FOSS4G Conference, Boston 2017

Summary notes and highlights from the conference

Conference August 14 – 18, 2017 Boston, MA

Conference by number

1,100 attendees

250 talks/sessions

56 workshops – two days.

I attended 4 separate workshops for two days and attended 24 talk sessions for three days. Below are my summary notes. Highlight of the conference were keynotes from Paul Ramsey and Joe Cheng  

• Workshop

1. Browser-based Geoprocessing with Turf.js and leaflet
   • turf.js client-side geoprocessing using turf.js library and integrating with leaflet.js library to build webmap browser based simple geoprocessing application.  
2. Classification of remote sensing images with the Orfeo ToolBox and QGIS
   • Orfeo is image processing application that can operate as standalone desktop application or use within QGIS desktop
3. From WebODM to QGIS
   • Open source Sfm software with web interface based on OpenCV algorithem. Software can take images taken from drone and process to develop cloud points and orthomoasic photos to 3D model. 
4. Processing Lidar and UAV point cloud in GRASS GIS
   • Exploring capabilities of GRASS GIS 7.2 lidar point cloud data processing.

Conference Sessions:

Leveraging Open Source software to build data/mapping tools in local government – Chris Whong NYC Planing Lab

• Talked about using various open source to build tool and applications
• using AuthO, turf.js Cartoserver, mapbox-gl, React, Digital Ocean.
• Lesson learned – use existing tools (no customized design/development) bootstrap, eliminate dependencies of large software 
• build modular (example of Jane maps, name after Jane Jacobs) 

Why your map sucks and you don’t even know it – Will Cadell, Sparkgeo.com 

• Map analytic company that analyzed 15 million user session. 
• Results: 25% of users don’t touch the map, once it’s loaded, user doesn’t move around
• Double loaded (multiple times) caused by bad coding
• 10% of activities happens at very first 1% of the duration
• Need to tell the “INTENT” of the story very quickly
• Most users do not click to explore the map but pan continously 

Let them expire! – One company’s experience with fewer proprietary software licenses – Paul Wickman GIS technology Director RESPEC

• Using QGIS to them was NOT alternative to Arc but other option
• Open Source option is not just free but means value
• “Price is what you pay, Value is what you get”
• Reasons for migrations
  • Per user scale model won’t work for current ESRI licensing model
  • It’s about business with value and felt that ArcGIS value has dimished in value
  • Look for value not ideology and need to thought out the plan.  “I don’t need the drill, just a hole”

An Introduction to containerizing geospatial apps with Docker – James Milner, 3D Repo

• Docker container is encapulated environment which runs application.https://docs.docker.com/get-started/
• Docker eliminate “It works FINE in my machine but NOT yours” issue. Get rid of dependency issue.
• Docker Container vs VM (virtual machine)
  • VM isolates everything (no OS access) but Docker container use native OS. 
  • Docker container means running docker images.
• By not having to load OS separately (not like VM), you don’t lose efficiency. 
• One can use docker image to load Geoserver + Postgis i.e. from Docker Hub https://hub.docker.com/r/crcsi/geoserver-postgis/

Offline First Mapping – Calvin Metcalf, AppGeo

• Using Web Service Worker, this can intercept http request and can save the map tile locally. 
• It is browser based script that runs in the background, separate from a web page
• Service workers can periodic background sync & push notifications that normally require native application but can do them through the browser. 
• Service workers currently only support Chrome and Firefox browsers.
• Here is a good introductory implementation of this technology in blog writeup 

Advanced geospatial technologies: The new powerful GRASS GIS 7.2 release – Helena Mitasova, NC State University.

• Not your old GRSS (Geographic Resources Analysis Support System)  from Army Corps of Engineering days, Fascinating story of GRASS GIS told by William Shatner back in 1987.
• Lot of new features and capabilities including raster/vector data analysis, cartographic tools, image processing, advance LIDAR processing tools and many more with latest version of GRASS GIS.

