GRASS GIS as Geocomputational Engine

FOSS4G NA 2024, St. Louis Missouri

2024-09-10

Corey White
corey.white@openplains.com

Center for Geospatial Analytics at NC State University

OpenPlains Inc.

Vaclav Petras
vpetras@ncsu.edu

Center for Geospatial Analytics at NC State University

Markus Neteler
neteler@mundialis.de

Mundialis

This talk was funded by the US National Science Foundation (NSF), award 2303651.

Corey White

What is GRASS?

  • GRASS is geospatial computational engine
  • GRASS is a modeling platform
  • GRASS is a development framework
  • GRASS is not your Papi’s Desktop GIS

GRASS as a Geoprocessing Engine

  • Fast, trusted, science based tools
    • +400 core tools
    • +400 contributor addons
  • Modern Tooling
    • Jupyter Notebooks
    • R Studio
    • Python and C API
    • GUI

What is GRASS used For?

  • Advanced geospatial modeling and analysis
  • Data Science
  • Data Engineering

Who is using GRASS

Civil and Environmental

  • Researchers
  • Modelers
  • Data Engineers
  • Climate Scientists
  • Startups

Features

  • Hydrologly
  • Map Calculations
  • Geomorphology
  • Remote Sensing
  • Machine Learning
  • Lidar

GRASS in the Wild

Case Studies

Desktop

  • Civil Engineering Firm
  • Lidar/Raster Processing
  • Power desktops to simplify processing

Cloud

  • Environmental Engineering Firm
  • Remote Sensing
  • Machine Learning

HPC

  • Academia
  • National Scale Data Processing
  • Urban Growth Forcasting

Desktop-based Geoprocessing

So fast you usally don’t need scale

  • Handles large raster base calculations with ease

Desktop Case Study

Bohannan Huston

Civil Engineeering firm based out of New Mexico

Modeling & Anaylsis
  • 1m QL1 Lidar Data
  • Vegitation Analysis
  • Flow Lines Deliniation

Desktop Case Study

Bohannan Huston: Tech Specs

Data: North Carolina Phase 4 Geiger - images: 22,519

Process Hardware Time (sec) Time (min)
SVF Generation Mac M2 Max 790.8 13.2
Mac M1 Max 889.2 14.8
Image Tiling Mac M2 Max 407.8 6.8
Mac M1 Max 519.1 8.7
Building Prediction Mac M2 Max 7537.5 125.6
Mac M1 Max 9641.4 160.7

Cloud-based Geoprocessing

Tools

  • actinia - GRASS REST API
  • Apache AirFlow - ETL
  • OpenPlains - Interative Geospatial Modeling

Cloud Case Study

Mudialis: Free Data With Free Software
  • GRASS GIS
  • GRASS GIS Addons
  • actinia

Cloud Case Study

Mudialis: Detecting buildings, green roofs and trees
  • Official cadastral plans are lagging behind reality.
  • Detection based on point clouds and 7.5 cm aerial imagery (RGB-NIR)
New GRASS GIS addons
  • m.analyse.buildings
  • r.extract.greenroofs
  • m.analyse.trees

Cloud Case Study

Mudialis: Image classification workshop with actinia

HPC-Based Geoprocessing

HPC - High Performance Computing

HPC Case-Study

NCSU Henry2 cluster
  • Nodes: 933
  • Cores: 13844
  • Python, R, GRASS GIS

HPC Case-Study

NCSU: CONUS Urban Growth Model

HPC Case-Study

NCSU: CONUS Urban Growth Model

Example: Scaling urban growth model

  • r.futures addon: FUTURES urban growth model
  • 30-m resolution (16 billion cells) on HPC
  • 50 stochastic runs annually from 2020 to 2100

Summary

  • GRASS is geospatial computational engine
  • GRASS is a modeling platform
  • GRASS is a development framework
  • GRASS is not your Papi’s Desktop GIS

Questions

Return Home