A machine learning approach for monitoring Brazilian optical water types using Sentinel 2-MSI


This project contains almost all the necessary steps and database used in the study to be published. Notice that steps from 5 to 9 the database here in GitHub is not available due to the large file size of satellite images. For reproducing the entire study, you should download the appendix data available in the publication.

Summary

01 Plot OWTs
02 SVM cross validation
03 Evaluation of algorithm noise sensitivity
04 Train classification algorithm
05 Define probability thresholds of novelty detection
06 Satellite and in situ match up of Rrs
07 Image Classification
08 ECP novelty evaluation
09 Funil owt frequency
10 Curuai owt frequency