NH7.1 | Spatial and temporal patterns of wildfires: models, theory, and reality
EDI
Spatial and temporal patterns of wildfires: models, theory, and reality
Convener: Joana Parente | Co-conveners: Andrea TrucchiaECSECS, Marj Tonini, Andrey Krasovskiy, Francesca Di Giuseppe, Marco Turco, Shelby CorningECSECS
Orals
| Tue, 16 Apr, 14:00–15:45 (CEST), 16:15–18:00 (CEST)
 
Room 1.14, Wed, 17 Apr, 08:30–12:30 (CEST)
 
Room 1.14
Posters on site
| Attendance Wed, 17 Apr, 16:15–18:00 (CEST) | Display Wed, 17 Apr, 14:00–18:00
 
Hall X4
Orals |
Tue, 14:00
Wed, 16:15
Wildfires represent a hazardous and harmful phenomenon to people and the environment, especially in populated areas where the primary cause of ignition is related to human activities. This has motivated scientists to develop spatial-temporal datasets and to produce risk and prognostic maps for governments and managers. A key tool in this respect is to assess the fire spatial-temporal distribution and to understand their relationships with the surrounding environmental, climatological and socio-economic factors.
Innovative algorithms and methodologies developed in the computational science field have proved to be useful in analysing spatially and temporally distributed natural hazards and ongoing phenomena such as wildfires. Moreover, considering the fast-growing availability of high digital geo-referenced data, it is important to promote methods and new tools for their study and modelling, in a wide-range of scales. A new exciting challenge is to convert available datasets into meaningful and valuable information and make this information interesting to stakeholders.
This session aims to bring together fire scientists, researchers of various geo-environmental disciplines, economists, managers, people responsible for territorial and urban planning, and policymakers. The main goal is to improve the understanding of the fire regime and to discuss new strategies to mitigate the disastrous effects of wildfires. We welcome empirical studies, new and innovative technologies, theories, models, and strategies for fire research, seeking especially to identify and characterize the spatial-temporal patterns of wildfires.
Research topics include, but are not limited, to the following:
• development of methodologies based on expert knowledge and data-driven approaches, for the recognition, modelling and prediction of structured patterns in wildfires;
• pre- and post-fire assessment: fire incidence mapping and spatial distribution; fire severity and damages; fire risk management;
• long-term wildfires patterns and trends: relation between wildfires and global changes such as climate, socioeconomic and land use/ land cover changes;
• fire spread models and fire-weather relationships, ranging from case studies to long-term climatological assessments;
• post-fire vegetation recovery and phenology.

This year's session is divided into three different themes/blocks, followed by a short debate between the audience and the presenters.

Joining the amazing presenters, we will have 1 solicited presenter: Carlos C. DaCamara.

Orals: Tue, 16 Apr | Room 1.14

The oral presentations are given in a hybrid format supported by a Zoom meeting featuring on-site and virtual presentations. The button to access the Zoom meeting appears just before the time block starts.
Chairpersons: Marj Tonini, Marco Turco
14:00–14:05
Fire, drivers & recovery (Marj Tonini & Marco Turco)
14:05–14:15
|
EGU24-10974
|
NH7.1
|
On-site presentation
Célia M. Gouveia, Alexandre M. Ramos, Mafalda C. Silva, Rita Durão, A. Serkan Bayar, and Joaquim G. Pinto

The frequency and intensity of temperature extremes have increased worldwide in the past couple of decades. The year 2018 was characterized by record-breaking temperatures in many parts of Europe during spring and summer, which lead to unusual and severe wildfires in central and northern Europe. For example, devastating fires destroyed large areas of intact forests, not only in countries with a long tradition of wildfires but also in countries, such as Sweden, Norway, Finland, and Latvia.  In 2022, Europe was characterized   by a prolonged spring drought, several summer heatwaves and fire activity without precedents for several European countries. In particular, a high number of fires occurred in Germany, Austria, Chechia, Hungary, Slovenia and Romania during the summer months, highlighting the increase of fire-prone conditions in the region linked with an increase of hot and dry conditions.

This work analyses the exceptionality of the 2022 fire season over central Europe. Fire Radiate Power (FRP) from MODIS, burned area and number of fires from EFFIS were used to characterize fire occurrence in 2022. We used ERA5 meteorological parameters, such as the maximum and minimum air 2m temperature, minimum relative humidity, and wind speed to evaluate the severity of the heat extremes from April to August. SPEI for the time scales of 6 and 12 months were computed using ERA5 data to evaluate the drought conditions in spring and summer over the central European region. The Canadian Fire Weather Index (FWI) and sub-indices available from ERA5 data, were used to assess the exceptionality of meteorological fire danger over the region in the summer of 2022.  Moreover, possible FWI trends and sub-indices were also analysed for the period from 1979 to 2022. The impact of drought on vegetation productivity during the spring and summer of 2022 was also evaluated. Results highlight the new fire dynamics in Europe in recent years, with new emergent hot spots, in central and northern European countries. It is thus extremely important to assess of trends of fire danger and changes in fire activity over this region to better define the related activities of fire monitoring, as well as the definition of planning activities and suppression measures towards climate mitigation and adaptation.

Acknowledgements: This study is partially supported by the European Union’s Horizon 2020 research project FirEUrisk (Grant Agreement no. 101003890) and by the Portuguese Fundação para a Ciência e a Tecnologia (FCT) I.P./MCTES through national funds (PIDDAC) – UIDB/50019/2020- IDL and DHEFEUS - 2022.09185.PTDC.

How to cite: Gouveia, C. M., Ramos, A. M., Silva, M. C., Durão, R., Bayar, A. S., and Pinto, J. G.: The exceptionality of the 2022 fire season over Central Europe, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10974, https://doi.org/10.5194/egusphere-egu24-10974, 2024.

14:15–14:25
|
EGU24-21377
|
NH7.1
|
On-site presentation
Hanieh Dadkhah, Divyeshkumar Rana, Ebrahim Ghaderpour, and Paolo Mazzanti

Wildfires present substantial threats to ecosystems and human settlements which increase the importance of monitoring for timely detection and assessment. This study was performed on the Campania provinces—Salerno, Avellino, Benevento, Caserta, and Napoli in Italy—employing a multi-sensor remote sensing approach to elevate wildfire analysis. The first objective is identifying fire patches through Moderate Resolution Imaging Spectroradiometer (MODIS) data from 2001 to 2020. Although, generating Difference Normalized Burn Ratio (DNBR) and Difference Normalized Difference Vegetation Index (DNDVI) maps from Sentinel-2 images. Integration of MODIS and Sentinel-2 outcomes enhances pinpointing fire-affected areas. Subsequently, Incorporating Landsat 9 images for Land Surface Temperature (LST) and Climate Hazards Group InfraRed Precipitation with Station data (CHIRPS) for precipitation trends in five provinces from 2001 to 2020 in Hotspot Polygons generated via First Order Contiguity edges corners algorithm. Pearson correlation coefficients between burnt area, LST, and precipitation are computed. A high correlation coefficient was observed between the mentioned parameters. Wildfire analysis reveals peak burnt areas in 2001, 2007, and 2017 in Avellino and Salerno Province. Change detection maps illustrate significant land cover changes from Forest to Savannas and Shrubland to grasslands in 2001. Avellino province reveals a decreasing trend in Grassland and an increase in Savannas, as the same as observations in Salerno Province. This study considers the analysis of wildfires, connecting burnt areas, climate variables, and land cover changes across the Campania provinces.

Keywords: Wildfire, Fire and Vegetation Indices, Land Cover Changes, Climate Change

How to cite: Dadkhah, H., Rana, D., Ghaderpour, E., and Mazzanti, P.: Integrating Multi-Sensor Remote Sensing Data for Comprehensive spatio-temporal WildfireAssessment In Campania Provinces – Italy, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-21377, https://doi.org/10.5194/egusphere-egu24-21377, 2024.

14:25–14:35
|
EGU24-6004
|
NH7.1
|
Virtual presentation
Carolina Purificação, Cátia Campos, Alice Henkes, Stergios Kartsios, and Flavio T. Couto

The study is a step forward in the characterization of meteorological environments that favour the evolution of large and extreme fires in Southern Portugal. The region has some fire-prone areas which are recognized by the mega fires occurred in 2003, 2005, and 2018. Two numerical simulations were performed using the Meso-NH non-hydrostatic research model and used to investigate in detail the atmospheric environments of two large fires that occurred on 18th July 2012 and 19th June 2020. The simulations were configured using two nested domains with a 375 km × 375 km grid domain (D1) at 2.5 km horizontal resolution and a 150 km × 150 km domain (D2) at 500 m resolution added before the start of the fires. The vertical grid was configured with 50 stretched levels following the terrain. The initial and boundary conditions are provided by the 6-hourly operational ECMWF analyses. The large-scale circulation has been characterised using data obtained from the ECMWF's Meteorological Archival and Retrieval System. In addition to the large-scale circulation, namely the positioning of the Azores anticyclone and the thermal low development over the Iberian Peninsula, the results have shown the important role played by regional orography in creating favourable fire weather conditions. For instance, the high-resolution simulations showed the high daytime temperatures and sometimes overnight, low humidity, and strong wind gusts that favour fire spread. In July 2012, the typical sea breeze circulation affected the fire evolution, whereas the intense downslope winds favoured the fire spread in June 2020. The study brings useful guidelines for interpreting the impact of different mesoscale environments that may produce large fires, namely the orographic effects that can increase the fire susceptibility and vulnerability of some regions. This study was funded by national funds through FCT-Foundation for Science and Technology, I.P. under the PyroC.pt project (Ref. PCIF/MPG/0175/2019).

How to cite: Purificação, C., Campos, C., Henkes, A., Kartsios, S., and Couto, F. T.: Meteorological environments leading to two large fires in Southern Portugal, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6004, https://doi.org/10.5194/egusphere-egu24-6004, 2024.

14:35–14:45
|
EGU24-11570
|
NH7.1
|
ECS
|
On-site presentation
Kerryn Little, Dante Castellanos-Acuna, Piyush Jain, Laura Graham, Nicholas Kettridge, and Mike Flannigan

Persistent positive anomalies in 500 hPa geopotential height (PPAs) are upper-air circulation patterns associated with surface heatwaves, drought, and consequently fuel aridity, elevated fire weather, and active wildfires. We examined the association between PPA events and surface fire weather and burned area at a pan-European level. Europe-wide, extreme fire weather and wildfires were on average 3.5 and 2.3 times more likely to occur concurrently with a PPA, respectively. PPAs were associated with 43% of pan-European area burned between March and October 2001–2021, and there was a latitudinal increase in the percentage of area burned during PPAs up to 60% over Northern Europe. Burned area was highest in the three days following PPA presence, and fuel moisture indices from the Canadian Fire Weather Index System lagged behind peak PPA strength, demonstrating the role of PPAs in pre-drying fuels. PPAs have been associated with significant wildfire events experienced across Europe, including the 2017 Portugal wildfires, the 2018 UK, Sweden, and Finland wildfires, and the 2021 Greece wildfires. Our findings demonstrate opportunities for developing early warning systems of wildfire danger, having implications for wildfire awareness and preparedness, informing policy, and wildfire management decisions like early mobilisation and resource sharing initiatives within and across Europe.

