| 03 Nov 2025
Exposure of Settlements to Wildfires in a Transboundary Wildland-Urban-Interface Region in Central Europe
Abstract. Climate change has been causing a noticeable rise in disastrous wildfires in Mediterranean and temperate forests. Although forest fires were of smaller size and less concern in Central Europe than for example in the Mediterranean, the region has recently experienced large wildfires. For example, a wildfire burned significant parts of two National Parks at the border between Germany and the Czech Republic in 2022. This event demonstrated the need for local stakeholders and the scientific community to adapt fire risk assessment and management to this new reality. Here we aim to create a wildfire exposure map for the trans-boundary national park region Saxon-Bohemian Switzerland between Germany and the Czech Republic. We use several Earth observation products, historical fire occurrence points, medium to very high fire danger weather and wind scenarios, and three different fire duration scenarios (1, 2, 3 days) to simulate fire behaviour and burn probability, and to finally assess the potential exposure of settlements to wildfires. Observations from the wildfire in 2022 were used to validate the modelling framework. The interactive wildfire exposure map was tested with the general audience and local stakeholders for its usability and usefulness. We generally found higher burn probability and potential flame length at settlements in Czech Republic. As our results and the experiences from the past fire have shown wildfires cross borders, demonstrating the need for coordinated trans-border wildfire management.
Competing interests: The authors declare that they have no conflict of interest.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
This is an interesting manuscript that describes the application and dissemination of a wildfire risk and exposure assessment in the border region between Germany and the Czech Republic. This region, like many others in Northern and Central Europe, has not been typically considered as a hub of wildfire activity (unlike Southern Europe). Consequently, both the public and government agencies lack the awareness and preparedness to address the possibility of fire becoming a more common natural hazard in the near future due to climate change. A large wildfire in 2022 which had devastating effects on both human property and local forest ecosystems has motivated the authors to study fire risk and exposure in the region. To that end, they utilized a common modelling approach based on a widely-used fire spread model, FlamMap, to create maps of burn probability and flame length across the entire region assuming three weather scenarios and three fire durations. They then intersected fire hazard maps with data on property and infrastructure to estimate potential wildfire for all settlements in the region. Finally, they created a web-based interactive map that allows stakeholders to learn about fire risk in the region; and presented this tool to a group of stakeholders, soliciting their inputs on the map.
Major comments
The main strength of this manuscript is its holistic approach to addressing a growing wildfire problem in a new region. in contrast to most studies, this study did not end with the completion of the fire risk map, but continued to dissemination and accepting feedback from relevant stakeholders. This provides valuable lessons to researchers in other regions where fire problems may arise in the future, as they seek to educate stakeholders on an emerging threat. This strength of the manuscript compensates for the fact that its scientific novelty is rather low. The methodology follows a standard approach for mapping wildfire hazard and exposure that's been around for more than a decade, so it's only novelty is the generation of a fire risk map for a region that didn't have one before. The development of the web-based map is important for dissemination but it's mostly a technical contribution. And the focus group approach to study the relevancy of the results is not novel in the context of fire risk analysis, though the insights gleaned form it may be used to improve publicly available fire risk maps in the future.
I have no qualms with the risk mapping approach, as it was based on mostly standard procedures, though these had to be adapted to the particularities of the specific study region (i.e., generation of new fuel models and selection of suitable weather data). I think that the authors' approach is reasonable given data limitations, and consequently the predictions of the models under different scenarios make sense.
In contrast, the choice of methodology for the purpose of validation was confusing to me. If the actual perimeter of the 2022 fire is known, why was the ground truth based on remote sensing data? Is it because the daily progression of the fire is unknown? To what extent can we trust remote sensing data as "ground truth" for validation purposes, given that RS fire products suffer from significant commission and omission errors? Also, I get why the validation was done on a day-by-day basis (to account for the static wind conditions in FlamMap), but constricting the prediction of fire spread to a single day (starting the actual fire perimeter at the day's start) can inflate the model's accuracy. If FlamMap cannot handle dynamic weather, perhaps we shouldn't expect it to reliably reconstruct actual fires but use it only for its intended purpose: revealing areas at risk due to existing fuel composition and configuration and prevailing fire weather conditions (typically assuming extreme weather, because these conditions results in the worst fire outcomes).
Another issue in the validation method is the choice of accuracy metric. The % of fire detections within the modelled fire perimeter would always be 100% if FlamMap predicts that the entire landscape burns. But this will obviously reflect a massive commission error, which cannot be accounted by the metric used here. Instead, both omission and commission errors should be reported, ideally based on actual fire perimeters (see my previous comment) rather than on remotely sensed data.
