Free archive of georectified and atmospherically corrected Landsat satellite images create a large range of opportunities for environmental research. However, the topographic effects in images are typically normalized regionally by end-users, and it remains uncertain if this procedure is always necessary. Our objective was to assess the effect of topographic normalization on the fractional tree cover (Fcover) modelling in a tropical mountain landscape, in Southeastern Kenya. We carried out topographic normalization by C-correction for all available Landsat images between June 2012 and October 2013, and examined if normalization improves Fcover regressions. The reference Fcover was based on airborne LiDAR data. Furthermore, we tested several vegetation indices and seasonal features (annual percentiles and means), and compared three digital elevation models (DEM). Our results showed that the fit of Fcover models did not improve after topographic normalization in the case of ratio-based vegetation indices (Normalized Difference Vegetation Index, NDVI; Reduced Simple Ratio, RSR) or Tasseled Cap Greenness but improved in the case of Brightness and Wetness, particularly in the period of the lowest sun elevation. RSR was the best vegetation index to predict Fcover. Furthermore, SRTM DEM provided stronger relationship with cosine of the solar incidence angle than ASTER DEM and regional DEM based on topographic maps. We conclude that NDVI and RSR are robust against topographic effects in the tropical mountain landscapes throughout the year. However, if Tasseled Cap indices are preferred, we recommend topographic normalization using SRTM DEM.
DOI:
https://doi.org/10.1016/j.jag.2016.05.008Altmetric score:
Dimensions Citation Count:
Publication year
2022
Authors
Adhikari H; Heiskanen, J.; Maeda E E; Pellikka P K E
Language
English
Geographic
Kenya
