40TB of public data became one conservation game

A grass fire was burning behind Raffael Hickisch during a video call on Tuesday, but his attention was on a different problem: turning fragmented environmental data into tools people can actually use.

Hickisch, a conservationist and cofounder of the Chinko Nature Reserve in 2014, has long relied on a patchwork of public datasets to track what is happening across protected land in the Central African Republic. That meant pulling together human settlement data, deforestation data, and NASA fire data in desktop software, then trying to make sense of it all.

“I hardly ever simulate it because it’s just a lot of work and my computer doesn’t even support it.”

That limitation has consequences on the ground. Protected areas often include seasonal grazing zones alongside areas that are fully off-limits, and enforcing those rules requires visibility across tens of thousands of square kilometers. As Hickisch put it, deforestation can be happening without governments seeing it simply because they do not know how to access satellite-derived information.

In January, Hickisch came across exe.dev through a friend. He took an idea he had written down in 2022—a platform for monitoring protected areas—and handed an expanded version to the company’s agent, Shelley. The result was a first prototype of Five Megapixel Conservation in a matter of hours, though Hickisch says getting the app to its current state took about 300 hours.

Recommended reading

Xona plans 258 LEO satellites for GPS-like service

The app, 5mp.globe, is designed to display global conservation activity based on GPX files that track the movements of park rangers, vehicles, and aircraft surveying protected areas. It aggregates that data daily, monthly, and annually into 100 sq km global pixels, then generates reports for protected area managers, governments, and nonprofits.

“I was super excited. I was like, 'Oh, this is crazy. Now I can do those things that I have proposed years before, but I don’t need the budget, just my time.'”

Austria LIDAR project and 100 parallel VMs

Hickisch then turned to his home country, Austria, which publishes annual nationwide LIDAR scans and high-resolution aerial photos. Individual files can be as large as 15 gigabytes, making them impractical to download wholesale. Using vsicurl, he found a way to read slices of the data instead.

He built another app that runs 100 VMs in parallel, each handling a different part of Austria, and applies a standard machine learning model to classify objects in the imagery. By combining tree height with altitude, Hickisch says he can estimate forest age and identify areas with the oldest forests or places most in need of protection.

After three months, he had covered about 30 percent of Austria and processed roughly 40 terabytes of raw data. The results are being stored on Zenodo, the open-access repository run by CERN.

A settlers-style game on top of real forest data

To make that dataset easier to engage with, Hickisch also built a simple settlers-style game. Players can buy parcels of land, reforest them, and uncover treasures. Click a parcel, and the game surfaces real LIDAR data from the Austrian study, including the height of the tallest trees there.

“It’s a stupid game, but the powerful thing is the two terabytes of data that are underlying it.”

Hickisch says the apps are open source, and he does not expect them to make money soon. What he wants instead is adoption: two countries using Five Megapixel Conservation to monitor their parks within the next few years. For him, the point is not just better conservation data, but giving people in places like the Central African Republic the ability to build their own software tools for the first time.