How to cite: Little, K., Castellanos-Acuna, D., Jain, P., Graham, L., Kettridge, N., and Flannigan, M.: Persistent positive anomalies in geopotential height drive enhanced wildfire activity across Europe, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-11570, https://doi.org/10.5194/egusphere-egu24-11570, 2024.

14:45–14:55
|
EGU24-8971
|
NH7.1
|
ECS
|
Highlight
|
On-site presentation
Outi Kinnunen, Leif Backman, Juha Aalto, and Tiina Markkanen

Climate change alters boreal forest dynamics. The risk of boreal forest disturbances are expected to increase by the end of the century compared to the current state. However, projecting the future impacts of climate change on forest disturbances inherently contains uncertainties related to the global climate models.

Here, we study the impact of climate change on forest fires and wind damage using ecosystem model (JSBACH) simulations from 1951 to 2100. The simulations are driven by output from three global climate driver models that have been bias-corrected and downscaled (CORDEX EUR-44 domain). The global models from CMIP5 were run under two forcing scenarios, RCP 4.5 and RCP 8.5. To tackle the uncertainty of climate change projections, we use six climate projections.

In our simulations the fire season in Fennosscandia is projected to extend in both spring and autumn. The fire season is estimated to lengthen by 20-52 days, starting 10-23 days earlier and ending 10-30 days later, by the end of the century. In general, it is expected that the number of fires and burnt area are projected to increase from the reference period (1981-2010) to the end of the century (2071-2100) due to rising temperatures, despite increases in precipitation. However, the amount and direction of change varies significantly between climate projections and locations.

Our preliminary results implicate that the risk for wind damage may change and affect to the number of fires. Wind damage affects the size of litter pools that change the amount of fuel available for fires. Finally, our aim is to study the interaction between forest fires and wind damage in boreal forests.

How to cite: Kinnunen, O., Backman, L., Aalto, J., and Markkanen, T.: Projected impacts of climate change in forest fires and wind damage in Fennoscandia, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8971, https://doi.org/10.5194/egusphere-egu24-8971, 2024.

14:55–15:05
|
EGU24-10990
|
NH7.1
|
Highlight
|
On-site presentation
Luiz Galizia, Christelle Castet, and Apostolos Voulgarakis

Wildfires are expected to increase under warmer and drier conditions, yet little is known about their potential effects on population at global scale. Here, we developed a novel framework based on statistical wildfire model and spatial demographic data to better understand how global warming alters population exposure to wildfires throughout the world. We sought to model annual burn rate with relevant explanatory variables, such as climate, land cover, and topography to simulate historical and future wildfire frequency at global scale. To do so, we used a Generalized Additive Model combined with historical climate data and an ensemble of CMIP6 climate projections under the SSP5-8.5 scenario. We then analysed population exposure to wildfires combining population count across the wildland–urban interfaces with the simulated historical and future wildfire frequency. Our results indicate that in the present day the highest population exposure to wildfires is in southeast Asia, parts of South America and Africa, due to the large number of people living in wildland–urban interfaces with a high wildfire frequency. All other things being equal, global warming was found to increase population exposure, with an expansion of the regions with high wildfire frequency in east and south Europe, southeastern Asia, parts of North and South America. The estimated increase in population exposure may also imply potential impacts on the built environment and human health in the absence of mitigation or adaptation measures.

How to cite: Galizia, L., Castet, C., and Voulgarakis, A.: Global warming increases population exposure to wildfires , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10990, https://doi.org/10.5194/egusphere-egu24-10990, 2024.

15:05–15:15
|
EGU24-6108
|
NH7.1
|
Virtual presentation
Cátia Campos, Flavio T. Couto, Filippe L.M. Santos, João Rio, Teresa Ferreira, Carolina Purificação, and Rui Salgado

Portugal is one of the European countries that faces significant challenges with wildfires. While lightning-triggered natural fires constitute a minority compared to anthropogenic ones, accurate forecasting of lightning occurrences is crucial for effective prevention. The study assesses the ECMWF model's capability to predict lightning in Portugal over four fire seasons [2019-2022]. Observed lightning data was obtained from the national lightning detector network, aggregated into 0.5° and 1° resolutions over 3-hour periods. The evaluation employs statistical indices from a contingency table to analyze the model's performance. Results indicate an overestimation of lightning occurrences by the ECMWF model, with a Bias greater than 1. The success rate for lightning prediction was 57.7% for a horizontal resolution of 1° and 49% for 0.5°. Additionally, the temporal analysis reveals a time lag between both data, with the model starting to predict lighting before its occurrence and finishing the prediction earlier. These findings are complemented by analyzing the spatial lightning distribution, which led us to identify some weather patterns associated with lightning activity during the study period. For instance, lightning activity was associated with the Iberian thermal low development overlapped by an Upper Level Low and the passage of large-scale features, such as frontal systems. The insights gained from this study have implications for the ECMWF lightning forecast applicability in the context of forecasting natural forest fires in Portugal. The research was funded by the European Union through the CILIFO project (0753-CILIFO-5-E) and also by national funds through FCT Foundation for Science and Technology, I.P. under the PyroC.pt project (PCIF/MPG/0175/2019).

How to cite: Campos, C., Couto, F. T., Santos, F. L. M., Rio, J., Ferreira, T., Purificação, C., and Salgado, R.: Assessing ECMWF Lightning Forecast in Portugal during fire seasons, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6108, https://doi.org/10.5194/egusphere-egu24-6108, 2024.

15:15–15:25
|
EGU24-15304
|
NH7.1
|
ECS
|
Highlight
|
On-site presentation
Li Zhao and Marta Yebra

Lightning-induced wildfires lead to significant loss of life and extensive property damage worldwide. This issue is especially critical in southeast Australia, where such wildfires account for 80-90% of the total area burned, emphasising the need for a comprehensive understanding of the contributing factors and mechanisms that drive these events. This study aims to investigate the complex interactions between climate, topography, lightning activity, and fire events in New South Wales (NSW), Australia. By analysing comprehensive datasets from 2017-2021, including ignition records, meteorological data, topographical information, and fuel characteristics, this research seeks to identify the key factors influencing lightning-attributed wildfires and predict the probability of lightning-caused fire occurrence. A Random Forest model is trained and tested to estimate the probability of fires caused by lightning strikes. Model performance was assessed through the Receiver Operating Characteristic, with an Area Under the Curve (AUC) around 0.7 in the validation datasets, indicating a good agreement between the estimated probabilities and the reported lightning-caused fires. The identified key factors that influence lightning fire ignitions include humidity, elevation, temperature, rainfall, soil moisture, and fuel moisture, highlighting the dominant influence of weather variables on wildfire ignitions. The preliminary results demonstrate a potential link between the geographic distribution of lightning-induced fires and the temperate climate zones, possibly due to the presence of dense vegetation and seasonal weather patterns. Our ongoing efforts focus on further refining the predictive model and conducting a more extensive analysis of the data to enhance our understanding of the dynamics of lightning-induced wildfires. Ultimately the study will provide insights for effective risk management and mitigation of lightning-caused wildfires in the regions prone to wildfires.

How to cite: Zhao, L. and Yebra, M.: Modelling and Analysis of Lightning-Induced Wildfires in Australia , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15304, https://doi.org/10.5194/egusphere-egu24-15304, 2024.

15:25–15:35
|
EGU24-18048
|
NH7.1
|
ECS
|
On-site presentation
Johannes Laimighofer, Mariana Silva Andrade, Pia Echtler, Sven Fuchs, Mortimer Müller, Maria Papathoma-Köhle, Harald Vacik, and Herbert Formayer

Increasing temperatures, due to climate change lead to more evapotranspiration which increases the possibility of severe drought periods. These trends enhance the risk of wildfire hazards even in humid regions like the Alps. Further, possible changes in the occurrence of thunderstorms can modify the ignition danger of lightning induced wildfires. This study aims to investigate the effect of climate change on the probability of wildfires ignited by lightnings including possible shifts in lightning probability for Austria.

The full analysis is performed on a 1x1 km grid over Austria. Fire ignition danger and drought periods are approached by computing the Fine Fuel Moisture Code (FFMC). Noon temperature and windspeed for the FFMC are estimated by a spatio-temporal GAM (generalized additive model) with a geographic varying cyclic B-spline. The occurrence of lightnings is approached by the Showalter Index, which is validated with data from the Austrian Lightning Detection and Information System (ALDIS) for the period 2011 to 2020. For the historical weather conditions the Spartacus dataset is used for the period 1981-2022. Regarding the future development, five different climate projections are compared.

The historical period showed on average no trend for days with high FFMC values (> 91) for Austria, but already 13% of the study area have a significant positive trend (tested by Mann-Kendall trend test). The trend is even more evident for the climate projections, which show a significant increase in days with FFMC values > 91 for 99% of the study area, with a sharp increase starting about 2050. Possible alterations in thunderstorm activity will strengthen the danger of forest fire ignitions of wildfires in Austria and are posing an increasing threat for forest management and society.

 

How to cite: Laimighofer, J., Andrade, M. S., Echtler, P., Fuchs, S., Müller, M., Papathoma-Köhle, M., Vacik, H., and Formayer, H.: Effect of climate change on lightning induced forest fires in Austria, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-18048, https://doi.org/10.5194/egusphere-egu24-18048, 2024.

15:35–15:45
Coffee break
Chairpersons: Shelby Corning, Andrey Krasovskiy
16:15–16:16
16:16–16:26
|
EGU24-8266
|
NH7.1
|
Highlight
|
On-site presentation
Stefan Doerr, Tadas Nikonovas, Cristina Santin, Gareth Clay, Claire Belcher, and Nicholas Kettridge

Fire weather indices are used widely as predictors for landscape fire potential. However, for the United Kingdom (UK: England, N-Ireland, Scotland and Wales) and comparable regions of humid-Atlantic Europe, they do not correlate well with fire occurrence. Here we explore the role of vegetation phenology as a key driver for fire occurrence in the UK.

