GIScience Jena at the EGU 2022 Conference

This blog post is work in progress. The following are the contributions of members of the GIScience group to the EGU 2022 conference. Abstracts with associated papers are highlighted more prominently.

Geospatial machine learning methodology

Brenning: Novel approaches to model assessment and interpretation in geospatial machine learning

• This talk shows how spatial prediction error profiles (SPEPs) and spatial variable importance profiles (SVIPs) can be used to explore the spatial behaviour of black-box machine-learning models. It furthermore proposes an approach for visualizing and interpreting machine-learning models from a transformed perspective, which is particularly useful when the feature space is high-dimensional or features are correlated.
• Spatial model assessment and interpretation:
  • blog post
  • arXiv preprint
• Model interpretation in transformed space:
  • blog post
  • arXiv preprint
  • R package wiml

Natural hazards and Earth surface processes

Goetz et al. (with A. Brenning): Modelling debris-flow source-area connectivity and impacted stream channels in the semi-arid central Andes of Chile using random walks and network analysis

• Associated Github repo sedconnect
• Related earlier journal paper in NHESS, associated Github repo with R package runoptGPP, and non-technical summary (in German)

Maraun et al. (with R. Knevels, H. Petschko and A. Brenning): A severe landslide event in the Alpine foreland under possible future climate and land-use changes

• Associated Nature blog post
• Associated journal article in Comm. Earth Env.

Wang & Brenning: Combining active-learning approaches with support vector machines for landslide mapping

• Active learning is a machine-learning strategy that aims to reduce the amount of training data required, directing the labeling expert to the most ‘relevant’ locations. We demonstrate the potential of this approach for landslide mapping.
• Associated journal paper in Remote Sensing
• Our next step is transfer learning, see preprint in Geoscientific Model Development

Lucchese et al. (with A. Brenning): Landslide Susceptibility Modeling of an Escarpment in Southern Brazil using Artificial Neural Networks as a Baseline for Modeling Triggering Rainfall

• Artificial neural networks are versatile modeling tools – this contribution is a first step towards modeling landslide susceptibility across a large region in southern Brazil… stay tuned…
• Results from a CAPES-funded research visit in Jena

Adeniyi et al. (with A. Brenning): Assessing machine-learning algorithms for digital soil mapping in an agricultural lowland area: a case study of Lombardy region

• Results from an ERASMUS+-funded research visit in Jena within the EC2U university alliance

Hydrological modelling

Künne et al.: Prediction of flow intermittence in Drying River Networks using a process-based hydrological model

• Related article: grant proposal in Research Ideas and Outcomes

Mimeau et al. (with A. Künne): Inter-comparison of climatological datasets for the hydrological modelling of six european catchments

Watson et al. (with A. Künne): Climate change and hydrological extremes: predicting and preparing for the impact on water resources

Climate and Earth system science (collaboration with MPI-BGC)

Bogdanovich et al. (with A. Brenning): Geographically varying temperature thresholds for societal attention and health impacts of heat waves

Ruiz-Vásquez et al. (with A. Brenning): Exploring the relationship between S2S temperature forecasting errors and Earth system variables

• Associated journal article: EGUsphere preprint

Additional contributions

Lima et al. (with J. Goetz & H. Petschko): Exploiting newly available landslide data to verify existing landslide susceptibility maps a decade after their implementation

Mishra et al.(with R. Knevels, H. Petschko and A. Brenning): Attribution of 2009 extreme rainfall & landslide event in Austria

Steger et al. (with R. Kohrs and J. Goetz): A data-driven approach to establish prediction surfaces for rainfall-induced shallow landslides in South Tyrol, Italy