Remote Sensing Trainee (Summer 2025)

Company details

The SPACEBEL Earth Observation Applications department provides and manages Earth observation services and geographic information systems enabling end-users to access useful information for high-quality digital monitoring.

The main activities of the department are

• The development and supply of services based on satellite and drone imagery for the forestry, agriculture and mining sectors.
• The development and supply of services for hazard management (floods, storms, ground movements, wild fires, etc.)
• Applications based on hyperspectral, multispectral, thermal, radar – high and medium resolution – imagery from satellite and airborne sensors, as well as drones.

Job Details

Function

As a remote sensing trainee, you will work on the consolidation of multi- and hyperspectral data for the detection of vegetation anomalies based on a hybrid model.

The aim of your internship is to acquire knowledge of data and advanced techniques enabling detailed characterisation of plot dieback over time.

One of the objectives of the project in this area is to prepare hyperspectral data for the CHIME space mission, which focuses on agroforestry and forest health.

The main goal is to create simulated datasets integrating various satellite and terrestrial sources. A processing pipeline is being developed to generate these data and a hybrid methodology combining physical models and machine learning is being used to extract information on vegetation. The validation of these data is based on detailed field measurements and forest health indicators provided by the DNF (Office Wallon de la Santé des Forêts - Walloon Forest Health Office) and the CRA-W (Centre Wallon de Recherches Agronomiques - Walloon Centre for Agronomic Research).

The purpose of the internship is part of the SENWISE project funded by the European Space Agency.

Your role will involve the following tasks:

• Use of different tools, model quality and performance analysis to support the recognition of anomalies related to vegetation health (in forest and agroforestry contexts);
• Processing of hyperspectral satellite images;
• Analysis of complementarity with Sentinel-2 Time Series;
• Validation of the method with available field data (via partnership with the DNF).

Profile

You will be using the following tools/techniques:

• Python (Pytorch, TensorFlow);
• GIS (QGIS);
• PROSAIL simulations ;
• Self-supervised autoencoder model ;
• Multi-layer perceptron (MLP);
• RMSE/MAE.

Offer

Internship period: ideally two months as of July (or August) 2025

Internship location: SPACEBEL offices in Angleur (Liège - Belgium).