We mapped satellite-derived fire occurrence and phenology climatology for 2012-2023 onto main fire-affected vegetation cover types within distinct precipitation regions for the UK. This enabled fire occurrence for fuels in different phenological phases to be explored across distinct ‘fuel’ types and regions.

Semi-natural grassland and dwarf shrub-dominated land emerged as the prominent fire affected ecosystems across much of the UK. We found that, critically, fire occurrence for vegetation at its maximum greenness were reduced by a factor of five to six compared to dormant vegetation, despite higher fire weather indices being typically associated with the former.

In contrast to most regions of the world that exhibit more extreme fire weather, fire activity in the UK’s humid Atlantic climate therefore seems strongly governed by vegetation phenology. This suggests that incorporating vegetation phenology is la critical step in the development of robust fire risk and behaviour prediction systems for regions with similar climate. It should be noted, however, that we also found evidence of that this fire-suppressing phenology barrier can be broken during extreme summer heat/drought events, which are likely to increase in frequency and severity under changing climate.  Hence fire weather indices remain critical predictors during the currently still rare extreme summer heat/drought events.

How to cite: Doerr, S., Nikonovas, T., Santin, C., Clay, G., Belcher, C., and Kettridge, N.: Spatio-temporal patters of fire in humid-Atlantic Europe: is vegetation phenology rather than fire weather the key driver? , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8266, https://doi.org/10.5194/egusphere-egu24-8266, 2024.

16:26–16:36
|
EGU24-11983
|
NH7.1
|
ECS
|
On-site presentation
Mariana Silva Andrade, Mortimer M. Müller, Katrin Kuhnen, and Harald Vacik

The altering climate patterns contribute to variations in precipitation, temperature and overall ecosystem conditions, influencing the composition and combustibility of forest fuels in the alpine regions. Changes in vegetation patterns, with shifts in species distribution and the prevalence of dry, flammable materials, increases the risk through wildfires. Rising temperatures and prolonged periods of drought enhance the likelihood of ignition and intensify fire behavior. Thus, this research aims to carry out a comprehensive vegetation analysis to characterize the different fuel types in Austria and to provide the scientific basis for developing a detailed forest fuel map, considering various types of vegetation, topography and land cover. We use statistical models to predict fuel characteristics based on vegetation type and empirical data collected during field surveys in the recent years. The spatial distribution of fuel types will be related to an analysis of the location of historical fire data for the time period 2001-2023. A statistical analysis is done to identify clusters, patterns, and relationships for different fuel types. This integrated methodology not only enriches our understanding of the complex interconnection between vegetation fire ignition and behavior, but also provides the scientific basis for developing targeted strategies in forest fire management and improving prevention measures in Austria.

How to cite: Silva Andrade, M., M. Müller, M., Kuhnen, K., and Vacik, H.: Relation between the distribution of fuel types and forest fires in Austria, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-11983, https://doi.org/10.5194/egusphere-egu24-11983, 2024.

16:36–16:46
|
EGU24-5375
|
NH7.1
|
ECS
|
Highlight
|
On-site presentation
Tiago Ermitão, Célia Gouveia, Ana Bastos, and Ana Russo

Over the past two decades wildfires have been increasingly disturbing many ecosystems worldwide. Among them, the Mediterranean-like climate regions have been strongly affected by recurrent events, as widely seen during the fire seasons of 2003, 2005, 2017 and 2022 in Portugal and northern Spain, as well as in Greece and southern Italy in 2007, 2021 and 2023. Additionally, Chile experienced significant fire seasons in 2015 and 2017, California faced destructive wildfires in 2018, 2020 and 2021, and Australia was affected by severe wildfires during 2019-2020.

Even though there is an observed increase of fire frequency over fire-prone regions, the Mediterranean ecosystems are in general well adapted to fire through several mechanisms to tolerate exposure to extreme conditions and recover from fire. However, climate change has been exacerbating the frequency and severity of climate extreme events, so that the pace of recovery of ecosystems from fires may be impaired, enhancing the potential of irreversible changes in vegetation communities. 

Here we assess the recovery of global Mediterranean vegetation after recurrent fires over the past two decades based on Enhanced Vegetation Index (EVI) retrieved from the MODIS sensor. To do so, we apply a statistical model to assess the recovery rate of vegetation repeatedly burned across different land cover types. Moreover, we study how fire severity, pre-fire state of vegetation and post-fire climate conditions modulate the recovery rates. Our results show a significant influence of fire severity on vegetation recovery rates globally across all Mediterranean regions, suggesting that higher severity levels may trigger the activation of the ecosystem's recovery mechanisms. Nevertheless, we also find a modulating effect of post-fire climate conditions, particularly air temperature and precipitation, on the recovery rates of burned vegetation, which highlights how compounding effects of changing disturbance regimes and climate change might destabilize ecosystems.

This study was supported by the doctoral Grant PRT/BD/154296/2022 financed by FCT under the MIT Portugal Program and was performed under the framework of DHEFEUS project, funded by Portuguese Fundação para a Ciência e a Tecnologia (FCT) (https://doi.org/10.54499/2022.09185.PTDC). The work was also funded by the FCT I.P./MCTES through national funds (PIDDAC) UIDB/50019/2020 (https://doi.org/10.54499/UIDB/50019/2020), UIDP/50019/2020 (https://doi.org/10.54499/UIDP/50019/2020) and LA/P/0068/2020 (https://doi.org/10.54499/LA/P/0068/2020). A.R. is supported by the FCT through national funds from the MCTES within the Faculty of Sciences of University of Lisbon, through https://doi.org/10.54499/2022.01167.CEECIND/CP1722/CT0006.

How to cite: Ermitão, T., Gouveia, C., Bastos, A., and Russo, A.: Global patterns of Mediterranean ecosystems recovery from recurrent fires, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5375, https://doi.org/10.5194/egusphere-egu24-5375, 2024.

16:46–16:56
|
EGU24-20928
|
NH7.1
|
On-site presentation
Avery Walters, Nawa Raj Pradhan, Ian Floyd, and Venkataraman Lakshmi

The 2007 Zaca Fire burned about 240,000 acres of land north of Lake Cachuma, which supplies water to Santa Barbara, CA. USGS Streamgage 11124500 was able to record pre and post-fire stream discharge for the affected Santa Cruz Creek Watershed, of which 67% was burned. 80% of this burn was severe, which raises concern for extreme flood events following the wildfire. It has been observed that the extreme temperatures in wildfires not only damage vegetation but soils as well -- wildfires much more so than comparatively mild prescribed burns. Our research proposes analyzing precipitation and stream discharge data from the affected Santa Cruz Creek Watershed to quantify the effects of such widespread and severe wildfire. This study uses the Hydrologic Modeling System (HEC-HMS) from the Army Corps of Engineers (ACE) to perform event-based, lumped modeling. It also uses Gridded Surface Subsurface Hydrologic Analysis (GSSHA) to perform physics-based modeling of the same watershed. Doing so should deepen our understanding of the effects of increasingly common and severe wildfires on watershed characteristics like infiltration and streamflow.

How to cite: Walters, A., Pradhan, N. R., Floyd, I., and Lakshmi, V.: Modeling Post-Wildfire Rainfall Events in the Santa Cruz Creek Watershed using HEC-HMS and GSSHA, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20928, https://doi.org/10.5194/egusphere-egu24-20928, 2024.

16:56–16:57
16:57–17:17
|
EGU24-6456
|
NH7.1
|
solicited
|
Highlight
|
On-site presentation
Ubirajara Oliveira and Britaldo Soares-Filho

Firefighting has become increasingly difficult and costly due to climate change. In response, new tools, including online platforms, are emerging to help prevent and promptly combat ever more destructive wildfires. While those initiatives only provide maps of fire risk based on environmental and climatic conditions, which in general have a medium predictive capability, fire propagation models, although successful in predicting fire behavior and spread, particularly at local scale, can become impractical during emergency situations, since they require lots of spatial data that must be obtained, processed and input by the user. To overcome these limitations, we have developed a fire-spread prediction system for the Brazilian Cerrado, the biome most affected by wildfires in South America. The system, named as FISC-Cerrado, automatically uploads hot pixels and satellite data to calculate maps of fuels loads, vegetation moisture, and post-probability of burning for simulating fire spread thrice a day for the entire Cerrado at 25 ha and for nine conservation units at 0.09 ha spatial resolution. Unlike the requirements to operate fire spread models, the user-friendly interface of FISC-Cerrado, alongside the automatization of the entire chain of tasks, allows its use by practitioners who do not have technical skills, such as GIS knowledge. Model results together with ancillary data, e.g., historical burned areas and annual CO2 emissions from fires, are available on an interactive web-platform (https://csr.ufmg.br/fipcerrado/en/), which is being used for daily operations by the fire brigades of the selected conservations units. 

How to cite: Oliveira, U. and Soares-Filho, B.: The FISC-Cerrado near-real time web-system for predicting fire spread, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6456, https://doi.org/10.5194/egusphere-egu24-6456, 2024.

17:17–17:27
|
EGU24-5991
|
NH7.1
|
Virtual presentation
Flavio T. Couto, Cátia Campos, Jean-Baptiste Filippi, Roberta Baggio, Carolina Purificação, Filippe L. M. Santos, and Rui Salgado

In 2017, Portugal was affected by several mega-fire episodes, which led the convective clouds formation, i.e., pyroCumulus (pyroCu) or pyroCumulonimbus (pyroCb). The pyroCb plays a crucial role in the fire front evolution through feedback processes between the atmosphere and the fire, including increased burn and spread rates by surface wind speed and direction variations. In order to investigate the pyro-convective activity during mega-fire events, numerical simulations were performed with the Meso-NH atmospheric model coupled to the ForeFire fire propagation model. The present study considers the mega-fires occurred in Pedrógão Grande and Góis on June 17, 2017, and in Quiaios on October 15, 2017. The experiments were configured into three nested domains with horizontal resolution of 2000 m (600 km × 600 km), 400 m (120 km × 120 km) and 80 m (24 km × 24 km) for the innermost model. The vertical resolution is the same for all the nested domains, with 50 levels and a first level above the ground at 30 m height. Initial and lateral boundary conditions for the outer domain were provided by ECMWF analysis, with updates every 6 h. Heat and water vapour were emitted into the atmosphere using the ForeFire model. In this case, the fire front evolution is directly imposed from a pre-defined time of arrival map (one-way coupling) and obtained from official reports. The results from the simulation of 80 m horizontal resolution showed that in the Pedrógão Grande mega-fire, the violent fire-driven convection manifested as a pyroCb cloud. The convective column penetrated the upper troposphere, and an intense outflow originated from the pyroCb cloud. In Quiaios mega-fire, the simulation also well represented the pyro-convection phenomenon, characterised by a northward-oriented smoke plume and the development of a pyroCu cloud. This study has provided important insights into the numerical modelling of pyroconvective clouds using Meso-NH/ForeFire simulations. This study was funded by national funds through FCT-Foundation for Science and Technology, I.P. under the PyroC.pt project (Ref. PCIF/MPG/0175/2019).