R-Studio:  Coding as first resort – Joe Cheng (CTO of RStudio)

• Most folks use coding as last resort after using GUI but speaker showed why it should be other way around. 
• GUI – developed design by developer/designer
• CLI – task directed by user themselves
• Use code to extend functionality of existing such as QGIS to repeat and to scale. 
• Main benefit of using code 
  • Concise
  • Highly interactive ( inform, return results after each line of code)
  • Fast iteration
• R – John Chamber originator  is to give user ability to interact and explore data. This programming language is heavily based on statistics and visualization. 
• R Studio contains many tools, map + text + graphic/table + narrative to compose data to tell story. 
• Shiny – can generate fully interactive web map using R studio.
• R Markdown – framework to crate html, report/PDF and dashboard. 

DIY Mapping with drones and open source in a humanitarian context – Dan Joseph, American Red Cross

• Many reasons for DIY drone instead of commercial, off-the-shelf product.
  • Fix your own machine out in the field
  • Ability to access parts locally
  • Lower cost
• Workflow as following 
  • Using DIY drone map to collect information 
  • then to process using OpenDroneMap to stich images together then share them through Open Aerial Map which then can be 
  • used in iD Editor to update OpenStreetMap database for everyone to use. 

LiveDroneMap – an Automatic Real-time UAV mapping solution – Janwoo Cheon, Seoul University, South Korea.

• Real-time drone imagery data being captured and process in real-time data to get result back instantly. 
• Live data processing to get instant on-the ground event evaluation without long-lag data processing  time. 

Data Driven Style (DDS) for fast GL maps – Molly Lloyd, Mapbox

• Using OpenGL (open graphic library) and WebGL, one can style using JSON file to stylize map features. 
• Mapbox using these two technologies to use in their Mapbox GL-JS as styling using JSON data file for their vector tile styling web map. 

Mapbox GL: How vector maps work – Vladimir Agafonkin, Mapbox

• Original creator of Leaflet talked about inner workings of vector tile. 
• He said that the new era of cartography is here.
  • Raster tile era lasted about 30 years which talks to server to get the information
  • Vector map era is here that leverages client side, using web browser to display web map. 
  • Currently there are 94.8% coverage of WebGL/OpenGL in US and 91.5% for the world. 
  • OpenGL renders triangle (Earcut) to display various features 

Migrating to Vector Tiles -Lizzi Silvinski, NBT Solutions. 

• Talked about integrating vector tile into existing raster tile system. 
• Goals for the project 
  • Dynmaic styling
  • GIS functionality 
  • Editing tools
•  Proof-of-Concept in github here. 
• Using leaftlet draw to edit in vectortile 
• Wrote own node.js to serve tiles. 

State of QGIS – Kurt Menke, Larry Shaffer

• Various features coming up QGIS 3.0 
• Major feature of having meta data catalog 
• Improve digitizing 
• Composer is completely rewritten and single layout with multiple pages
• Summary from blog post 

Vector tiles from OpenStreetMap with OpenMapTiles and TileServer GL – Petr Pridal, Klokan Technologies GmbH

• Provide various vector tile format for numerous data products and web map serving options.
• OpenStreetMap data in vector tile format available by various region
  • Whole planet data vector ~ 54 GB
  • West Coast ~ 1.9 GB
  • Washington State ~ 205 MB
  • Also can extract customized area 
• Hillshade and contour lines are also available for download
• Workflow is to download OSM vector tile and serve using either OpenMapTIles Mapserver of docker image or use  Tileserver-GL locally
• Downloadable cloudless satellite photo – 250 GB, 20m pixel resolution with zoom 13 level. 

Using open-source tools and high-resolution geospatial data to estimate landscapes’ visual attributes – Helena Mitasova, Anna Petrasova, NC State University.

• Using Geomorphons  – model based using machine/computer vision to interpret topology. 

Coupling a geospatial Tangible User Interface (TUI) and an Immersive Virtual Environment (IVE) using open-source geospatial and 3D modeling tools – Helena Mitasova, Anna Petrasova, NC State University. 

• System of software (GRASS GIS) and various hardware to do the analysis of natural environment planning and analysis. 
• Background information here which was developed in NC State University OSGeo Research and Education laboratory.
• Great tools/use case  to apply in environmental analysis and modeling.