes. A texture measure based on homogeneity is used to eva-luate the topographic correction result. This study has demonstrated promising in reducing topographic effects on the Landsat ETM image with the pixel-based Minnnaert correction method. Introduction The topographic effect has long been recognized as a problem for quantitative analyses of remotely sensed data (Teillet et al., 1982; Ekstrand, 1996; Tokola et al., 2001), therefore, topographic correction has become one of the important image preprocessing steps in the application of remotely sensed data in mountainous regions. Common atmospheric calibration models, such as the 6S (Second simulation of the satellite signal in the solar spectrum) model (Vermote et al., 1997) and image-based dark object subtraction models (Chavez, 1996), are based on the assumption that the surface is a flat horizon with a cloud-free atmosphere. These models can eliminate the impacts caused by sensor instruments, solar irradiance, solar zenith angle, atmospheric scattering, and absorption, but cannot eliminate the topographic impacts. In order to reduce topographic effects on remotely sensed data, Pixel-based Minnaert Correction Method for Reducing Topographic Effects on a Landsat 7 ETM Image Dengsheng Lu, Hongli Ge, Shizhen He, Aijun Xu, Guomo Zhou, and Huaqiang Du considerable research with Thematic Mapper (TM) and Satellite Probatoire d’Observation de la Terre (SPOT) data has been conducted in mountainous areas (Teillet et al., 1982; Jones et al., 1988; Leprieur et al., 1988; Civco, 1989; Itten and Meyer, 1993; Meyer et al., 1993; Sandmeier and Itten, 1997; Gu and Gillespie, 1998; Gu et al., 1999; Allen, 2000; Tokola et al., 2001; Blesius and Weirich, 2005; Gitas and Deverux, 2006). Methods used for reducing topographic effects include (a) band ratio (Holben and Justice, 1980), (b) topographic correction models, such as Minnaert correction model and statistical-empirical approach (Civco, 1989; Colby, 1991; Allen, 2000; Bishop and Colby, 2002; Riano et al., 2003), (c) integration of digital elevation model (DEM) data and remote-sensing data (Walsh et al., 1990; Franklin et al., 1994), and (d) a combined correction model of atmospheric and topographic effects (Conese et al., 1993; Richter, 1997; Sandmeier and Itten, 1997). Although different approaches have been used for topographic correction in previous research, an effective approach to reduce topographic effects is not available. Topographic effects result from the differences in illumination due to the position of the sun and the angle of the terrain. This causes variation in brightness values. A steep slope often produces serious shadows in the aspect facing away from the sun resulting in significant variation of illumination for areas with different slopes and aspects. A simple approach to reduce the topographic effects is to use the cosine model (Jensen, 1996), which assumes that the surface reflects incident solar energy uniformly in all directions, i.e., a Lambertian reflector. Any variations in land surface reflectance are caused by the amount of incident radiation. However, the cosine model only models direct irradiance of a pixel on the ground. It ignores diffuse sky light that may weakly illuminate a shadowed area. This implies that weakly illuminated areas in the terrain receive a disproportionate brightness when the cosine model is applied. Teillet et al. (1982) noted that the cosine model is not particularly useful in areas of steep terrain where incident angles may approach 90°. Allen (2000) also indicated that shaded and weakly illuminated slopes may appear white, and almost all north-facing slopes become over-illuminated using the cosine model. An alternative way to account for non-Lambertian behavior is the use of a Minnaert-correction model which makes use of PHOTOGRAMMETRIC ENGINEER ING & REMOTE SENS ING Novembe r 2008 1343 Hongli Ge, Aijun Xu, Guomo Zhou, and Huaqiang Du are with the School of Environmental Technology, Zhejiang Forestry University, Lin’An, Zhejiang, China 311300 ([email protected]). Dengsheng Lu is with the School of Environmental Technology, Zhejiang Forestry University, Lin’An, Zhejiang, China, and the Anthropological Center for Training and Research on Global Environmental Change (ACT), Indiana University, Bloomington, IN 47405. Shizhen He is with the East China Institute of Forest Inventory and planning, State Forestry Administration, Jinhua, Zhejiang, China. Photogrammetric Engineering & Remote Sensing Vol. 74, No. 11, November 2008, pp. 1343–1350. 0099-1112/08/7411–1343/$3.00/0 © 2008 American Society for Photogrammetry and Remote Sensing 06-085.qxd 10/11/08 3:51 AM Page 1343