🌎 LIVE-GIS

Background

Geographic information systems (GIS) allow their users to integrate, visualize, explore, and understand geospatial data. Geospatial data includes, maps, satellite images/datasets, and any other datasets with spatial locations on the surface of the earth.

The dominant solution in this space is ArcGIS (from esri), closed commercial software which, for many researchers, is prohibitively expensive and lacks full integration with open-source tools. Open source projects such as QGIS provide a free and full-featured desktop tool for experts, but can be intimidating for new users, and don't integrate seemlessly with browser-based tools.

Open-source tools such as leaflet.js, xarray, and GeoPandas within JupyterLab provide key components for a complete GIS solution, but connecting these tools to each other and to other visualization libraries is a challenge -- the challenge that LIVE-GIS will solve.

Current Capabilities

The ipyleaflet project integrates leaflet.js into the JupyterLab ecosystem as a Jupyter Widget. This allows users to interactively refine their visualizations with the full power of Jupyter notebooks and any other Python GIS libraries/tools.

While this provides a powerful environment for visualization of a small number of datasets, even the most basic analysis tasks require command line use and an understanding of the underlying Python APIs for the libraries.

SAVE

A prototype LIVE environment, The Search Analysis and Visualization Environment (SAVE) project integrates glue-jupyter viewers based on bqplot with ipyleaflet in a Jupyter environment to provide for easy authoring and display of GIS dashboards using open source tools. Configuration and layout of different viewers is supported by Voilà Gridstack.

SAVE is supported by a collaboration between researchers at Harvard and at Google. SAVE was used to produce the visualizations and dashboards for several of the Climate Justice Design Fellows at Harvard University. Those projects can be viewed at the Climate Justice Design Fellowship website.

WorldWide Telescope (WWT)

An alternative to the purely web-based ipyleaflet-based approach is available in the desktop version of glue, using WorldWide Telescope (WWT). In addition to its astronomical utility, WWT can also overlay data on the Earth's surface. The flight-paths-over-Boston image seen here comes from the "Learn to fly glue, fast" video, which showcases the use of the web-based WorldWide Telescope viewer as a plug-in to the desktop version of glue.

Jupyter GIS

Several projects are already underway at The Eric & Wendy Schmidt Center for Data Scince & Environment with the goal of adding critical pieces of the GIS analysis stack into tools that can integrate alongside existing tools running in JupyterLab.

The Future

The full LIVE-GIS environment will combine open-source GIS-specific viewers with the general-purpose visualization tools available in all LIVE environments. The glue back-end will provide a simple interface for connecting datasets, including via smart AI-enabled linking, and for propagating selections in one viewer into other viewers. Controls for exploring and subsetting time-series data will be available alongside simple geospatial manipulations, such as calculating summary statistics over regions-of-interest and basic geometric manipulations.

Multi-dimensional linked-data exploration

LIVE-GIS will include support for gridded geospatial data (e.g. satellite data, model forecasts) through xarray and cloud-first formats such as Zarr and Cloud Optimized GeoTiffs. The main map viewer will be leaflet via ipyleaflet, but other map viewers (such as Google Maps Platform) will be evaluated for inclusion based on technical considerations and community requests. The GeoPandas library will be used to reading geographic data and manipulating geometries.

N-D labeled arrays and datasets in Python

An open-source library for interactive maps

Python tools for geographic data

The goal is to create an environment that focuses on exposing the most common GIS analysis and visualization tasks in a simple graphical user interface, while providing command-line access through Jupyter to the underlying libraries so that experts can perform more complicated manipulations. By building on web-first technologies, the same environment used for analysis tasks will also be able to produce standalone web pages for the display and sharing of results.

Projects such as CryoCloud and Pangeo provide curated software environments that can be deployed as JupyterHubs in the cloud through infrastructure-enabling projects such as 2i2c. These projects provide a pipeline for getting new versions of LIVE-GIS quickly into the hands of people working on real-world geoscience problems, enabling rapid feedback and development.

Page updated

Report abuse