Using high resolution DSM data to correct the terrain illumination effect in Landsat data

Terrain affects optical satellite images through both irradiance and bidirectional reflectance distribution function (BRDF) effects. Slopes facing the sun receive enhanced solar irradiance and appear brighter compared to those facing away from the sun. For anisotropic surfaces, the radiance received at the satellite sensor from a sloping surface is also affected by surface BRDF which varies with combinations of surface landcover types, sun, and satellite geometry (sun and sensor view, and their relative azimuth angle) as well as topographic geometry (primarily slope and aspect angles). Consequently, to obtain comparable surface reflectance from satellite images covering mountainous areas, it is necessary to process the images to reduce or remove the topographic effect so that the images can be used for different purposes on the same spectral base. The most common method of normalising for the topographic effect is by using a Digital Surface Model (DSM) and/or a Digital Elevation Model (DEM). However, the accuracy of the correction depends on the accuracy, scale and spatial resolution of DSM data as well as the co-registration between the DSM and satellite images. A physics based BRDF and atmospheric correction model in conjunction with the 1-second SRTM (Shuttle Radar Topographic Mission) derived DSM product released by Geoscience Australia in 2010 were used to conduct the analysis reported in this paper. The results show that artefacts in the DSM data can cause significant local errors in the correction. For some areas, false shadow and over corrected surface reflectance factors have been observed. In other areas, the algorithm is unable to detect shadow or retrieve an accurate surface reflectance factor in the slopes away from the sun. The accuracy of co-registration between satellite images and DSM data is important for the topographic correction. A mis-registration error of one or two pixels can lead to large error of retrieved surface reflectance factors in the gully and ridge areas (retrieved reflectance factors can change from 0.3 to 0.5 or more). Therefore, accurate registrations for both satellite images and DSM data are necessary to ensure the accuracy of the correction. Using low resolution DSM data in conjunction with high resolution satellite images can fail to correct some significant terrain effects. A DSM resolution appropriate to the scale of the resolution of satellite image is needed for the best results.