The JRC presented a new method to correct surface shapes of satellite imagery to study mountainous terrains.© EU, 2009 (A-M. Petrescu)
New method to study mountainous terrains with satellite imagery
JRC scientists obtained remarkable results in correcting surface shapes of satellite imagery to study mountainous terrains, with a combination of topographic correction algorithms and statistical methods. The JRC presented its new method in a recently published article in the IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.
Proper pre-processing of remote sensing imagery is required to limit possible discrepancies due to atmospheric, radiometric and topographic effects, such as confusion between clouds and bright soils. Whereas most of these factors have been studied and their correction methods established, topographically induced illumination differences, confusion between dark shadows and water bodies for example, are still being investigated and there is no universal technique widely accepted in the remote sensing community.
To fill this gap, the JRC tested two digital elevation models (3D representations of a terrain's surface), a pre-classification/stratification approach to identify the strata according to the vegetation covers and several statistical correction methods, in study areas from 3 continents (Asia, Africa and South America) with different land covers. The pre-classification/stratification approach was used to split the different land cover types into strata which were corrected individually with the selected topographic correction method in order to achieve better reduction of the terrain effects (for example illumination effects due to the orientation of the slope, occlusions by mountains, …). Images from four different sensors systems were tested and processed to encompass different land cover types, temporal variations in solar illumination and a range of reliefs.
The obtained methodology represents an optimal solution for an operational topographic correction system with global dataset processing capabilities. Some improvements could still be made to increase the separation accuracy among the strata and the correctness of water detection.