How to cite: Couto, F. T., Campos, C., Filippi, J.-B., Baggio, R., Purificação, C., Santos, F. L. M., and Salgado, R.: Towards a better understanding of pyroconvective clouds using Meso-NH/ForeFire coupled model, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5991, https://doi.org/10.5194/egusphere-egu24-5991, 2024.

17:27–17:37
|
EGU24-2647
|
NH7.1
|
Virtual presentation
María Isabel Asensio, José Manuel Cascón, and José Manuel Iglesias

We present a comprehensive simulation toolset for the analysis and prediction of wind fields, wildfire spread, and the propagation of their resulting smoke. It is composed of three physical simulation models that work together: HDWind, PhyFire and PhyNX.

The wind field simulation model, HDWind, is a mass consistent vertical diffusion wind field model based on an asymptotic approximation of the Navier-Stokes equations, providing a 3D wind field (which satisfies the incompressibility condition in the air layer) governed by a 2D equation that id adjusted to meteorological data obtained in a small number of points by solving an optimal control problem. PhyFire is a simplified 2D one-phase fire spread simulation model based on the principles of mass and energy conservation and that considers the radiation and convection (i.e., driven by wind and terrain slope) means of propagation, featuring the most relevant 3D effects, the influence of humidity, ambient temperature, wind, and the fuel types as well as their moisture content. The atmospheric dispersion model PhyNX is an urban scale Eulerian non-reactive multilayer air pollution model, able to describe convection, turbulent diffusion, and emission, considering the 3D wind field provided by the HDWind, and the smoke emission provided by PhyFire. The three models are solved using mainly the finite element method and some numerical and computational procedures to reduce the computational cost.

The required data to feed the simulation models such as cartographic and meteorological information are obtained from online geospatial information systems (GIS) in an automated way with little user intervention, thanks to the integration of the models with the geospatial library GDAL/OGR, which enables easy interpolation with most used standard GIS formats and services. An integration of this toolset into an easy-to-use webgis platform for their exploitation for professionals in the field of wildfire prevention will be demonstrated.

How to cite: Asensio, M. I., Cascón, J. M., and Iglesias, J. M.:  A comprehensive wind-fire-smoke simulation tool based on physical models and geospatial information., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-2647, https://doi.org/10.5194/egusphere-egu24-2647, 2024.

17:37–17:47
|
EGU24-20768
|
NH7.1
|
ECS
|
Virtual presentation
Sibo Cheng and Rossella Arcucci

The growing frequency of wildfires globally has highlighted the importance of immediate fire forecasting. Traditional high-accuracy fire spread simulations, like cellular automata and computational fluid dynamics, are detailed but require extensive computational resources and time. Consequently, there has been a significant push towards developing machine learning-based fire prediction models. These models, while effective, tend to be specific to certain regions and demand a large volume of simulation data for training, leading to considerable computational demands across various ecoregions.

In response, this study introduces a generative approach using three-dimensional Vector-Quantized Variational Autoencoders. This method is designed to create spatial-temporal sequences predicting the progression of future wildfires in specific ecoregions. The effectiveness of this model was evaluated in the context of the Chimney fire, a notable recent wildfire in California. The results demonstrate that the model effectively produces realistic and structured fire scenarios, incorporating influential geophysical factors like vegetation and terrain slope. Additionally, the data generated by this model were used to develop and train a surrogate model for wildfire spread prediction. This surrogate model was successfully validated using both simulated data and actual data from the Chimney fire incident.

How to cite: Cheng, S. and Arcucci, R.: VQ-VAE generative model of spatial-temporal wildfire propagation, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-20768, https://doi.org/10.5194/egusphere-egu24-20768, 2024.

17:47–18:00

Orals: Wed, 17 Apr | Room 1.14

The oral presentations are given in a hybrid format supported by a Zoom meeting featuring on-site and virtual presentations. The button to access the Zoom meeting appears just before the time block starts.
Chairpersons: Joana Parente, Andrea Trucchia
08:30–08:35
Fire Danger/Risk assessment (Joana Parente & Andrea Trucchia)
08:35–08:55
|
EGU24-17121
|
NH7.1
|
solicited
|
Highlight
|
On-site presentation
Carlos C. DaCamara, Mariana Ponte Oliveira, Sílvia A. Nunes, Ricardo M. Trigo, and Isabel F. Trigo

Meteorological fire danger has been steadily increasing over Europe in the last decades, not only over the Mediterranean South that is recurrently affected by extreme fire weather and where the largest fire events take place, but also over Central, Eastern and Northern countries that are facing more and more events. The two most recent examples are the devastating fires in Rhodes and northern Greece in 2023, and those in France, Spain, Portugal, Slovenia and Czechia in 2022 when the total of burnt area almost reached the record value of 2017. The increase in severity of fire events is of major concern for all European countries, but special attention should be devoted to Central Europe where large fires, usually driven by the compound effect of droughts and heatwaves (e.g., 2018, 2022), are posing new challenges at the levels of fire management and fire forecasting.

We present a statistical model of energy released by wildfires that allows calibrating the Canadian Fire Weather Index (FWI) over three major land cover types (forest, shrub, and agriculture) covering an area encompassing Central Europe (3.5º-17ºE and 45º-62ºN). The model consists of a doubly truncated lognormal body distribution with generalized Pareto tails (DaCamara et al., 2023) that incorporates FWI as a covariate of its parameters. For each land cover type, the model is fitted to the set of observed values (from 2001 to 2022) of the logarithm of Fire Radiative Power associated to hotspots as detected by the MODIS instrument on-board Terra and Aqua platforms. For each model, goodness of fit is evaluated by using the Anderson-Darling test to assess the strength of the evidence against the null hypothesis that the sample follows the distribution.

The fitted models allow estimating for each land cover type the probability of exceedance of a predefined threshold of log(FRP) for each day and grid point. Five classes of fire danger (low, moderate, high, very high, and extreme) for each land cover type are then defined by analyzing the spatial and temporal variability of the distribution of pixels among classes as well as the distribution among classes of FRP associated to hotspots, such that classes of higher fire danger tend to concentrate in the fire season, and fires with high values of FRP occur in pixels classified in the classes of high, very high and extreme danger. The procedure is further validated by examining several case studies that were chosen because of unusually intense fire events or because of the high number of occurrences.

 

This work was supported by EUMETSAT Satellite Application Facility on Land Surface Analysis (LSA SAF) and by Instituto Dom Luiz (IDL), a research unit financed with national funds (PIDDAC) by FCT (UIDB/50019/2020).

 

References:

 

DaCamara, C. C., Libonati, R., Nunes, S. A., de Zea Bermudez, P., & Pereira, J. M. C. (2023). Global-scale statistical modelling of the radiative power released by vegetation fires using a doubly truncated lognormal body distribution with generalized Pareto tails. Physica A: Statistical Mechanics and Its Applications, 625. https://doi.org/10.1016/j.physa.2023.129049

How to cite: DaCamara, C. C., Oliveira, M. P., Nunes, S. A., Trigo, R. M., and Trigo, I. F.: Early warning meteorological fire danger over Central Europe, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17121, https://doi.org/10.5194/egusphere-egu24-17121, 2024.

08:55–09:05
|
EGU24-15610
|
NH7.1
|
ECS
|
On-site presentation
Oscar Mirones, Jorge Baño-Medina, Joaquín Bedia, Swen Brands, and Mario Santa Cruz

The accurate prediction of the Fire Weather Index (FWI) is vital for effective wildfire management and climate-resilient planning. Multisite fire hazard forecasts are crucial for resource allocation, early intervention in high-risk areas, and identifying potential “megafire” threats from multiple simultaneous fire spots. Therefore, it is very important to account for the spatial consistency of these forecasts. This study examines the performance of Convolutional Neural Networks (CNNs) as a Statistical Downscaling (SD) technique for predicting FWI in different locations in the Iberian Peninsula. We contrast CNNs with two conventional SD methods: Generalized Linear Models and analogs. Using daily observed FWI data as predictands and ERA-Interim fields as predictors under a cross-validation setup, we discover that the CNN-Multi-Site-Multi-Gaussian (CNN-MSMG) model outperforms in daily FWI forecasting. This model integrates the covariance structure of the predictands into the CNN design, producing spatially consistent FWI forecasts. Furthermore, CNN-MSMG shows desirable features for estimating fire hazard in the climate change scenario, such as strong spatial consistency of extreme events and the capacity to generalize to new climate situations. These findings have important implications for improving FWI forecast accuracy and strengthening wildfire risk evaluation under climate change.

How to cite: Mirones, O., Baño-Medina, J., Bedia, J., Brands, S., and Santa Cruz, M.: Spatially Consistent Fire Weather Index Predictions using Convolutional Neural Networks in Diverse Iberian Locations, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15610, https://doi.org/10.5194/egusphere-egu24-15610, 2024.

09:05–09:15
|
EGU24-18455
|
NH7.1
|
ECS
|
On-site presentation
A. Serkan Bayar, Alexandre M. Ramos, Célia Gouveia, and Joaquim G. Pinto

Increasing temperatures and harsher drought conditions in recent decades have enhanced the risk of wildfire in many regions across the globe. Recent fire activity in Central Europe raised concerns about the possible expansion of the fire weather danger conditions under climate change outside the present-day fire-prone regions, such as the Mediterranean Basin. Here, we employ the widely used Canadian Fire Weather Index (FWI) system to assess the historical and future trends in the fire weather danger for Central Europe. Calculation of the originally proposed FWI requires utilizing noon-time temperature, relative humidity, wind, and accumulated precipitation. Using the ERA5 reanalysis dataset, we make sensitivity analyses with different combinations of alternative input data for noon-time meteorological parameters and estimate their biases.

This study uses an ensemble of regional climate models (RCM) from the EURO-CORDEX domain. We first compare the results from ERA5 with the RCM ensemble for the historical period. Then, we analyze future projections for Central Europe under different global warming levels (+2 K and +3 K). Results indicate that the fire-prone areas consistently increase under warmer climate conditions, including emerging fire-prone regions in Central and Northern Europe.

