Data base: Atmospheric Boundary Layer Height Determination through Digital Image Analysis from low-cost sensors in Urban Environments
Description
Atmospheric raw data in the Valle de Aburrá, Medellín - Colombia. Backscatter coefficient data from ceilometer, part of the SIATA (from Spanish for Early Warning System of Medellín and the Valle de Aburrá) and JPG landscape images of urban zone of the valley.
Files
Steps to reproduce
Each image was initially read to extract its metadata, including the date and time of acquisition, focal length, geographical coordinates (latitude and altitude), RGB information, and image dimensions. The images subsequently underwent a series of processing steps. First, images were filtered according to their average brightness to exclude excessively dark or saturated scenes. The selected images were then converted to greyscale and their histograms were normalised to enhance radiometric consistency. Morphological binarisation was subsequently performed using a reference grey level to distinguish atmospheric structures from the background. Finally, morphological perimeter filters were applied to identify the pixels corresponding to the radiometric transition associated with the Atmospheric Boundary Layer. These pixels constituted the input required for the geometric estimation of ABL height. The geographical reference points were combined with the image-derived ABL pixels through the geometric interpolation model. This procedure transformed image pixel positions into real-world elevations, allowing quantitative estimation of the Atmospheric Boundary Layer height. Backscatter coefficient profiles were processed using noise reduction filters and signal smoothing techniques to improve data quality. The processed backscatter data were then represented as two-dimensional range-corrected images, enabling visual identification of the ABL height at the same observation times as the smartphone photographs.