Seminar Paper on BOA Image Calibration & Topographic Correction

This seminar paper was written in the course Analysis & Modelling (Remote Sensing) in my second semester of CDE.

Raw image data derived from satellites is in many cases not ready to be used for analyses or calculations. Often a series of pre-processing steps is necessary to prepare the image data for the actual study purpose, among those are bottom-of-atmosphere image correction and topographic correction. This paper explores the principles of these two calibration concepts by providing an overview of available methods, tools and important considerations for their use. Sen2Cor, a free atmospheric correction tool for Sentinel-2 imagery is used to demonstrate the capabilities of image calibration techniques by conducting corrections on imagery and by explaining every step of the workflow in detail. Although using most tools, both proprietary and free, does not require a deep understanding of the background processes, it is sill advisable to have a basic understanding of them. With this, from the multitude of available algorithms, tools and methods, a more well-thought-out decision can be made as to what is the best choice or a specific project.

A. Bottom-of-Atmosphere Image Calibration

Gas absorptions and aerosol scattering influence electromagnetic radiation passing through the atmosphere before reaching a satellite sensor. These effects can be removed with atmospheric correction methods to retrieve surface reflectance values. Such corrections are essential when performing tasks like time-series analyses to make images comparable across space, time and sensors.

B. Topographic Correction (TOC)

Topographic conditions let equal surfaces appear lighter or brighter depending on the slope and aspect. Topographic correction removes these differences by calculating reflectance values for a perfectly flat surface. Advanced TOC strategies use DEMs and land cover classes as ancillary data to derive reflectance values considering diffuse irradiances, incident angles and land cover class surface roughness (Stratified Non-Lambertian TOC).

The result of a topographic correction compared with its original image (Torralba I. S. 2015. Evaluation of topographic correction algorithms on satellite images.)