How to cite: Bayar, A. S., Ramos, A. M., Gouveia, C., and Pinto, J. G.: Projections of the Fire Weather Danger over Central Europe using EURO-CORDEX simulations, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-18455, https://doi.org/10.5194/egusphere-egu24-18455, 2024.

09:15–09:25
|
EGU24-14686
|
NH7.1
|
ECS
|
Highlight
|
On-site presentation
Anasuya Barik and Somnath Baidya Roy

We developed a comprehensive fire risk assessment framework for Indian forests, divided into five distinct forest zones (Himalayan, Northeast, Central India, Deccan, and Western Ghats) characterized by diverse climatic conditions and forest types. This framework focused on three primary triggering factors: weather, fuel availability, and anthropogenic ignition.

For the weather factor, we considered the Fire Weather Index (FWI) module of the Canadian Forest Fire Danger Rating System with ECMWF's ERA5 reanalysis as meteorological inputs over the period 2003-2021. As fire weather is a dominant factor in causing fires, we developed a robust system to predict fire weather danger. We evaluated the simulated FWI against MODIS active fire data and observed that FWI was a good enough metric for fire weather danger assessment. FWI was categorized into five danger classes through an ensemble approach based on logistic regression, FWI percentiles, percentage of fires, and K-means clustering. We introduced machine learning techniques to reduce the subjective decisions in these methods. This increased the efficiency of the danger rating system to detect fire probability well by 30-50%. A rigorous evaluation of the danger classes revealed that there was no overlap of central tendencies between different methods in the ensemble. The defined danger classes demonstrated coherent values for evaluative parameters, with a consistently high hit rate, low hits due to chance, moderate correct rejections, and an acceptable false alarm ratio.

Addressing fuel availability, we used vegetation indices (MODIS normalized difference and enhanced vegetation indices) and topographic features (aspect, elevation and slope from FLDAS land surface model). The anthropogenic ignition factor consisted of population density and land use information. In India, fragmented forests cohabitate with human settlements and agricultural lands. To quantify the impact of anthropogenic ignition on fire occurrences, we computed the percentage of built-up and agricultural area within each grid cell. We used machine learning predictive algorithms such as multiple linear regression with interactions, support vector machines, decision trees and neural networks to integrate these triggering factors with fire count as the target variable, We selected the highest-performing system as the risk assessment framework.

This country-scale fire risk assessment provides insights into regional exposure variations and serves as a foundational step towards establishing an operational fire risk assessment system for India. This framework will be of help to operational fire management agencies, enabling enhanced prediction of fire danger and informed decision-making.

How to cite: Barik, A. and Baidya Roy, S.: Machine learning based fire danger assessment framework for Indian forests , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14686, https://doi.org/10.5194/egusphere-egu24-14686, 2024.

09:25–09:35
|
EGU24-795
|
NH7.1
|
ECS
|
On-site presentation
Mariana Ponte Oliveira, Sílvia A. Nunes, Carlos C. DaCamara, Ricardo M. Trigo, and Isabel F. Trigo

The Satellite Application Facility for Land Surface Analysis (LSA SAF), that is part of EUMETSAT’s ground segment, operationally disseminates daily forecasts of meteorological fire danger over Mediterranean Europe. The so-called Fire Risk Map (FRM) product relies on estimates of the probability of exceedance of predefined thresholds of daily released energy by active fires as derived from a Generalized Pareto model that uses FWI as covariate for the scale parameter; FWI, the Fire Weather Index (FWI), is part of the Canadian Fire Weather Index System and has proven to be very suitable to rate fire danger over Europe.

The aim of this study is to extend the procedure to Northern and Central Europe making use of a statistical model of Fire Radiative Power (FRP) as derived from MODIS observations over Europe covering the period 2000-2022. Following the approach developed by DaCamara et al. (2023), the statistical model consists of an 8 parameter, doubly truncated lognormal body distribution with generalized Pareto tails, using FWI as a covariate of its parameters.

First, Europe is divided into eight regions according to the recorded number of hotspots and to the averaged FRP over the study period. Each one of those regions is then stratified into three subregions according to the dominant land cover type, i.e., forest, shrub, or agriculture. For each of the subregions, a statistical model is fitted to the sample of historical records of FRP together with the associated sample of FWI values obtained from the Copernicus Emergency Management Service.

The fitted models are then applied to Europe to generate monthly climatological values of probability of exceedance of prescribed thresholds of FRP. This information is used to define appropriate limits for classes of fire danger (i.e. low, moderate, high, very high and extreme) for each subregion of Europe. Finally, these classes are validated by analyzing the distribution of recorded FRP among the classes and by examining maps for extreme fire events.

 

This work was supported by EUMETSAT Satellite Application Facility on Land Surface Analysis (LSA SAF) and by Instituto Dom Luiz (IDL), a research unit financed with national funds (PIDDAC) by FCT (UIDB/50019/2020).

 

References:

DaCamara, C. C., Libonati, R., Nunes, S. A., de Zea Bermudez, P., & Pereira, J. M. C. (2023). Global-scale statistical modelling of the radiative power released by vegetation fires using a doubly truncated lognormal body distribution with generalized Pareto tails. Physica A: Statistical Mechanics and Its Applications, 625. https://doi.org/10.1016/j.physa.2023.129049

How to cite: Ponte Oliveira, M., A. Nunes, S., C. DaCamara, C., M. Trigo, R., and F. Trigo, I.: Assessing meteorological fire danger over Europe based on a statistical model of satellite-derived fire radiative power, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-795, https://doi.org/10.5194/egusphere-egu24-795, 2024.

09:35–09:45
|
EGU24-1324
|
NH7.1
|
On-site presentation
Octavian Dumitru, Gottfried Schwarz, and Chandrabali Karmakar

This work investigates the occurrence, parameters, and consequences of fires in satellite images that can be directly exploited by several combinations of different multispectral image bands.

When we want to understand the semantics of a recorded digital image, we can cut it into smaller-size image patches and routinely classify these image patches via common unsupervised or supervised image classification techniques. In addition, when we include some clever interactive learning steps to attach semantic labels to the hitherto mathematically classified image patches, this should allow for a highly automated and powerful image understanding procedure.

On the other hand, starting with simple examples, the application-oriented analysis and exploitation of Sentinel-2 images can combine and display selected colour bands and their combinations. This has already been discussed in many (mostly GIS-oriented) publications ranging from the straightforward assignment of directly available pseudo-RGB colour bands up to advanced machine learning approaches for the extraction of content-related information (such as image feature descriptors or indices) [1-4]. Further, we will also refer to a few recently published advanced information extraction tools [5-10].

As an alternative to these (mostly conventional) image classifications, we describe a powerful semantic image classification technique that starts with the generation of topics (instead of classes) that was originally described by [11].Here, the resulting topic maps can be further combined and be used for colour band displays and their interpretation. When we combine the properties and capabilities of Sentinel-2 images with topic interpretation techniques, the most interesting question is whether a semantic interpretation based on topic maps outperforms common feature-based approaches.

To this end, we selected several Sentinel-2 multi-band images comprising different geographical areas affected by fires. This presentation shows the actual impact of various band combinations of Sentinel-2 channels and illustrates the band-dependent appearance of Fires, Smoke, Clouds, and other specific categories linked to the investigated continental areas. The basic algorithm being used for this investigation is Latent Dirichlet Allocation that has been applied as a data mining tool to discover patterns in the data, combined with automated band selection approaches.

The combination of automated image classification and multi-colour visualization seems to be an interesting alternative to Deep Learning.

[1] https://gisgeograpy,com/sentinel-2-bands-combinations

[2] https://worldofittech.com/sentinel2-bands-and-combinations

[3] https://giscrack.com/list-of-band-combinations-in-sentinel-2a

[4] https://eo4geocourses/github.io/IGIK_Sentinel2-Data-and-Vegetation-Indices

[5] A. Revill et al., The Value of Sentinel-2 Spectral Bands for the Assessment of Winter Wheat Growth and Development, Remote Sensing, 11(17), 2018.

[6] K. Kowalski et al., A generalized framework for drought monitoring across Central European grassland gradients with Sentinel-2 time series, Remote Sensing of Environment, 286, 2023.

[7] M.K. Vanderhoof et al., High-frequency Time Series Comparison of Sentinel-1 and Sentinel-2 Satellites for Mapping Open and Vegetated Water Across the United States, Remote Sensing of Environment, 288, 2023.

[8] E.C. Rodriguez-Garlito et al., Mapping Invasive Aquatic Plants in Sentinel-2 Images Using Convolutional Neural Networks Trained with Spectral Indices, JSTARS, 16, pp.2889-2899, 2023

[9] Z. Chen, et al., Mapping Mangrove Using a Red-Edge Mangrove Index (REMI) Based on Sentinel-2 Multispectral Images, TGRS, 61, pp.1-11, 2023.

[10] A. Temenos, Interpretable Deep Learning, GRSL, 20, pp.1-5, 2023.

[11] D.M. Blei, et al., Latent Dirichlet Allocation, Journal of Machine Learning Research, 3, pp.993-1022, 2003.

How to cite: Dumitru, O., Schwarz, G., and Karmakar, C.: Automated Selection of Sentinel-2 Spectral Bands for Fire Detection, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-1324, https://doi.org/10.5194/egusphere-egu24-1324, 2024.

09:45–09:55
|
EGU24-3838
|
NH7.1
|
ECS
|
On-site presentation
|
Byeongcheol Kim and Seonyoung Park

Forest fires pose significant threats to both human safety and the natural ecosystem. Detecting and accurately estimating the extent of the burned area is crucial for effective response planning. Remote sensing emerges as a valuable solution for estimating the burned area, with various satellites employed in previous studies. Unlike these satellites, microsatellites offer a promising alternative with higher spatiotemporal resolution. In this study, we utilized PlanetScope imagery and implemented the U-Net model. PlanetScope provides images at a 3m spatial resolution and revisits the same area every day, offering a distinct advantage in accurately estimating burned areas across different scales of fire events. However, PlanetScope lacks a Shortwave Infrared (SWIR) band commonly used in forest fire studies. To address this limitation, a virtual SWIR band was introduced in this study. To enhance accuracy in specific regions, a virtual SWIR band was created using machine learning techniques using the SWIR images from Landsat and Sentinel-2. Our approaches were tested in four study regions. The U-Net model was employed to generate burned area prediction maps, and each model's performance was evaluated using several metrics, including intersection-over-union (IoU), mean IoU, recall, precision, F1-Score, and the Kappa coefficient. In this study, we not only validated the effectiveness of our proposed methods but also identified the potential to enhance the accuracy of burned area estimations, particularly for microsatellites lacking a SWIR band.

