Vegetation monitor

Rijkswaterstaat

It involves detection of vegetation types along the floodplains of major rivers. The system uses an AI image detection algorithm based on satellite images. By regularly detecting vegetation types in recent satellite images, a comparison can be made with the legally established standard in the vegetation layer. In doing so, the AI system operates in a

Last change on 24th of December 2025, at 9:29 (CET) | Publication Standard 1.0

• Publication category: Impactful algorithms
• Impact assessment: AIIA
• Status: In use

General information

Theme

Space and Infrastructure

Begin date

06-2025

Contact information

[e-mail]

Link to publication website

www.rijkswaterstaat.nl

Responsible use

Goal and impact

The aim is to determine five different vegetation types in surfaces along the floodplains of major rivers. This is done based on the most recently available high-resolution satellite data.

Considerations

The end users are members of Rijkwaterstaat Oost-Nederland's Uiterwaarden team. They can perform their monitoring tasks more efficiently, thanks to improved access to information. Using this algorithm has two main advantages. First, this approach saves a lot of manual work in the field. Thus, tracking vegetation along floodplains is more efficient. Second, this approach increases the precision of monitoring.

A disadvantage is that satellite images may include recordings of private land. Monitoring is limited to only the area of the vegetation layer. Thus, intrusion on private individuals is minimised.

If vegetation is found to be too densely vegetated, it may pose a flood hazard. The algorithm helps detect this. Therefore, it is important to apply this monitoring.

Human intervention

The AI system has a supporting role. The users are members of team Uiterwaarden. The algorithm helps them determine whether vegetation is too dense. In doing so, the algorithm makes suggestions for further investigation. In this case, a visual survey will also be carried out by the area manager.

Risk management

Incoming source data is monitored (in realisation phase). Due to the resolution, it is not possible to recognise persons on the satellite images. There are no restricted areas (e.g. military) within the considered area. Thus, no personal or confidential data are present in the source data. Our processing of the source data also does not provide any confidential information.

The algorithm can only be used (and not modified) by designated users within Rijkswaterstaat. Authorisation authentication is set up for this purpose.

The AI system is easy to explain based on the outcomes. These show how well the model detects the different vegetation types.

Legal basis

The legal basis follows from the task of RWS as the party responsible for managing this part of the floodplain.

Links to legal bases

Decree and Explanatory Memorandum Public Works Layout, Rivers and Vegetation Layout Section, Update 2025: https://open.rijkswaterstaat.nl/publish/pages/206624/besluit_en_toelichting_vegetatielegger_2024-1.pdf

Impact assessment

AI Impact Assessment (AIIA)

Operations

Data

Satellite data over several seasons. The algorithm uses only the area of the floodplains of major rivers. This is recorded in the vegetation layer.

Technical design

The algorithm applies AI image detection to satellite images. With this, it detects vegetation types along the floodplains of major rivers. The result is compared with the legally established standard.