0619059 - ÚVGZ 2026 RIV US eng J - Journal Article
Kudláčková, Lucie - Linda, R. - Balek, Jan - Štěpánek, Petr - Zahradníček, Pavel - Poděbradská, Markéta - Možný, Martin - Hlavsová, Monika - Žalud, Zdeněk - Trnka, Miroslav
Reliability of fire danger forecasts for Czech agricultural and forestry landscapes.
FIRE ECOLOGY. Roč. 21, č. 1 (2025), č. článku 20. ISSN 1933-9747
Institutional support: RVO:86652079
Keywords : weather prediction models * climate-change * index * Czech Republic * Fire danger * ffdi * Fire occurrence * fwi * Number of wildfires * Prediction * Weather forecast
OECD category: Environmental sciences (social aspects to be 5.7)
Impact factor: 3.6, year: 2023 ; AIS: 1.183, rok: 2023
Method of publishing: Open access
Result website:
https://fireecology.springeropen.com/articles/10.1186/s42408-025-00362-7DOI: https://doi.org/10.1186/s42408-025-00362-7

BackgroundThe increasing threat of fire caused by ongoing climate change requires accurate and timely prediction for the effective management of extreme fire situations. The limited research on the connection between fire danger metrics and the occurrence of wildfires in the forested and agricultural landscapes of the Czech Republic underscores the need to better understand how to properly quantify fire danger in the context of Central Europe. This study focused on assessing the accuracy of fire danger prediction with respect to the number of wildfires in different geographic regions of the Czech Republic and provided new insights into central European fire ecology.ResultsWe found that the fire season in the Czech Republic has two peaks, in spring and summer, with regional differences in the total number of wildfires. Analyses of fire danger via the Canadian Fire Weather Index (FWI) and Australian Forest Fire Danger Index (FFDI) for the years 2018-2022 revealed that the IFS numerical weather prediction model is the most suitable for conditions in the Czech Republic. A linear regression model showed a high predictive capability for the total number of wildfires in the Czech Republic, with an observed R-squared value of 0.81 and a mean absolute error (MAE) of 5.19 wildfires with a 95% confidence interval (CI) of 4.94-5.44. Additionally, the second model, which utilized a linear model with random effects to account for regional variability, had an R-squared value of 0.34 and an MAE of 1 wildfire (95% CI +/- 3), indicating that the inclusion of regional correction coefficients (random effects) enhanced the prediction accuracy.ConclusionsThis study provides key insights into fire danger prediction in relation to the number of wildfires. With this model, it is possible to predict how many wildfires may occur at specific values of the FWI and FFDI in individual regions (NUTS 3) of the Czech Republic. This information can be used for more effective readiness planning for human resources and fire equipment while also contributing to the enhancement of general knowledge in the field of fire science in the context of central Europe.
Permanent Link: https://hdl.handle.net/11104/0365832
Postprint:

Scientific data in ASEP :
Data pro článek. Reliability of fire danger forecasts for Czech agricultural and forestry landscapes