How to cite: Kim, B. and Park, S.: Burned area detection based on Planet imagery using virtual SWIR band, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-3838, https://doi.org/10.5194/egusphere-egu24-3838, 2024.

09:55–10:05
|
EGU24-7851
|
NH7.1
|
On-site presentation
Christopher Marrs, Kristina Beetz, Johanna Kranz, Konrad Bauer, Evripidis Avouris, Markéta Poděbradská, Daniel Kinalczyk, and Matthias Forkel

Until recently, forest fires were considered a rare phenomenon in the temperate forests of Central Europe due to the moderate summer temperatures and the humid climate. However, many of those forests (e.g. monocultures of Picea abies, Norway Spruce) were affected by bark beetle infestations in the past years and recent fires such as in the Bohemian-Saxon Switzerland in 2022 raised widespread debates about the effects of forest mortality on fuel accumulation and hence fire occurrence and severity. Here we mapped and investigated fuel types, fire severity and started to continuously monitor fuel moisture in the Bohemian-Saxon Switzerland. We enhanced a European fuel type classification with a class for dead and dying spruce and mapped fuel types. Satellite observations from VIIRS, Sentinel-2 and Landsat were used to map fire intensity and severity of the fire from 2022.

We found the highest fire intensities at sites with dead spruce forests and single beech trees. Burn severity was moderate with high variability across all fuel types but highest severities occurred in dead spruce stands. Fire severity derived from satellite observation correlated positively with char height and torched trees, especially seen in dead spruce stands, which was likely caused by the high amount of dry fine woody debris and the initial natural regeneration. Our results demonstrate that surface fuel accumulation from past bark beetle disturbances resulted in more intense fires and higher burn severity. The results demonstrate that the recent rapid changes in Central European temperate forests cause a need for a dynamic mapping and monitoring of fuel types and fuel moisture for fire risk assessment and for cross-border fire risk management in landscapes previously not considered as fire-prone.

How to cite: Marrs, C., Beetz, K., Kranz, J., Bauer, K., Avouris, E., Poděbradská, M., Kinalczyk, D., and Forkel, M.: Investigating fuels, fuel moisture and fire severity in the Bohemian-Saxon Switzerland region (Czech Republic/Germany): the need for dynamic fire risk assessment and management , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7851, https://doi.org/10.5194/egusphere-egu24-7851, 2024.

10:05–10:15
|
EGU24-10384
|
NH7.1
|
ECS
|
On-site presentation
Eva Preinfalk and John Handmer

As socio-natural hazards, future extreme wildfires across Europe are exacerbated by two main drivers: climate change and socioeconomic dynamics. While modelling studies account for changes in fire danger and area burned under different climate scenarios, they largely disregard the impacts of land use change, or the interaction with adaptation through changes in vegetation management. This creates uncertainties regarding the role of anthropogenic processes and the reliability of projections under various policy scenarios. Next to the interaction with wildfire hazard, socioeconomic processes are shaping all dimensions of wildfire risk. Building on the IPCC notion that risk arises at the intersection of hazard, exposure and vulnerability, we screen the relevant empirical literature to identify key socioeconomic drivers of wildfire risk in a European context and bring this together with the Shared Socioeconomic Pathways (SSP) perspectives on plausible socioeconomic futures. The resulting wildfire risk scenario space serves two main purposes: (i) providing a qualitative navigator for incorporating socioeconomic uncertainty in model-based wildfire risk assessments and (ii) establishing boundary conditions for evaluating the feasibility of management strategies. Applying the SSP framework for envisioning plausible development trajectories, we systematically investigate the role of socioeconomic dynamics in determining future wildfire risk. Sustainable land use practices and profitable agricultural value chains reduce hazard (e.g. SSP1), while factors like poor environmental regulation (e.g. SSP5) and increasing pressure on land abandonment as competitive value chains disappear (e.g. SSP4), increase this dimension of wildfire risk. Exposure remains high across scenarios for different reasons. Ineffective land use planning contributes to the expansion of human settlements in areas dominated by unmanaged flammable vegetation (e.g. SSP3, SSP5), with a further escalation of livelihood exposure on poorly managed agricultural land (e.g. SSP4). Vulnerability becomes a distinctive driver of wildfire risk in scenarios with low rates of economic development and poor investment in human capital (e.g. SSP3, SSP4). While increased socioeconomic welfare may enhance coping capacities in the context of wildfires (e.g. SSP1, SSP5), the prioritization of business-related objectives in institutional risk management (e.g. SSP5) poses a risk of neglecting other critical aspects. As wildfires transition from a climate hazard into a potential disaster at the intersection with exposure and differential vulnerabilities, we emphasize the importance of addressing all three dimensions of risk. By expanding the view of future wildfire risk, we show that challenges to wildfire risk management differ significantly between scenarios. Social, economic and socioecological challenges may lead to paradoxical situations in managing wildfire risk. In scenarios, where vulnerability reduction has maximum leverage in reducing risk, socioeconomic challenges hinder the feasibility of implementing the measures necessary to achieve it. Similar dilemmas may arise in the context of hazard and exposure. By considering multiple plausible futures, this work stresses the importance of considering socioeconomic dynamics in shaping wildfire risk and keeping the design of risk management strategies open and flexible to adapt to changing circumstances.  

 

How to cite: Preinfalk, E. and Handmer, J.: Fuelling the fires - An exploration of the drivers and the scope for management of European wildfire risk under the Shared Socioeconomic Pathways , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10384, https://doi.org/10.5194/egusphere-egu24-10384, 2024.

Coffee break
Chairpersons: Joana Parente, Francesca Di Giuseppe
10:45–10:46
10:46–10:56
|
EGU24-1627
|
NH7.1
|
Virtual presentation
Gaofeng Fan and Zhonghua He

Wildfire risk prediction is a critical component of disaster prevention and mitigation, often closely associated with local human activities in most regions. Recent studies demonstrate that employing joint modeling techniques using diverse datasets alongside Convolutional Neural Networks-Long Short-Term Memory Networks (CNN-LSTM) produces favorable predictive results. This approach effectively tackles certain drawbacks of fire weather indices (FWI), notably the insufficient consideration of surface coverage and coarse resolution. However, previous research inadequately explored variations in the impact of influencing factors across different categories and spatial orientations, neglecting the internal structural features within the samples. This study focuses on the six eastern provinces of China, utilizing a multi-source dataset comprising satellite-monitored wildfire products from 2012 to 2022, along with terrestrial ecology, terrain, and simulated meteorological elements. By introducing channel and spatial attention mechanisms, high-resolution imagery, and visual transformer model, this research optimizes the CNN-LSTM wildfire prediction model. Results indicate a noteworthy enhancement, elevating accuracy, Kappa coefficient, and AUC of ROC curves from 91.15%, 80.87%, and 97.01% to 93.30%, 85.63%, and 98.15%, respectively. This refined model not only refines high-risk prevention areas highlighted by FWI but also enhances understanding of mountain trails in hilly terrains. Consequently, it reduces false alarms in regions such as non-harvesting agricultural fields, reinforcing predictive risk assessment concerning potential human activities within forested areas. Sensitivity analysis reveals that while the impact of internal sample structural features on wildfire risk prediction is lower than meteorological elements, it surpasses the influence of terrain and terrestrial ecology elements. Thus, this study has developed a methodology integrating multiple attention mechanisms and sample structural features, furnishing high-precision daily kilometer-level wildfire risk prediction products. This approach holds substantial promise for the precise prevention and control of regional wildfires.

How to cite: Fan, G. and He, Z.: Deep Learning Modeling of Human Activity Affected Wildfire Risk by Incorporating Structural Features: A Case Study in Eastern China, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-1627, https://doi.org/10.5194/egusphere-egu24-1627, 2024.

10:56–11:06
|
EGU24-14466
|
NH7.1
|
ECS
|
On-site presentation
Aryalakshmi Madhukumar, Jayaluxmi Indu, and Lanka Karthikeyan

Wildfires are becoming increasingly frequent and devastating in many tropical forests. In India, nearly 6% of forest cover is highly fire-prone, and 36% is prone to frequent fires. These frequent wildfires have compound impacts on forests, which include changes in biodiversity and, forest functionality. In the short term, the burnt ecosystem cannot have the same functionality as that of a pre-fire situation. Understanding forest ecosystem health status after a fire event can help increase our ability to manage the fire seasons. 
In the current study, time series data from optical remote sensing is used to assess the forest ecosystem resilience. The study area chosen is the forest region in Uttarakhand, India, including Jim Corbett National Park, that witnessed a severe forest fire in 2016. The active fire pixels during the 2016 fire event in the area were identified, and resilience over the identified pixels was measured based on the concept of engineering resilience.  Results are presented as resilience quantified for all burnt pixels based on three different indices which represent resistance and recovery namely, time taken for recovery, maximum impact of fire event, and cumulative impact. Further, we identified how resilience differs with the type of forest cover in the study area and how well each type of forest recovers from the fire event. The findings suggest that in the Indian scenario, deciduous broadleaf forests have a longer recovery followed by evergreen broadleaf and evergreen needleleaf, while grasslands and broadleaf cropland have shorter recovery times and impacts. From this work, we aim to study forest resilience in the Indian scenario and how well this can be compared with other areas where similar climatic conditions exist. The current work has potential applications in risk governance, ecosystem management, etc. and in evaluating the post-fire processes and primary factors driving the processes. Extensive data feeds available from current satellite platforms enable the post-fire dynamics study to be more accurate, thus more informed, and faster choices by stakeholders.

How to cite: Madhukumar, A., Indu, J., and Karthikeyan, L.: Indicator-based measurement of resilience and analysis of spatial trend in resilience to forest fire in Uttarakhand, India, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14466, https://doi.org/10.5194/egusphere-egu24-14466, 2024.

