Speaker
Description
Current static mapping approaches of ecosystem conditions are inadequate, given the fast pace of ecosystem transformations under the climate change regime. Our project “Time-varying AI-based mapping of ecosystem conditions and extents using multi-source Earth observation data cubes”, TEE Cube, aims to bridge this gap by developing a dynamic approach for mapping ecosystem conditions. The study focuses on the Bavarian Forest National Park in Germany. We analyse time series of spectral indices and plant biochemical traits using multispectral satellite imagery and hyperspectral data. We use Artificial Intelligence (AI) techniques to assess retrospective changes and to predict missing time steps of plant traits, providing detailed, long-term information on ecosystem dynamics.
We collect and process spaceborne multispectral data (Landsat and Sentinel-2) to generate a regional data cube covering the park from 1984 to 2024. Additionally, we combine spaceborne hyperspectral data (EnMAP, PRISMA, EO-1 Hyperion, and DESIS) and preprocessed airborne hyperspectral image archives (HySpex) with biochemical and biophysical traits to generate a regional biological condition data cube.
We develop a regional demonstrator data cube using the Earth System Data Cube (ESDC) framework to integrate diverse Earth Observation (EO) data relevant to mapping ecosystem extent and condition (Figure 1). In the next steps, we aim to establish a thorough comprehension of ecosystem dynamics and states by seamlessly incorporating multispectral and hyperspectral data-derived plant traits. The utilisation of advanced AI techniques specifically tailored for the analysis of three-dimensional data cubes will facilitate the reconstruction of plant traits and enable the discernment of long-term trends and patterns. These advancements will substantially enrich scientific understanding in remote sensing, AI, and ecosystem mapping.
Figure 1. Preliminary results of the showcase on the constructed VI data cube and hyperspectral plant traits data cube. (a): Landsat OLI-derived NDVI data cube; (b): EnMAP-derived chlorophyll data cube.
| Status Group | Postdoctoral Researcher |
|---|
