Internship experience at ICube-SERTIT
This report describes the activities carried out during the summer internship at ICube – SERTIT (Strasbourg, France). ICube – SERTIT is responsible for producing geo-information from space imagery. One of the main services offered by SERTIT is the Rapid Mapping Service. During disasters, this system is triggered to address the demands of risk management, crisis, and reconstruction of damaged areas.
My main duty consisted of bringing new insights and updating a fire detection algorithm developed by SERTIT. The algorithm is change detection driven and it is based on several indexes, such as Normalized Burn Ratio (NBR), Burn Area Index (BAI) and Normalized Difference Vegetation Index (NDVI), developed in a Jupyter Notebook (python). Despite being adapted to predict fire on Landsat and Sentinel-2 data, this internship focused mainly on Sentinel-2 imagery.
The internship’s tasks consisted in: understanding and illustrating the workflow of the implemented fire detection chain; identifying the algorithm's flaws during the post-processing stage; modifying the code to improve the zonal statistics thresholds, which used to exclude genuine fire clusters; Implementing the BAIS2 spectral index; Assessing accuracy by adjusting the thresholds on multiple test sites to find a generic set of values with the best output quality (measured by F1-score); Delivering a commented code ready to be used in operation in SERTIT’s rapid mapping service.
The algorithm was tested at several testing sites around the globe, obtaining comparatively higher F1 scores after the changes than before. However, a few testing sites, such as those in high latitude areas, remain a challenge, leaving room for improvement.
This internship was one of the most important experiences as a graduate student in Geoinformatics. It was my first opportunity to work fully in Python in a professional environment. Merci beaucoup, SERTIT, for providing me with such an interesting project and an extremely competent and thoughtful team!
Recomendations to future interns
- Start looking for potential internship positions and/or places you would like to work early enough;
- Do not hesitate to ask for advices and/or recommendations letters to the programme coordinators and/or professors;
- Ask as many questions as you can during your internship, it is also a learning experience!;
- Trust your skills!