11:06–11:16
|
EGU24-3221
|
NH7.1
|
ECS
|
On-site presentation
Aimée Guida Barroso, Gean Paulo Michel, Franciele Zanandrea, Márcio Vinicius Aguiar Soares, Gabriel Ferreira Subtil de Almeida, Marcio Cataldi, Priscila Esposte Coutinho, and Livia Sancho

Wildfires represent a significant threat to natural ecosystems, biodiversity, and communities worldwide. Disruption in precipitation regimes and temperature rise caused by climate change are key factors that worsen and increase wildfire incidents. In Brazil, recent studies have shown the majority of fire incidents are initiated by anthropogenic action, as a consequence of agricultural expansion, deforestation and land disputes. Although the human use of fire as an illegal tool is difficult to predict, the occurrence of dry meteorological conditions, prone to uncontrolled spreading of fires, can be studied employing climate modeling, providing a useful instrument to aid authorities in preventive measures and improved responses to mitigate these impacts, contributing to more efficient and sustainable management of fire-related risks. The Standardized Precipitation Index (SPI) is a useful tool for assessing precipitation variability, allowing the analysis of drought period duration, distribution, and severity. The SPI uses precipitation data to standardize the deviation of accumulated precipitation from the historical average in each location. This process yields negative or positive values, which correspond to water deficits or surpluses, respectively. Aiming to identify areas in Brazil where predicted disruption in rainfall patterns, in face of climate change, may create drier conditions and increase vulnerability to fire incidents, we evaluated precipitation trends, comparing historical simulations from the 6th phase of the Model for Interdisciplinary Research on Climate (MIROC6) and future scenarios data from the Intergovernmental Panel on Climate Change (IPCC). We focused our analysis on 3 climate change scenarios, referred to as Shared Socioeconomic Pathways: SSP2-4.5, SSP3-7.0, and SSP5-8.5. These scenarios encompass anticipated global socioeconomic transformations up to the year 2100, based on different projections of greenhouse gas emissions, and offer an assessment of the climate outlook for current society. Thus, we calculated SPI indexes for the time spans 1960-1990 and 2020-2050, examining the variations in rainfall patterns across the country during both periods. Using SPI derived from MIROC6 climatological data, it is possible to identify past patterns that are the basis for understanding future changes' impact. The results from SPI climatological data are consistent with the climate and seasonal rainfall patterns historically observed in Brazil, where Northeast and Central Brazil exhibit greater water deficits. The scenarios employed suggested that the historical patterns of droughts would be worsened in severity in central Brazil and the areas of influence would be extrapolated, creating drier meteorological conditions to the Southern and East portions of Amazonia and the Southeast of Brazil. The SPI indexes calculated to the projected scenarios reinforce the understanding of the impacts of climate change, suggesting the pathway SSP55-8.5, with higher emission of CO2, implicates in increased occurrences of extreme events, particularly prolonged and severe droughts in regions that suffer from wildfires. Identifying regions with an increased likelihood of prolonged drought events in the projected future is a valuable instrument for examining fire hazard and mitigation plans within a country such as Brazil, which encompasses diverse climates and biomes across its territory with resources of significant conservation value.

How to cite: Guida Barroso, A., Michel, G. P., Zanandrea, F., Vinicius Aguiar Soares, M., Ferreira Subtil de Almeida, G., Cataldi, M., Esposte Coutinho, P., and Sancho, L.: Standard Precipitation Index (SPI) applied to Socioeconomic Pathway Scenarios (SSPs) as a tool to map the distribution of droughts and potential fire hazard areas in Brazil in the face of climate change, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-3221, https://doi.org/10.5194/egusphere-egu24-3221, 2024.

11:16–11:26
|
EGU24-6060
|
NH7.1
|
On-site presentation
Sven Fuchs, Pia Echtler, David Hausharter, Matthias Schlögl, and Maria Papathoma-Köhle

Changes in temperature and precipitation in Austria due to climate change are expected to increase the days of fire weather in the near future. Extreme wildfire events are not common in Austria, nevertheless, given climate change and an increase in the number of events in the last years, authorities and decision-makers require tools to identify vulnerable hotspots and to reduce the upcoming risk in the Wildland Urban Interface. A number of projects ran by the University of Natural Resources and Life Sciences in Vienna focus on the assessment of vulnerability at different levels (national and local) and different elements at risk (industrial and residential buildings, free spaces and infrastructure). We present herein the current research landscape on the field in Austria and more specifically the projects PHLoX (StartClim), REVEAL (Waldfonds), and FIREPRIME (DG ECHO). Each project is based on expert knowledge and data-driven approaches and deals with different elements at risk and vulnerability indicators emphasising the need for participatory methods for wildfire risk management. We demonstrate how these projects can serve as a blueprint for increased wildfire resilience in Europe and beyond.

How to cite: Fuchs, S., Echtler, P., Hausharter, D., Schlögl, M., and Papathoma-Köhle, M.: The research landscape of wildfire vulnerability in Austria, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6060, https://doi.org/10.5194/egusphere-egu24-6060, 2024.

11:26–11:36
|
EGU24-6539
|
NH7.1
|
On-site presentation
Maria Papathoma-Koehle, Pia Echtler, Sven Fuchs, Matthias Schlögl, Mortimer Müller, and Harald Vacik

Changes in temperature and precipitation in the European Alps are reflected in an increasing number of wildfire events and burnt areas. Therefore, apart from conducting research on the behaviour of wildfires in regions with high fire danger, it is important to analyse the vulnerability of settlements, buildings, and infrastructure also in areas with less experience with the impacts of an increasing wildfire hazard. Studies focusing on the vulnerability of the built environment do exist, but they are mostly limited to the interaction of buildings with fire, rather than offering a tool to measure this vulnerability for planning and conducting risk reduction measures. We attempt to close this gap by assessing the physical vulnerability of elements at risk located at the Wildland Urban Interface (WUI) in several case study areas in the Austrian Alps. In the absence of empirical data and by using a co-creation approach, we engage experts from various domains (firefighters, managers, planners, and government officials) to develop a tool for wildfire risk management. The tool is based on indicators to assess the vulnerability of different characteristics of elements at risk, including residential buildings, hotels, industry, critical infrastructure, and cultural heritage. Applications in each case study area are designed to demonstrate the usability of the tool in various disaster risk reduction activities and different contexts with a high or low wildfire danger. A final workshop is planned to ensure the dissemination of the results in the entire wildfire community. The resulting REVEAL decision support tool should be applicable not only in other regions of Austria, but also in other European regions that do not have experience and empirical data related to the impacts of wildfire events on infrastructure.

How to cite: Papathoma-Koehle, M., Echtler, P., Fuchs, S., Schlögl, M., Müller, M., and Vacik, H.: Assessing Wildfire Vulnerability in the absence of empirical data: the REVEAL Project, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6539, https://doi.org/10.5194/egusphere-egu24-6539, 2024.

11:36–11:46
|
EGU24-6775
|
NH7.1
|
ECS
|
On-site presentation
Khabat Khosravi and Aitazaz Farooque

Abstract

Wildfire Susceptibility Assessment (WSA) is one of the critical approaches to wildfire risk management. In this study, we employed a hybrid approach by integrating two distinct statistical models, namely Frequency Ratio (FR), Weight of Evidence (WoE), with Shannon Entropy (SE) (i.e., FR-SE and WoE-SE) for WSA. To meet the aim, 18538 historical wildfire data were collected and separated into two sections for model development and validation. Next, 13 wildfire-influencing parameters, including slope degree, aspect, topographic wetness index, elevation, evapotranspiration, land use/land cover, normalized differences vegetation index, distance from the lake, precipitation, distance from the rivers, distance from the roads, soil moisture, and mean annual maximum temperature were prepared and feed the models. Finally, model performance were evaluated using the validation data set and receiver operating characteristic (ROC) curve technique. Findings shows that the integration of models has improved the modeling performance, as WOE-SE model has the highest performance (96.5%), followed by WoE (96.3%), SE-RF (95.9%) and RF (95.2%) model respectively. Result of SE model showed that mean annual maximum temperature has the highest impact on the wildfire occurrence across Canada, while topographic wetness index is the lowest effective parameter.

Keywords: Wildfire, statistical models, Canada, Shannon Entropy, Frequency ratio, Weight of Evidence.

How to cite: Khosravi, K. and Farooque, A.: Canada’s Wildfire Susceptibility Assessment Using Statistical Data-Driven Models, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6775, https://doi.org/10.5194/egusphere-egu24-6775, 2024.

11:46–11:56
|
EGU24-614
|
NH7.1
|
ECS
|
On-site presentation
Zarmina Zahoor, Jonathan Eden, Matthew Blackett, and Yung-Fang Chen
 

Wildfires are becoming more intense and frequent, with record-breaking fire seasons witnessed across the world in recent years. Amid rising global temperatures, the challenge to understand, communicate and ultimately reduce wildfire risk is critical. A recent report published by the United Nations Environment Programme noted a particular increase in fire prevalence across regions that were not previously considered fire-prone, including the Indian subcontinent. In Pakistan, wildfire has gradually emerged as a significant environmental and societal threat. However, it is unclear how such threats will evolve under climate change, and to what extent Pakistan’s ongoing afforestation projects, such as the Ten Billion Tree Tsunami, take changes in risk into account. 

Here, we explore how meteorological conditions conducive to wildfire are likely to respond to a changing climate throughout Pakistan. Following an initial spatiotemporal analysis of wildfire occurrence based on satellite-derived data between 2001 and 2020, we identity hotspots of fire activity across the forested regions of the Baluchistan, Kashmir, Khyber Pakhtunkhwa and Punjab provinces. Using the fire weather index (FWI) derived from the simulations of 14 global climate model ensembles from the 6th phase of the Coupled Model Intercomparison Project (CMIP6), we then quantify changes in fire danger throughout the 21st century under four climate change scenarios defined by the Shared Socioeconomic Pathways (SSPs). We show that the magnitude of seasonal mean FWI is projected to increase by as much as 10% by the end of the century under the highest emissions scenario, with up to 20 additional days of extreme fire weather projected per year.  

Our conclusions advise on how forest management strategies and afforestation projects across Pakistan should account for potential changes in wildfire risk associated with a changing climate. We introduce a prototype online portal as a mechanism to disseminate results and communicate future risk to a range of potential stakeholders. Further work will focus on the resilience of wildfire forecasting and early warning systems in a changing climate.  

 

How to cite: Zahoor, Z., Eden, J., Blackett, M., and Chen, Y.-F.: Spatiotemporal analysis and projections of wildfire risk across Pakistan under different climate change scenarios, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-614, https://doi.org/10.5194/egusphere-egu24-614, 2024.

11:56–12:06
|
EGU24-16436
|
NH7.1
|
ECS
|
Highlight
|
On-site presentation
Alba Marquez Torres and Diego Bengochea Paz

In an era of increasing wildfire incidents worldwide, fire risk mapping has emerged as a crucial tool for ecosystem management and environmental safeguarding against the significant loss of socio-ecological value. Our research introduces a novel daily global fire risk model, combining the probability of fire ignition as a fire hazard model with an analysis of exposure and vulnerability. This model was calculated with over 4 million historical fire and non-fire ignitions recorded between 2000 and 2020 and tested with more than 24 million ignition points. It integrates key explanatory variables encompassing climatic conditions, agro-environmental factors, terrain, and social drivers at the time of fire ignition, processed through advanced machine learning techniques, such as the XBoost Random Forest algorithm.

