Paper_Coastal_V3

This repository contains the scripts and results used for the analysis of progradation and erosion in the Atrato River delta, supported in the paper 'Detection of erosion and progradation in the Colombian Atrato River Delta by using Sentinel-1 synthetic aperture radar data', where the methodology can be known in detail. At the root of the repository there are two folders, the first called 'Images' which contains 7 folders with images of the step-by-step process of study, and a second called 'Scripts' where 9 scripts are housed, with which the images of the first folder were obtained.

1. In the folder 'Images/1.Original' there are 32 SAR images captured by Sentinel-1 in this page, taken in pairs from 11/06/2016 to 30/05/2023 in the VH and VV polarizations. By running the 'Scripts/1.Rescale_Intensity.py' script, it is possible to have a better visualization of these images that are stored in 'Images/2.Rescaled'(32 images).

2. The SAR images contain a characteristic speckle noise that must be eliminated, for this, we used an autoencoder located in 'Scripts/Autoencoder_despeckling.h5' and executed by 'Scripts/2.Filter.py' script that uses the images in 'Images/2.Rescaled' folder (32 images). This autoencoder receives images of 512x512 pixels, by which it is necessary to divide each image, and after being filtered, they are saved in the folder 'Images/3.Filtered'(32 images).

3. The process of composing the images after filtering them can be detailed in this page. This process is executed by 'Scripts/3.Compose.py' script, which takes the images from 'Images/3.Filtered' folder (32 images), and they are saved in 'Images/4.Composite' folder (16 images).

4. To classify the water and non-water of the composite images, the Otsu algorithm was implemented in the script 'Scripts/4.Classify.py' was used. The band for classification chosen was VH, the selection of this band is supported by the histogram analysis of the script 'Scripts/4.1Histogram_analysis.py'. The chosen images are in 'Images/4.Composite' folder and are saved in 'Images/5.Classified'. The white area is the 'water' class, and the black area is the 'non-water' class.

5. To continue with the analysis, it was necessary to align all the images, as a reference, the oldest was taken dated 11/06/2016 and additionally, for display issues, 50 pixels were removed from each side of the image. This algorithm is executed in 'Scripts/5.Resgistration.py' processing the images in the folder 'Images/5.Classified' and saving them in 'Images/6.Registration'.

6. The 'Scripts/6.Heat_map.py' script uses the images from 'Images/6.Registration' to calculate the zones where there were changes in time, these changes can be seen in the heatmap saved in 'Images/7.Result/Heat/Heat_map.png'.

7. In the 'Scripts/7.Measure_Area.py' script the area in km^2 of the area classified as water and not water in 'Images/6.Registration' images are measured, and these measurements are saved in an Excel file located at 'Images/7.Result/Areas.xlsx'.

8. In the 'Scripts/8.Areas_Erosion_Progradation.py' script, the area in km^2 of the erosion and progradation zones is measured, taking the images from 'Images/6.Registration', these measurements are saved in an Excel file located at 'Images/7.Result/Areas_Erosion_Progradation.xlsx'.