FOSS4GNA2018: The free & open source software for geospatial conference, St. Louis, MO.
I've just returned from a week in St. Louis, for FOSS4GNA, the Free & Open Source Software for Geospatial conference, where the predominant topics this year were increasing integration of R and RStudio into the geospatial toolkit, big geospatial data management and analysis, and the management and analysis of an increasing array of high-resolution and high-cadence satellite imagery sources. A number of the presentations and workshops were worth passing along to Stanford 'geo' users, and are noted, below.
New data sources
- Amazon is making global-scale datasets available in AWS, including OpenStreetMap queryable in Amazon Athena and Landsat 8 imagery in a public S3 bucket, as well as dozens of other datasets available through their Registry of Open Data on AWS
- Students and researchers can apply for free access to Planet.com's Daily multispectral image of the Earth at 3m resolution at: https://www.planet.com/markets/education-and-research/
- The National Geospatial-Intelligence Agency is making DigitalGlobe's high resolution imagery (5 areas of interest: Rio, Las Vegas, Paris, Shanghai & Khartoum) freely available as part of the Spacenet Challenge on AWS. The next Spacenet challenge is "Road Network Extraction."
Workshops and tutorials
- Planet.com's Sara Safavi and Dana Bauer, presented a 3-hour workshop using Planet's custom Jupyter Notebook-based Interactive Guides for leveraging their daily image of the Earth at 3m resolution to create mosaics, produce subsets and even run simple machine learning tasks.
- R, and in particular RStudio, as a powerful desktop GIS alternative was a strong theme, and Tina Cormier, Remote Sensing Scientist at TellusLabs, delivered a great demonstration of using R Notebooks, in RStudio, for geospatial data science. Find her R Notebook code from the presentation, here: https://github.com/tacormier/FOSS4GNA_2018.
- Tom Buckley of Metropolitan Transportation Commission presented a 3-hour introductory level workshop on "R Studio as a Traditional Desktop GIS Replacement," at which I also found out about Robin Lovelace's newly available "Geocomputation with R."
- The Go Spatial team's workshop on "Running a vector tiles stack" with their tegola server, written in Go, provides a low-barrier solution to creating, serving and styling vector tiles in a local environment, or as a web service.
Other highlights included:
- Drew Bollinger's demonstration of DevelopmentSeed's label-maker (which prepares labeled image chips for machine learning) provides a simple means of leveraging OpenStreetMap.org as a vast machine learning training data source.
- Paul Ramsey's "groovy" & irreverent [Spatial] SQL Festival talk (with fantastically verbose speaker notes), highlights some novel, useful and efficient ways to use PostGIS for wrangling spatial data in PostgreSQL. Stanford affiliates can try pure PostGIS "in the cloud" using their stanford.edu email address at https://stanford.carto.com.
For those interested in Free & Open Source Software for Geospatial, the FOSS4G International Meeting will be held Dar es Salaam, Tanzania at the end of August, this year.