I3 Project
INTERDISCIPLINARY | INTEGRATED | INTERACTIVE
Instructors: As. Prof. Hermann Klug
Automated Atmospheric Correction of Satellite Imagery for Inland Water Quality Assessment
Purpose of the project
This project is part of the summer semester 2024 I3 project of the Msc. Applied Geoinformatics at the University of Salzburg. It aims to automate the atmospheric correction process of satellite imagery for water applications. This will be done by incorporating “ACOLITE”, a well-established AC processor, in a Python script, to allow for optimized time series analysis. Additionally, to test the hypothesis that Sentinel-2 MSI Level-1C data atmospherically corrected using ACOLITE produce more accurate water spectra compared to Sentinel-2 MSI Level-2A, a comparison of spectral signatures will be performed. Lake Balaton (Hungary) will be used as a study area. As a complementary (optional) objective, different WQ indices will be calculated for the lake to assess the script’s applicability on retrieving water quality parameters.
Abstract
Atmospheric correction (AC) of satellite imagery over inland waters is a complex, computationally intensive, and time-consuming process, but it is also necessary to remove atmospheric effects and retrieve the pure signal from the water's surface. Different processors have been developed to perform this task, one of which is ACOLITE, a well-known AC software developed to correct images for atmospheric effects over inland and coastal waters. This study aims to integrate ACOLITE into a Python script to automate the process of AC and facilitate time-series analysis of satellite imagery for water quality (WQ) monitoring. Lake Balaton was chosen as a study area, an eutrophic lake with complex optical characteristics that underwent a major algal bloom event during the summer of 2019. To assess the performance of the AC, Sentinel-2 MSI Level-1C images from July 2019 were processed and spectral signatures were extracted and compared to Sentinel-2 MSI Level-2A data, which are atmospherically corrected for land applications. To further evaluate the workflow, the Maximum Chlorophyll Index (MCI) and the Normalized Difference Chlorophyll Index (NDCI) were calculated. Comparison between the signatures showed no major distinction and further analysis is required to determine which dataset is more suitable for inland WQ assessment. A qualitative comparison of the indices with in-situ chlorophyll-a concentration measurements confirmed that both indices are effective tools for chl-a monitoring.
You can find detailed information on the methodology, scripts and project results in the project GitLab page and Wiki as well as the paper.