Francesco Martinuzzi
I am Francesco Martinuzzi, a PhD student in Physics and Earth Sciences at Leipzig University in Germany. I am under the supervision of Prof. Miguel D. Mahecha and Dr. Karin Mora at the Remote Sensing Centre for Earth System Research RSC4Earth. My research is kindly funded by the Center for Scalable Data Analytics and Artificial Intelligence ScaDS.AI. I am also part of the ELLIS Society as a PhD affiliate.
In my PhD project I explore the consequences of extreme events on the biosphere using Machine Learning models and Dynamical Systems theory.
Sessions
In the rapidly evolving field of remote sensing, spectral indices are pivotal for monitoring and deciphering Earth system dynamics, including vegetation dynamics, water bodies ecosystems, and fire regimes. As the range of available spectral indices expands, the need for comprehensive catalogues and computational tools becomes crucial. Enter the Awesome Spectral Indices (ASI) - a novel, machine-readable, and open-source catalogue of spectral indices for Remote Sensing applications . ASI also promotes accessibility and application by integrating with three official Application Programming Interfaces (APIs) tailored for Python, Julia, and Google Earth Engine's JavaScript API. This integration facilitates the computation of a vast number of indices, adaptable to varied user experiences, data availability, and processing needs.
Our interactive workshop is structured into five engaging segments, each designed to showcase the potential of ASI in various environments:
- Introduction to Awesome Spectral Indices (15 mins): This segment unveils the ASI catalogue, highlighting its standardisation, the comprehensive range of spectral indices, and their properties. Participants will also get a demonstration of “Espectro”, a Streamlit web app, for querying spectral indices.
- ASI Python Mastery (30 min): Dive into “spyndex”, ASI’s Python API. This hands-on session guides attendees through exploring and computing spectral indices for several Python objects such as pandas Data Frames, numpy arrays, and xarray data arrays. Participants will leverage real-world data from STAC, such as Sentinel-2 and Landsat-9, applying parallel computing techniques with dask. This session is versatile, accommodating both Google Colab and personal Python environments.
- ASI in Google Earth Engine (20 min): Google Earth Engine (GEE) stands as a colossus in remote sensing analysis. Here, we introduce “spectral”, a JavaScript module for GEE, offering access to ASI spectral indices. Practical exercises will involve computing indices for images and image collections using GEE’s extensive data catalogue. (GEE account required).
- Harnessing ASI in Julia (30 min): Julia's focus in high-performance scientific computing makes it ideal for Earth system applications. “SpectralIndices.jl” brings ASI's power to Julia, enabling the computation of indices with remarkable efficiency . Participants will access indices and make computations for Arrays, Data Frames and YAXArrays. Participants should have a Julia environment pre-installed for this session, ideally with the necessary dependencies already downloaded.
- Collaborative Outro (25 min): The workshop concludes with an interactive discussion, brainstorming enhancements for ASI (from standard improvements to computational methodologies), real-world applications and integration with other software or initiatives. This collective discussion will be gathered into a white paper, aspiring for publication.
This workshop invites participants from all skill levels, from beginners to experts, offering a unique opportunity to explore the world of spectral indices and their application in remote sensing. Join us in harnessing the power of open-source software to explore and understand our planet like never before.