Orbital Insight Hackathon 2020

Once a year at Orbital Insight, we have an internal hackathon so everyone in the company has a chance to showcase the possibilities of geospatial analytics, including new technical improvements and ideas that inspire our product roadmap. This year, we were fortunate enough to have everyone from our global offices in one place at our headquarters in Palo Alto, CA, where teams were able to collaborate and discuss ideas throughout the week, making Friday our “hack day” where projects were presented company-wide.

Some decided to be on a team and complete a project as a group, while others decided to go solo and do a project on their own. This year, there were two main categories: Best Use Case and Best Technical Improvement.

Each presentation was incredible, but we’d like to take a moment to highlight the winners of Best Use Case — “The Mallacootas” and their impressive work on the Australian wildfires.

Team Mallacoota and Australian Wildfires
The Mallacootas taking a break from brainstorming in order to pose for a quick team photo.

Exploring Data Around Australian Wildfires:

Exploring data around Australian wildfires with artificial intelligence

Massive wildfires have burned over 15 million acres of land across Australia in the current fire season. Two dozen people have been killed. Tens of thousands have been displaced. Hundreds of millions of animals from across Australia’s unique ecosystems have lost their lives. The smoke plume from the fires in southeastern Australia has darkened skies as far away as New Zealand, across an area wider than Europe.

The Mallacootas took a structured approach to identify the areas of highest need and impact, using available open-source data. They analyzed “active fires” detections using Short Wave Infrared (SWIR) imagery across three months (November 2019 through the end of January 2020) and compared those to the same months in the previous year, finding an over 90% increase in overall fire detections with 451,094 in the previous year compared to 860,319 in the current period.

Active fire detections, Australian wildfires

They used active fires data points to ADM2 boundaries and calculated the difference in total detections between the two time periods, finding a clear outlier for this current time period. From this, they validated the extreme difference in wildfires and also found the inflection point of the increase (in late November). After finding the “where” and “when” the fires increased, they used Orbital Insight GO’s land use classification capabilities to understand the area and environment prior to their impact. With those results, they were able to understand the buildings and populated areas across the state, quantifying the value of this through comparing with OSM records.

Next, they used the above foundational results to tip-and-cue many follow on analyses, choosing two primary routes: assessing feature damages (period-to-period) and semantic land-use change analysis for buildings and forests, and assessing population movement and migration (measured through geolocation data).

Geolocation trends based on land use
Above: This map shows the number of buildings in the area based on population as measured by daily unique device counts (DUDC). This indicates the differences in device count as people are evacuating, showing people are moving to the coast of Mallacoota to evacuate by boat.
Australian fires geolocation trends
Deadly Australian wildfires closing in, geolocation data shows foot traffic patterns and ship detection

Key Takeaways:

With a two-day turnaround, The Mallacootas provided transparency and data on:


  • Initial dispersal — accounting for all populations, mapped and unmapped
  • Possible destruction across those on a per-building basis
  • Patterns-of-life deviation and possible migration

Forestry and Vegetation

  • Macro-level damage assessment/dispersal

Beach Town Refuge and Evacuation

  • Foot traffic, ship tracking, and news and media corroboration in Mallacoota

With the unfortunate increase in extreme weather events, Orbital Insight has applied geospatial data and AI in innovative ways to track and quantify environmental trends. Outputs of this nature are leveraged for future logistical planning and disaster recovery.

Other Projects

Other impressive projects under each category were:

Best Use Case:

  • Exploring and Comparing Twitter Data and Geolocation Data
  • Geoparsing of news articles and satellite tasking emails
  • Correlating Restaurant Reviews with Geolocation Data
  • Measuring Methane Levels from Satellite Imagery
  • Broad Spectrum Geolocation-Based Land Use Classifier

Best Technical Improvements:

  • Faster Algorithm for Geo-hashing
  • Improving Planet Ship Detector
  • Tracing Our APIs Using Jaeger
  • Using Tilt to Deploy GO 100% Locally for Rapid Development

A big thank you to everyone who participated. It was no surprise for us to see the outstanding work and dedication put into each project that was presented, and we are extremely proud to have so many talented people on our team.

After a great week full of team events and collaboration, concluding with our hackathon, we’re inspired and looking forward to working hard this year in order to bring you geospatial data solutions that fit your daily needs, easier and faster than ever before.

You’ve seen our culture and what we’re doing at Orbital Insight. We’re hiring!




Perspective is a powerful thing.

Recommended from Medium

Building footprint comparison in Cambodia: Orbital Insight and OpenStreetMap

Regression Performance Evaluation Methods

Weighted Polynomial Regression with NumPy

How to Master Pandas for Data Science

Visualizing UFO Sightings with Plotly.py

Introduction to Linear Regression using Python

Choose a good place to stay in Seattle

Predicting Trucks’ Brands with TensorFlow

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Orbital Insight

Orbital Insight

Geospatial analytics for an interconnected world.

More from Medium

Op-ed: Tackling biases in natural language processing

Storytelling via Temporal Networks

NLP for genre predictions on FFnet: an antithesis to utilitarianism