High-density point cloud algorithm to map trees in Amsterdam

Related theme: Data Intelligence Solutions

Amsterdam Municipality is working hard to update its database of green spaces. It encountered some challenges with the Gisib database and it was unclear how the dataset on trees matched with the situation on the ground. The municipality uses the database as a basis for its maintenance and safety policies. TAUW had previously completed a project for Amsterdam Municipality to map green spaces, and in 2023 we were asked to make a specific inventory of trees in the public space.

Challenge

For the previous project, we used aerial photos to determine areas with grass cover, shrubs and bushes. However, the tree cover makes it difficult to assess the amount of trees in a certain area, rendering aerial photos useless.

Drones were not possible either, because of flight restrictions from nearby Schiphol Airport. Field work was the only option left.

Solution

We worked with GeoZICHT to measure point clouds and design an algorithm that helped us extrapolate tree trunks. First we mapped the surroundings using a laser scanner. The data from the scanner was fed into a laptop and recreated digitally.

We then developed a high-density point cloud algorithm (HDPA) that was able to classify the point clouds into different objects, such as buildings, streets, shrubs and trees. In the following phase, we were able to isolate the trees from the visualisation. The algorithm was designed to detect cylindrical shapes. Based on various parameters the algorithm then determines whether these are actually trees or not.

That information was then relayed back to the municipal database and for the duration of a year, we conducted further fieldwork to see how accurate the algorithm was, adjusting the HDPA to its current level of accuracy.

Impact 

With the HDPA, the Municipality of Amsterdam has a good, up to date overview of the trees in the city. It will be far easier to conduct inspections of tree health and carry out mandatory maintenance.

The data set is also a valuable asset in heat stress modules. Knowing the density of trees in specific neighbourhoods helps policymakers to decide if climate adaptive measures are necessary or not.

The HDPA is now ready to be rolled out. Other municipalities will be able to benefit from this tool. And the application can easily be modified to map objects other than trees, such as street lanterns or rubbish bins.

Partnership 

TAUW, GeoZICHT and the external party engaged worked together on this innovative project with the municipality. The client appreciated the use of AI as an innovative tool while having a realistic view of the end product.

“The partnership between the City of Amsterdam and TAUW represents a pioneering approach to urban forestry management. By combining our efforts, we've successfully implemented the High Density Algorithm, leveraging state-of-the-art Lidar technology to map and monitor the city’s trees with remarkable accuracy. This collaboration has not only improved our urban landscape but also enhanced our environmental stewardship. The precise data obtained allows for more informed decisions regarding tree preservation, planting, and maintenance, ensuring that Amsterdam remains a green and vibrant city. This is a shining example of how technology and cooperative governance can lead to substantial environmental benefits.”