Anna is a geospatial research software engineer with PhD in Geospatial Analytics. She develops spatio-temporal models of urbanization and pest spread across landscape. As a member of the OSGeo Foundation and the GRASS GIS Project Steering Committee, Anna advocates the use of open source software in research and education.
Although GRASS GIS has been used for big data processing for a while now, you may think that some esoteric knowledge is needed to take full advantage of its computational power. The purpose of this talk is to demonstrate simple ways to parallelize your computations in GRASS GIS, that are applicable whether you are working on your laptop or HPC. I will give an overview of the state of parallelization of individual tools, show benchmarks, and introduce you to other GRASS GIS parallelization tricks. I will use examples relevant to land change modeling and share our experience with simulating urban growth at 30m pixel across the contiguous United States (16 billion cells) using FUTURES simulation implemented in r.futures addon. This talk is for all levels of expertise, although basic Python or GRASS GIS knowledge will be advantageous.
GRASS GIS is a well established, all-in-one geospatial number cruncher with Python interface, command line, and GUI, with new major version 8.0 released in spring 2022.
FUTURES is an open source urban growth model specifically designed to capture the spatial structure of development. It can accommodate the input of a variety of datasets with different spatial extents and can be coupled to other models. FUTURES is implemented in r.futures GRASS GIS addon.
The new GRASS GIS version 8.2 is a special edition including all new features developed during Google Summer of Code 2021. One of the enhancements is the parallelization of several raster modules by means of OpenMP, an implementation of multithreading to speed up massive data processing. Another exciting new feature is much improved, the Jupyter notebook support. Here, a new python package (grass.jupyter) is available which allows to interactively visualise maps and time series given the integration with folium.
The graphical user interface in version 8.0 introduced faster and more streamlined startup without a need for a welcome screen. For even more convenience, version 8.2 adds an experimental single window layout with familiar look-and-feel.
Related to raster data, a new metadata class called semantic labels can now be added to raster maps. Examples of semantic labels are aerial or satellite spectral bands, dataset names in remote sensing products (ndvi, evi, lst, etc), or any custom names.
At community level, we have developed a student grant program and, thanks to the move to GitHub, we have welcomed numerous new contributors.