A physics-based atmospheric and BRDF correction for Landsat data over mountainous terrain Fuqin Li a, ⁎, David L.B. Jupp b , Medhavy Thankappan a , Leo Lymburner a , Norman Mueller a , Adam Lewis a , Alex Held b a National Earth Observation Group, Geoscience Australia, GPO Box 378, ACT, 2601, Australia b CSIRO, Marine and Atmospheric Research, GPO Box 3023, ACT 2601, Australia abstract article info Article history: Received 29 September 2011 Received in revised form 21 June 2012 Accepted 23 June 2012 Available online 21 July 2012 Keywords: Landsat BRDF Atmospheric correction Terrain illumination correction Steep terrain affects optical satellite images through variations it creates in both irradiance and bidirectional reflectance distribution function (BRDF) effects. To obtain the corrected land surface reflectance and detect land surface change through time series analysis over rugged surfaces, it is necessary to remove or reduce the topographic effects. In this paper a physics-based BRDF and atmospheric correction model that handles both flat and inclined surfaces in conjunction with a 1-second SRTM (Shuttle Radar Topographic Mission) de- rived Digital Surface Model (DSM) product was applied to 8 Landsat scenes covering different seasons and terrain types in eastern Australia. Visual assessment showed that the algorithm removed much of the topo- graphic effect and detected deep shadows in all 8 images. An indirect validation based on the change in cor- relation between the data and terrain slope showed that the correlation coefficient between the surface reflectance factor and the cosine of the incident (sun) angle reduced dramatically after the topographic cor- rection algorithm was applied. The correlation coefficient typically reduced from 0.80–0.70 to 0.05 in areas of significant relief. It was also shown how the terrain corrected surface reflectance can provide suitable input data for multi-temporal land cover classification in areas of high relief based on spectral signatures and spec- tral albedo, while the products based only on BRDF and atmospheric correction cannot. To provide compar- ison with previous work and to validate the proposed algorithm, two empirical methods based on the C-correction were used as well as the established SCS-method to provide benchmarks. The proposed method was found to achieve the same measures of shade reduction without empirical regression. Crown Copyright © 2012 Published by Elsevier Inc. All rights reserved. 1. Introduction Topographic correction of satellite images over mountainous or hilly areas is very important (Liang, 2004), especially when the data are to be used for land cover mapping and monitoring over time. Steep terrain affects optical satellite images through both irradiance and Bi-directional Reflectance Distribution Function (BRDF) effects (Dymond et al., 2001; Gu & Gillespie, 1998). Slopes facing toward the sun receive more solar irradiance and appear brighter in satellite images than those facing away from the sun (Iqbal, 1983, Chapters 10, 11) where the darker pixels are often said to be “shaded”. In addition, for anisotropic surfaces, the radiance received at the satellite from in- clined surfaces is also affected by surface BRDF. That is, the signal re- sults from the combined effects of surface land cover structure interacting with the sun and satellite geometry (sun and view and its relative azimuth angle) as well as topographic geometry (e.g. slope and aspect angles). Finally, all of these factors affect the inver- sion of land surface parameters and applications that aim to detect land surface conditions and changes through time series analysis. Following some years of development and improvement, physics- based models for BRDF and atmospheric correction are relatively mature (Li & Strahler, 1985; Li et al., 1995; Li et al., 2010; Schaaf et al., 2002; Vermote et al., 1997). However, they have been applied mostly to sur- faces that are essentially flat and seldom to significantly inclined sur- faces. In the past, reported Digital Surface Model (DSM) or Digital Elevation Model (DEM) based topographic correction has mostly been undertaken separately from BRDF and atmospheric correction. Typical examples are terrain illumination correction for Lambertian surfaces (Dozier & Frew, 1981, 1990) and the “C-correction” (cosine) method which is based on an empirical relationship between observed radiance from inclined surfaces and the cosine of the incident angle (Richter, 1997; Teillet et al., 1982). Gu and Gillespie (1998) proposed the Sun-Canopy-Sensor (SCS) topographic correction which accounts for some of the BRDF effects over forested mountain areas and Shepherd Remote Sensing of Environment 124 (2012) 756–770 ⁎ Corresponding author. Tel.: +61 2 62495867; fax: +61 2 62499910. E-mail address: Fuqin.Li@ga.gov.au (F. Li). 0034-4257/$ – see front matter. Crown Copyright © 2012 Published by Elsevier Inc. All rights reserved. doi:10.1016/j.rse.2012.06.018 Contents lists available at SciVerse ScienceDirect Remote Sensing of Environment journal homepage: www.elsevier.com/locate/rse