Machine learning based unmixing using the EnMAP-Box

Tutor: Akpona Okujeni & Sebastian van der Linden (Humboldt-Universität zu Berlin)

Tutorial description: This tutorial will focus on the use of machine learning based regression methods combined with synthetically mixed training data from spectral libraries for land cover fraction mapping. Such unmixing approaches are particularly relevant to exploit forthcoming spaceborne imaging spectrometer data, which bring along coarser spatial resolution compared to common airborne imagery. The expected wealth of data requires an efficient use of spectral libraries or related data sources. The workflow is fully implemented in a toolbox that was developed as part of the EnMAP mission preparation. This EnMAP-Box is delivered as plugin for QGIS3. Simulated EnMAP data, spectral libraries and reference information will be used for this exercise. The tutorial will (1) provide a brief general introduction into the work with the EnMAP-Box, (2) introduce a built-in spectral library viewer for handling of spectral libraries and meta data, (3) and demonstrate the “library based land cover mapping tool” for spectral unmixing.

For information about the unmixing approach and the EnMAP-Box see:

Okujeni, A., van der Linden, S., Tits, L., Somers, B., & Hostert, P. (2013). Support vector regression and synthetically mixed training data for quantifying urban land cover. Remote Sensing of Environment, 137, 184-197.

Okujeni, A., van der Linden, S., Suess, S., & Hostert, P. (2017). Ensemble Learning From Synthetically Mixed Training Data for Quantifying Urban Land Cover With Support Vector Regression. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 10(4), 1640-1650.

General overview:

Download and installation guide:

Requirements: Participants need to bring their own laptops with the latest versions of QGIS3 and the EnMAP-Box preinstalled. Exercise data will be provided.

Number of participants: max. 20

Date: 5 February 2019, 13:30 – 17:00

Location: Lecture room at Global Change Research Institute (CzechGlobe), Bělidla 986/4a, 603 00 Brno

Fee: Free of charge

Registration: Subscribe for the tutorial during abstract submission / registration via ConfTool