From Chaos to Clarity: How a Geo-Data Cube Can Help Fix the Netherlands
All we had to do was eat, poop, reproduce, and let others do the same. Yet we face housing shortages, growing inequality, and climate pressure. A geo-datacube shows how to untangle these challenges.
These days, we have data on everything: population, economy, education, nature, you name it. Yet all that information often remains locked away in separate tables and dashboards. Lining up those datasets side by side and trying to make sense of them takes time, energy, and usually a fair amount of frustration. In this article, I’ll show how a geo-data cube makes it effortless to bring all that data together and visualize it in a clear and structured way. This gives you instant insight at a glance, enabling you to put that knowledge directly to use for smarter policy decisions.
I developed these concepts during my ten years of work experience with Earth observation, GIS data, and AI, combined with a Master’s in GIS & Remote Sensing at Wageningen University.
What is a geo-data cube?
The concept originates from the world of scientific climate models: mountains of satellite imagery and weather data from different sources are merged into one large spatial dataset, enriched with a time dimension. This is what we call a geo-data cube.
A geo-data cube allows you to examine, for example, how temperature, precipitation, and soil moisture evolve per square kilometer over multiple years. All of this information is neatly tied together in one organized raster.
In science, such cubes make it possible to detect relationships that are otherwise hard to recognize, such as between prolonged drought and vegetation loss.
Put simply, a geo-data cube overlays a fine-grained grid across a region and fills each cell with attributes spanning multiple years and themes. Think of it as a spreadsheet for each location, spread geographically across the whole region. This principle is now being applied to societal challenges in South Holland.
Figure 1: Example of a geo-data cube for climate models, combining satellite data with rainfall, land use data, and a time dimension.
Simplifying analyses with hexagons
The grid I use consists of hexagonal cells (based on Uber’s H3 hexagon grid:
https://h3geo.org/. Hexagons are equidistant to their neighbors and suffer less from edge effects compared to squares. This results in fairer, cleaner analyses, and the system is scalable and hierarchical.
Each hexagon has a unique index, a 15-character alphanumeric code. This makes it easy to determine which hexagons are adjacent. Because the grid works efficiently across multiple spatial resolutions, you can seamlessly combine dozens of themes and run analyses.
Figure 2: A hexagonal grid with unique indexing.
Examples of data in a geo-data cube (2018–2023):
Demographics: population, age, density (CBS, 100m x 100m grid with statistics).
Economy: employment, GDP per capita (CBS regional key figures).
Education: accessibility of schools, number of pupils (CBS).
Land use: agriculture, nature, built-up areas (LGN, based partly on satellite data (https://lgn.nl/).
Water quality: measurements from all water boards and Rijkswaterstaat on surface water (nitrogen, phosphate, etc from https://www.ihw.nl/informatievoorziening-kaderrichtlijn-water-krw).
How to extract insights from a geo-data cube
Once the data is integrated into the geo-data cube, you can, for instance, run a multi-criteria analysis (MCA). Think of identifying areas where water quality is poor precisely where land use is intensive. Or highlighting locations where education facilities lag behind and economic participation is low.
This results in maps where high-risk cells light up immediately. An MCA can therefore provide crystal-clear, location-based insights.
Using LLMs in a way that actually makes sense
With a geo-data cube, you can use a language model (LLM) like ChatGPT to ask simple questions such as:
“Where in South Holland is population density high?”
“Where is water quality poor?”
“Where are few schools located?”
The model translates your question into a search query, scans all relevant data, and returns a clear answer. Suddenly, you no longer need to be an expert in databases or statistics to understand these insights.
This approach makes it possible for anyone not just to look for problems but also to explore solutions, and then communicate everything visually on a map.
Recognizing trends and exploring scenarios
Because a geo-data cube contains multi-year data, it enables you to track trends and think through scenarios. Are differences between regions growing or shrinking? Do you see that investing in one theme, such as education, also creates positive spillovers in other themes like economic growth or quality of life?
The geo-data cube allows you not just to look back, but also to think ahead—testing different values and seeing whether valuable insights emerge for policymaking.
Figure 3: Tracking trends and running scenario analyses can support planning for the future.
The value for policymakers or curious citizens
Instead of staying stuck with average tables and vague generalizations, a geo-data cube lets each cell on the map speak, giving every patch of land its own meaning. This way, policies can target exactly those places where the warning signs are most urgent, rather than applying broad one-size-fits-all solutions.
At a glance, you can see how themes like water quality, education, and economy overlap. All data in this approach is built from publicly available datasets and open-source code, accessible to anyone who wants to see what’s happening in their own area.
This makes policy choices more evidence-based, transparent, and easier to justify. As new questions arise, you can simply add more data and adapt.
This approach not only reveals where inequalities accumulate, but also where policies are effective or new opportunities arise. The visual maps generated from the geo-data cube make policy discussions accessible to everyone: whether you’re a policymaker, data professional, or resident.
The future of the geo-data cube
The geo-data cube concept can be applied in so many ways that we are only at the beginning of its potential. AI and LLM technologies are advancing rapidly, and policymakers and researchers are quickly recognizing their usefulness.
At the Province of South Holland, we are exploring how to apply this technique to identify opportunities and solutions for major challenges. The technology is evolving fast, and interest in it is growing just as quickly. That’s why I started this blog, where I’ll share more about it in the future.
If you have specific questions or would like me to write a post on a particular aspect of the geo-data cube, feel free to reach out. If you want to follow developments more generally, I invite you to follow my updates. And of course, I’m always open for a coffee.
Get started yourself
A geo-data cube transforms a region like South Holland from abstract averages into meaningful micro-stories, making hidden inequalities and emerging opportunities visible.
Collaborate with me by sending a message, or follow my upcoming Jupyter notebooks to try it yourself.
PS: If you’re already curious to see examples of data cubes, check out the use cases from this EU Horizon project: https://fairicube.nilu.no/ , or this ESA blog on satellite image data cubes: https://eo4society.esa.int/resources/euro-data-cube/.