While the manuscript's presentation is really thorough, its flow can be much improved by trimming repetitive information and excessive data about some of the methods which delve too deeply in technical descriptions of well-known tools such as FlamMap and background on fuel models. Importantly, the results contain much superfluous information as each sub section tends to begin with description of methods (e.g., lines 352-353, lines 404-429, 440-445, 462-470) and/or a detailed description of the figures that should be left to the caption (e.g., lines 353-354, lines 323-326). It will be much easier to follow if results are simply described, following with a pointer to the relevant figure. Methods should not be described in the results section. Moreover, the results section contains some discussion materials which are out of place (e.g., lines 361-369). In general, a clearer distinction between the methods, results and discussion sections is needed.
Specific comments
Line 16: " ... to create an interactive wildfire exposure map"
Line 24: since the map was tested with audiences, there should be a sentence that mentions the results of this test at the end of the abstract.
Line 29: a reference is needed to support the claim that wildfires are increasing in the WUI. For example: Radeloff, V.C., Helmers, D.P., Kramer, H.A., Mockrin, M.H., Alexandre, P.M., Bar-Massada, A., Butsic, V., Hawbaker, T.J., Martinuzzi, S., Syphard, A.D., Stewart, S.I. (2018) Rapid growth of the U.S. Wildland Urban Interface raises wildfire risk. Proceedings of the National Academy of Sciences 115:3314-3319.
Lines 36-37: How many buildings were destroyed in these settlements? This is important for context.
Line 76: "we aim to provide a quantification of the" is cumbersome. Perhaps replace with: "we aimed to quantify wildfire exposure for different settlements in the region"
Lines 79-81: This text is unnecessary and may be deleted.
Lines 89-91: These two sentences can be combined and written more concisely, e.g., "The study area also includes several settlements that are not part of the conservation area of saxon switzerland but are surrounded by it (Figure 1)."
Line 94: Good, but figure 1 shows many historical fire ignitions. Please describe the fire regime of the study area, i.e., how many fires per year, fire size characteristics, main ignition sources.
Line 99: If the ignition cause of the 2022 fire is known, it should be stated.
Figure 1: The caption should mention the range of years for these ignitions.
Line 125: Did the model account for the potential of spotting? In principle, FlamMap can do so but was it implemented in this study? This is important because spotting is a major driver of fire spread under extreme weather conditions.
Lines 142-144: Random ignitions are not the standard in informed fire simulations, so there's no need to write why they weren't used. In the real world, fire ignitions are never randomly distributed. It's sufficient to start by stating that historical fire ignitions were used, and these had a clear spatial pattern in the study area.
Line 144: How many years does 'historical' reflect?
Lines 152-155: Unclear. Are you balancing the total number of ignitions or ignition density (N/Area)? The former is only correct if both parts have the same area.
Section 2.3.3: This is unrelated to the modeling stage and should be moved to the beginning of the exposure assessment section.
Line 272: How many fires were modelled overall to facilitate the calculation of burn probability per simulation? Is this the number of ignitions? In general, are all ignitions occurring in any single model run or each model run is using one ignition point? Please clarify.
Lines 343-344: Is the lack of fires crossing the river due to a missing spotting component in the model, or an outcome of the width of the river combined with prevailing wind directions?
Figure 3: it would be clearer if WBI numbers were replaced by names, i.e., Medium, High, and Very High.
Lines 352-353: See my major comment above. This sentence belongs in the methods and not here (and if it repeats info from the methods, it can simply be deleted).
Lines 353-355: There's no need to describe the contents of a figure anywhere outside its caption.
Lines 361-369: This paragraph belongs in the discussion section.
Figure 5. Perhaps mention in the main text the large variation around the median BP in the case of Sebnitz. Also, in the caption, mention the number of simulations from which the median and range were calculated.
Lines 404-429: This is all methods
Lines 440-445: This is all methods
Lines 462-470: This is all methods
Technical corrections
Lines 71-81: This paragraph should be in past tense.
Line 91: "Mio" is Million?
Lines 170-171: Append to the end of the previous paragraph (a paragraph cannot contain a single sentence).
Line 271: "ran FlamMap"
Line 338: hazard, not exposure. The outputs of FlamMap do not consider values at risk which are required to map exposure.
Figure 3 caption: should be "wildfire hazard rasters"
Figure 4 caption: The caption is in the wrong order. It should start by explaining the general purpose of the figure, e.g., "Hazard values (BP, FL) for different municipalities based on their residential home ignition zone under the extreme weather scenario. The five municipalities with the highest BP and FL are highlighted in red."
Line 491: replace exposure with hazard.