The impact of urban trees on air pollution: an Amsterdam case study

According to the National Institute for Public Health and Environment (RIVM), ten to fifteen thousand people in the Netherlands die as a consequence of air pollution or noise every year. Especially in and around Amsterdam the health risks are severe. In fact, Amsterdam is among the cities in western Europe with the worst air quality. To combat air pollution, it is vital to know which pollutants show up where and in what concentration. Also the distribution of urban green plays an important role in this.

 

Currently, the best satellite instrument for measuring air quality is the Dutch TROPOMI instrument that was launched by the European Space Agency in 2017. With 7 km x 3,5 km, this instrument has over six times the spatial resolution of its predecessor, the OMI instrument. One of the most important aspects of reducing air pollution is to investigate which policies are most effective in controlling emissions into the atmosphere. This can be achieved by trend monitoring. Unfortunately, since high-resolution satellite data has only recently become available, detailed trend analyses at a regional level from before 2017 are not possible. However, what if an artificial intelligence (AI) algorithm could be taught to transform the low-resolution OMI data into data of higher quality that matches the resolution of the TROPOMI instrument? Although this might sound like science fiction, it is exactly the kind of challenge that fits the tech company Sobolt.

Sobolt is one of the leading AI companies in the Netherlands. We are specialized in extracting value from data with a focus on sustainability issues. Over the last year, supported by the Copernicus Incubation Program, Sobolt has used deep learning to upgrade the historical OMI data to the same resolution as the TROPOMI data. With these upgraded OMI observations, new trends and patterns in air pollution can be identified.

Earlier this year, researchers from Utrecht University and Erasmus University Rotterdam validated the artificially upgraded OMI data against real TROPOMI data. The upgraded data represents realistic air quality data with a resolution similar to that of TROPOMI. However, more development is required and planned to safely replace the OMI data.

 

Fig. 1: An OMI air quality map of the Netherlands (left) with the corresponding artificially upgraded map (right).

 

An important factor in air quality is the distribution of trees. Urban trees can improve air quality by reducing the air temperature and by removing pollutants from the air. Moreover, they can reduce energy consumption in buildings, which leads to a decrease in air pollutant emissions. In general, larger trees have a more positive impact on the environment than smaller ones due to a larger leaf area.

To map the distribution of trees, Sobolt has partnered up with Bomenwacht, a high-quality consulting company in the area of trees. Over the last few months, multiple cities across the Netherlands have been mapped. The trees are automatically extracted with AI from 3D point clouds that are acquired by a lidar sensor that is mounted to the roof of a car. Subsequently, the trees and their physical characteristics are processed by i-Tree, a state-of-the-art, peer-reviewed software application developed by the USDA Forest Service to calculate the beneficial impact of urban trees and forests on the environment.

 

Fig. 2: A visualization of the automated extraction of urban trees from lidar point clouds.

 

From an i-Tree Eco analysis on the ca. 250,000 registered trees in Amsterdam, it is estimated that every year 77 metric tons of pollutants are removed from the air. The pollutants include ozone, nitrogen dioxide, sulfur dioxide, and particulate matter of less than 2.5 microns. The nitrogen and sulfur dioxide removals are equivalent to the annual nitrogen and sulfur dioxide emissions from 3,700 and 17,800 cars, respectively. The pollutant removal in Amsterdam translates into an annual value of € 6.5 million.

Besides having a positive effect on air quality, trees also capture atmospheric carbon, which helps to mitigate climate change through a reduction of global warming. From the i-Tree analysis it follows that in Amsterdam an estimated 3,900 metric tons of carbon are sequestered yearly, i.e. stored in tree tissue through growth. This amount is equivalent to the annual carbon emissions of 3,000 cars. The annual value of carbon sequestration amounts to € 630,000.

Furthermore, vegetation can reduce the portion of precipitation that does not infiltrate into the soil. This portion is referred to as surface runoff and can be especially problematic in urban areas with a large number of impervious surfaces where it can lead to pollution of water bodies. The trees of Amsterdam help to reduce surface runoff by roughly 160,000 cubic meters per year, which translates to an annual value of roughly € 300,000. In general, the value of urban trees can be increased through proper tree management.

 

Fig. 3: A map of trees for a part of Amsterdam. Each tree, represented by a green point, carries attributes regarding pollution removal, carbon sequestration, avoided surface runoff, etc. This visualization was created with CARTO.

 

Naturally, this type of analysis is not limited to Amsterdam, but can be done for any urban area in the world where tree data is available. With improved historical air quality data and better knowledge on the distribution of urban green, one is able to discover patterns in air quality that would otherwise be undiscoverable. Sobolt strives to assist urban planners to make better decisions for tree management and air quality policy.

Are you interested to learn more? Visit www.sobolt.com/airquality | www.treetracker.ai | www.bomenwacht.nl