A compromise between resolution and coverage

Aerial data is collected by aircraft, typically planes or drones, flying relatively close to the ground. This proximity allows for a high degree of detail, with resolutions of up to 5 cm x 5 cm. These highly detailed snapshots are vital for high-precision tasks across a variety of sectors, such as urban planning and infrastructure development. As acquiring new data with an aircraft is relatively expensive, data is often available on a country-wide basis, with updates every 1-3 years. To give an impression of the detail that can be captured, here's an aerial image of the Lego House:

A high-resolution aerial image of the Lego House in Billund, Denmark (https://legohouse.com/).

Satellite data is, as the name implies, collected by satellites in orbit around the Earth. They provide a broad and consistent view of the Earth's surface, albeit at a lower resolution than aerial data due to their distant position. Although satellite data may not match the raw detail of aerial imagery, its strength lies in its broad coverage and regular data capture. Sentinel-2, for instance, covers the entire globe every 5 days. In addition to the capturing images in the visible spectrum, Sentinel-2 also captures data in a variety of other wavelengths, such as infrared, which enables us to analyze aspects such as plant health or water quality from space. Let's take a look at satellite imagery 3 days apart to see the Danish Spring weather in action:

The city of Aalborg, Denmark, on March 5.

Aalborg, Denmark, 3 days later on March 8 after a heavy snowfall.

Why are these types of data suddenly important?

Multiple fundamental improvements have led to a significant decrease in the cost of acquiring, distributing, and analyzing Earth observation data, including aerial and satellite data. For example, reusable rockets have made satellite launches significantly cheaper, and technological improvements have made satellites significantly smaller. Once the data have been acquired, increase internet bandwidth and advances in cloud computing have made distribution and analysis much cheaper. So, data that was previously only available to governments and large corporations have now become available to everyone.

On top of the improvements to the cost and data quality, AI and deep learning have revolutionized our ability to analyze and interpret aerial and satellite data. We will dive much more into this in later articles, but the short version is that where we once needed a human to analyze the data, we can now use AI to do it for us. This means that we can analyze much more data, much faster, and with much higher accuracy than ever before. Or, in a language for business executives, we can now carry out advanced analyses much cheaper than previously.

Just to give a taste of the scale of the analyses, the Sentinel-2 programme captures approximately 1.6 TB of imagery per day. That is a lot of data. We will dive much more into specific applications in later articles, such as analyzing all agricultural fields in Denmark, annotating sports fields, and monitoring carbon capture in Timor Leste (hint: see this LinkedIn post from our collaboration with Klimate.co).

Bringing it All Together

Aerial data provides high-resolution snapshots of the Earth's surface, while satellite data provides a broad and consistent view of the same. In Atla.ai, we aim to democratize access to these data, making them more accessible and affordable, without long contract negotiations, license hassles, or complex APIs. On top of the raw data, we also provide specialized APIs with custom analyses to solve specific tasks.

Stay tuned and follow us on LinkedIn as we start to announce our upcoming services, with advanced analytics layers on top of the raw data, accessible to everyone. And remember to keep an eye out for our next articles, where we'll dive deeper into specific applications of aerial and satellite data, and how these data, combined with new AI methods, can help bring new possibilities to businesses.