I am working on Sentinel 2A images. Do these images need pre-processing? For example, topographic correction, atmospheric correction using well known methods?
You need to know, at which processing level the image is already available for atmospheric correction. Topographic correction does not depend on sensor. It is basically related to which time of the year the image was acquired. May be you can check this paper. https://www.researchgate.net/publication/303824782_The_effect_of_topographic_normalization_on_fractional_tree_cover_mapping_in_tropical_mountains_An_assessment_based_on_seasonal_Landsat_time_series
I have some script for topographic normalization written in R for sentinel image.
Best regards
Hari
Article The effect of topographic normalization on fractional tree c...
If you compare Sentinel-2 data of different dates, for example in a mult-temporal approach (change detection), your data should be radiometrically calibrated.
sen2cor is a free and open processing scheme which creates TOA reflectance (top of atmosphere) from L2A products of Sentinel-2. Reflectance data is corrected from variance caused by sun angle, daytime ect. Smaller haze is also removed but thick clouds remain.