Natural fire model

Safety region Noord- en Oost-Gelderland

The wildfire model is a prediction model that gives an indication of how high the risk is that a wildfire will occur in the safety region of North and East Gelderland. It is used by crisis officers to prepare for the period ahead.

Last change on 24th of June 2025, at 7:13 (CET) | Publication Standard 1.0

• Publication category: Other algorithms
• Impact assessment: ...
• Status: In use

General information

Theme

Nature and Environment

Begin date

Field not filled in.

Contact information

[e-mail]

Responsible use

Goal and impact

The purpose of the wildfire model is to provide an estimate of the wildfire risk for the coming week. The outcomes of the wildfire model are visualised internally within VNOG in a Power BI dashboard and a Power apps canvas app. These products are created within the Safety Information Exchange (VIK). These products are primarily intended for crisis officers to prepare for possible risks for the coming period. The Power apps canvas app integrates automated notifications, notifying users when an increased risk is expected. Any follow-up actions are decided by crisis officers.

Considerations

The model was developed to provide an indication of the wildfire risk for the coming week, and to do so in a data-driven way. This informs the user in a neutral and objective way or so that it becomes clear whether additional in-depth analysis is needed.

Human intervention

The outcomes of the model are used by people who can use it as a gauge of whether they should delve further into the subject. The model itself does not make any operational decisions and is purely to support users. The users themselves are responsible for any follow-up actions.

Risk management

The main risk is that the model may deviate from the actual risk and fail to inform users. At the point when users become completely dependent on the model and start reading in only when the model pans out, this can create potential risks. Use of the model is therefore accompanied by an explanation of the model's interpretation and additional user responsibilities.

Legal basis

The Safety Regions Act (WVR), in particular Article 10 (tasks and powers of the safety region) and Articles 45-50 (information and communication), elaborated in the

Disaster and Crisis Information Decree .

Links to legal bases

Wet veiligheidsregio’s: https://wetten.overheid.nl/BWBR0027466/2025-02-12

Elaboration on impact assessments

At present, no formal risk classification has been completed.

Impact assessment

Risicoclassificatie

Operations

Data

To develop the model, various historical data from KNMI weather stations were retrieved (precipitation, humidity, temperature, wind speed). This data is enriched with incident data from GMS on wildfires and roadside fires. This data was collected from the years 2017 to 2023.

Technical design

Logistic regression was performed based on the weather data and incident data. Before training the model, the data was first divided into a test and training set and all variables were scaled to the same level. Then the train set was used to train the model and the test set use to validate the model.

The model has weather data as input and outputs a value between 0 and 1 that indicates the risk of wildfires, with 1 being a very high risk and 0 being a very low risk.

During the pilot of this project, the first version of the model ran for a year. After a year, an evaluation of the model was done to evaluate how well the model worked.

External provider

Safety region Noord- en Oost-Gelderland