Webinar: Xarray & Dask for Geoscience

Access to very large geospatial datasets is becoming more and more common, e.g. with the availability of Satellite and Lidar imaging. Specialised code then needs to be written to handle these large volumes, leveraging parallel architectures to accelerate computations while handling memory constraints. Some toolboxes can reduce the complexity of the task. This webinar will present two of them:

  • the Xarray Python package providing a high-level data structure to manipulate annotated multidimensional arrays
  • the Dask Python package, for out-of-core (larger than RAM) computations, deployable on a laptop as well as on an HPC

The webinar will also discuss two case studies. Rose Pearson will present an application to hydrologically conditioned digital elevation maps (DEM) for flood mapping and Jan Schindler will detail results from a benchmark for a very large image analysis, with various Dask backends.



Some familiarity with Python is recommended to get the most out of this webinar.



About the speakers

Dr. Rose Pearson is a remote sensing scientist at Taihoro Nukurangi - NIWA and a visiting researcher at the Geospatial Research Institute. Her work primarily focuses on combining geospatial data, primarily LiDAR point clouds, to produce hydrologically conditioned DEMs and roughness maps for use in river flood modelling. Her research interests centre on surface generation and attribute mapping from a wide array of spatial and geospatial datasets.

Dr. Jan Schindler is a remote sensing & data scientist at Manaaki Whenua – Landcare Research. He investigates natural features, processes, and human impacts on the Earth's surface and in the atmosphere by working across disciplines. He employs data science techniques including ML/DL for spatial pattern analysis and environmental modelling and solves scale and integration challenges of disparate environmental datasets.

More Information

If you have any questions or would like more information about this session, please email training@nesi.org.nz

Also, subscribe to the NeSI training mailing list to stay updated about future training opportunities.


Event Date: 
Thursday, October 27, 2022 - 11:00 to 11:45