© Hanna Meyer

Uebersat: Spatio-temporal transferability of satellite-based AI-models

  • Logo of the Federal Ministry for Economic Affairs and Energy
    © BMWi


    AI methods are increasingly used in the context of satellite-based earth observation to generate spatiotemporal environmental information, for example for monitoring land use, biodiversity patterns or effects of climate change. AI models are usually trained on the basis of local field observations with the aim to make predictions for a larger area and/or a new time for which no reference data are available. However, the transferability of the models to new locations and/or new times is rarely questioned in current AI-applications and models are often applied to make predictions far beyond the geographic location of the training samples. Especially in heterogeneous landscapes the new locations might differ considerably in terms of their environmental characteristics from what has been observed in the training data. This is problematic, since machine learning algorithms can fit very complex relationships, but at the same time are weak at extrapolation. Predictions for new locations/times that differ in their characteristics from the training data must therefore be considered very uncertain, which calls for a method to assess the area of applicability of AI-models.
    In this project new methods for the analysis and improvement of the transferability of satellite-based AI models in space and time will be developed, with the aim to assess and increase the quality of earth observation products through the use of innovative AI techniques. The new methods will be integrated into cloud-based processing chains to optimise data-driven applications and deliver more reliable monitoring results.

  • Information

    Project lead: Hanna Meyer, Edzer Pebesma (IfGI)
    Team@GEOI: Marvin Ludwig, Jonathan Bahlmann, Hanna Meyer, Edzer Pebesma
    Funding: BMWi
    Term: 2021 - 2023

  • Publications

    Peer-Review Papers

    • Mila C, Mateu J, Pebesma E, Meyer H. 2022. ‘Nearest neighbour distance matching leave-one-out cross-validation for map validation.’ Methods in Ecology and Evolution n/a. doi: 10.1111/2041-210X.13851.
    • Meyer H, Pebesma E (2021): Predicting into unknown space? Estimating the area of applicability of spatial prediction models. Methods in Ecology and Evolution 2021. doi: 10.1111/2041-210X.13650.

    Conference Contributions

    • Meyer H, Pebesma E (2021): Estimating the Area of Applicability of Remote Sensing-Based Machine Learning Models with Limited Training Data. Contributed to the IEEE International Geoscience and Remote Sensing Symposium IGARSS, Brussels, Belgium. doi: 10.1109/IGARSS47720.2021.9553999.
    • Meyer H, Ludwig M, Pebesma E (2021): A new method to assess the area of applicability of spatial prediction models. GfÖ Virtual Annual Meeting 2021, 30 August to 1 September.
    • Ludwig M, Pebesma E, Meyer H (2021): Increasing the transferability of global spatial prediction models. GfÖ Virtual Annual Meeting 2021, 30 August to 1 September.
  • Theses

  • Teaching

    Mapping the area of applicability of spatial prediction models (OpenGeoHub2021)

    The OpenGeoHub Summer School took place online from 1 to 3 September 2021. We gave a workshop on „Mapping the area of applicability of spatial prediction models“. All course material as well as video recordings of the lecture/practical session is available online.