Further enhancing our model's robustness, we incorporate a suite of socio-ecological models, previously developed using machine reasoning, an AI algorithm based on semantics, through the k.LAB platform. These models cover critical areas such as vegetation carbon mass, pollination, outdoor recreation, and soil retention, enabling us to identify regions where humans and nature are most vulnerable to fire hazards.

Adhering to FAIR principles, our approach ensures that our data and models are findable, accessible, interoperable, and reusable. This commitment not only advances scientific research but also promotes broader application and collaboration. The global fire risk model provides temporally and spatially explicit results on a daily basis, offering a dynamic and precise tool for understanding and preventing fire risks.

This research has significant implications for policymaking and emergency response planning. By offering a detailed and dynamic understanding of fire risks, stakeholders can make informed decisions that can mitigate the impact of wildfires. The combination of diverse datasets, advanced analytical techniques, and a focus on practical applications makes this model a valuable resource in the global effort to address the increasing challenges of wildfires.

How to cite: Marquez Torres, A. and Bengochea Paz, D.: Advanced Wildfire Risk Mapping: A Novel Global Approach Using AI and Socio-Ecological Data, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16436, https://doi.org/10.5194/egusphere-egu24-16436, 2024.

12:06–12:26
12:26–12:30

Posters on site: Wed, 17 Apr, 16:15–18:00 | Hall X4

The posters scheduled for on-site presentation are only visible in the poster hall in Vienna. If authors uploaded their presentation files, these files are linked from the abstracts below, but only on the day of the poster session.
Display time: Wed, 17 Apr 14:00–Wed, 17 Apr 18:00
Chairpersons: Joana Parente, Andrey Krasovskiy, Francesca Di Giuseppe
Fire, drivers & recovery
X4.61
|
EGU24-8496
|
NH7.1
|
ECS
|
Highlight
|
|
Andrina Gincheva and the ONFIRE group

We present the ONFIRE Dataset (Gincheva et al., 2023), a gridded monthly burned area (BA) data product with national wildland data from several regions: Australia (since 1950), Canada (since 1959), Chile (since 1985), Europe (since 1980) and the United States (since 1984), covering up to the year 2021. This database is organised on a uniform 1° × 1° grid, providing a consistent spatial resolution for global analysis. Records from different sources and regions have been extracted and harmonised using open and reproducible methods. The data remapping and validation process ensures consistency and comparability between different regions. This dataset complements existing remotely sensed databases, offering users the opportunity to explore and analyse changes in fire regimes. The ONFIRE Dataset is accessible on Zenodo (https://zenodo.org/records/8289245; Gincheva  & Turco,  2023).

References

Gincheva, A., Pausas, J. G., Edwards, A., Provenzale, A., Cerdà, A., Hanes, C., ... & Turco, M. (2023). A monthly gridded burned area database of national wildland fire data (ONFIRE).

Gincheva, A., & Turco, M. (2023). ONFIRE dataset: Monthly Gridded Burned Area data (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.8289245

Acknowledgements

A.G. thanks to the Ministerio de Ciencia, Innovación y Universidades of Spain for Ph.D. contract FPU19/06536. M.T. acknowledges funding by the Spanish Ministry of Science, Innovation, and Universities through the Ramón y Cajal Grant Reference RYC2019-027115-I and through the project ONFIRE, grant PID2021-123193OB-I00, funded by MCIN/AEI/ 10.13039/501100011033 and by “ERDF A way of making Europe”. S.J. acknowledges funding by the MCI/AEI Ramón y Cajal Grant Reference RYC2020-029993-I. M.B., A.P., and M.M. acknowledge the support of the European Union - NextGenerationEU in the framework of the National Biodiversity Future Center of Italy; A.P. and M.M. acknowledge the support of the EU project FireEUrisk, grant no. 101003890. M.E.G acknowledges research support provided by ANID/FONDECYT N° 1231573 and ANID/FONDAP 15110009; COD 1522A0001. R.L. was supported by FAPERJ (Grant E-26/200.329/2023) and CNPQ (Grant 311487/2021-1). M.M.B. acknowledges funding from the New South Wales Government (NSW Bushfire and Natural Hazards Research Centre) and the Australian Research Council (DP 220100795). F.M. and E.C. were supported by the European Space Agency FireCCI project. 

How to cite: Gincheva, A. and the ONFIRE group: ONFIRE Dataset: Harmonizing Decades of Wildland Fire Data, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-8496, https://doi.org/10.5194/egusphere-egu24-8496, 2024.

X4.62
|
EGU24-9132
|
NH7.1
|
Highlight
Marj Tonini, Giorgio Meschi, Andrea Trucchia, and Paolo Fiorucci

In the southern European countries, the combination of climate change, substantial shifts in land use/land cover, and socio-economic factors acting over the last decades are anticipated to increase the frequency, scale, and intensity of wildfires unless enhanced prevention and control strategies are implemented. Statistical and data-driven approaches are widely used by researchers to evaluate the main variables controlling wildfires occurrences and spreading. Lately machine learning proved to be highly performant due to its flexible and non-linear nature, capable of capturing the complexity of the wildfire process. Nevertheless, conventional classification and regression methods like Support Vector Machine, pixel-based Neural Network, and Random Forest (RF), are global modelers, not calibrated to deal with the spatial heterogeneity of the investigated area. Thus, these algorithms turn out to be incapable of adequately addressing the spatially varying underlying relationship between wildfires pattern distribution and the predisposing variables.

While many studies seek to assess the importance of the predictor variables both at regional [1, 2] and at supranational level [3], up to now there is a lack of studies attempting to account for the spatial heterogeneity (i.e. non-stationarity) when modeling wildfires spatial patterns as function of geographical features.  To fill this gap, the present work explores the local feature importance of geographical independent predisposing variables on the spatial distribution of burned area density in the Mediterranean area. To this end, we have used the last development of Geographical Random Forest (GRF) [4], which integrates a parallelizable RF function, a procedure for the bandwidth optimization, and an option to spatially weight the local observations. As dependent variables we considered the percentage of burned pixels per map unit. Both geo-environmental features (i.e., variables providing information on the topography and land cover) and anthropogenic features (e.g., distances to urban areas and road network) have been select as predictors. The importance of these independent variables has been assessed by evaluating the Mean Decrease Accuracy (MDA) by using the Out of Bag samples available in RF: higher values mean that the model strongly benefits from the given variable when performing predictions. The spatial variation of each predisposing factor was illustrated by mapping the corresponding MDA values over the geographical space. Finally, the implemented model has been validated by using the root mean squared error computed over an independent testing dataset.   

[1] Trucchia A, Izadgoshasb H, Isnardi S, Fiorucci P, Tonini M, 2022. Machine-Learning Applications in Geosciences: Comparison of Different Algorithms and Vegetation Classes' Importance Ranking in Wildfire Susceptibility. Geosciences, 12 (11) p. 424. 

[2] Bustillo Sánchez M, Tonini M, Mapelli A, Fiorucci P, 2021. Spatial Assessment of Wildfires Susceptibility in Santa Cruz (Bolivia) Using Random Forest. Geosciences, 11 (5) p. 224. 

[3] Trucchia A, Meschi G, Fiorucci P, Provenzale A, Tonini M, Pernice U, 2023. Wildfire hazard mapping in the eastern Mediterranean landscape. International Journal of Wildland Fire. 32, 417-434. 

[4] Georganos S, Kalogirou S, 2022. A Forest of Forests: A Spatially Weighted and Computationally Efficient Formulation of Geographical Random Forests. ISPRS Int. J. Geo-Inf. 2022, 11, 471. 

How to cite: Tonini, M., Meschi, G., Trucchia, A., and Fiorucci, P.: Local feature importance of predisposing variables to explain the spatial heterogeneity of wildfires density in the Mediterranean area , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-9132, https://doi.org/10.5194/egusphere-egu24-9132, 2024.

X4.63
|
EGU24-6133
|
NH7.1
Marco Turco, John T. Abatzoglou, Sixto Herrera, Yizhou Zhuang, Sonia Jerez, Donal D. Lucas, Amir AghaKouchak, and Ivana Cvijanovich

This study delves into the increasing extension of summer forest fires in California, primarily driven by anthropogenic climate change (Turco et al. 2023). Historical data indicate a fivefold increase in summer burned area (BA) in forests in northern and central California from 1996 to 2021 relative to 1971 to 1995. Using the latest simulations developed for climate change attribution and detection studies and accounting for the uncertainties arising from the data-driven climate-fire model, climate models, and internal climate variability, we have investigated the impact of anthropogenic climate change on the observed increase in BA in California’s forests. We detect the signal of combined natural and anthropogenic forcing on the observed BA starting in 2001 while finding the observed BA changes to be inconsistent with internal variability or natural forcing alone. We estimate that climate simulations that included both human and natural forcings yield 172% more BA from 1971 to 2021 than models without anthropogenic forcing, with a remarkable +320% increase from 1996 to 2021. Considering the significance of anthropogenic climate change for the rise in forest BA in California, we pose a crucial question: what will the future of fires look like with ongoing climate changes? Addressing this, we evaluate how fuel limitations resulting from fire-fuel feedbacks might alter future fire trajectories under the influence of anthropogenic climate change. Dynamic models incorporating various feedback strengths suggest an expected further increase in annual average forest BA, ranging from 3 to 52% compared to the mean of the last two decades (2001-2021), which also marks the highest 20-year records since 1971. This highlights the imperative for proactive adaptation strategies. Our findings underscore the urgent need to address the impacts of climate change within fire management and policymaking.

How to cite: Turco, M., Abatzoglou, J. T., Herrera, S., Zhuang, Y., Jerez, S., Lucas, D. D., AghaKouchak, A., and Cvijanovich, I.: Nearly all of the increase in summer forest fires in California since 2001 is directly attributable to human-caused climate change, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-6133, https://doi.org/10.5194/egusphere-egu24-6133, 2024.

X4.64
|
EGU24-12598
|
NH7.1
|
ECS
|
Highlight
|
Linda Emily Weis, Michael Dietze, Anette Eltner, Daniel Schwindt, Kristina Beetz, Daniel Wolf, Annika Busse, and Elisabeth Dietze

Soil water repellency is a common phenomenon often associated with wild fires, which leads to a temporal change of forest ecosystems, for example, by enhanced overland flow, soil erosion and limited plant growth. Forest fires are expected to play an increasingly larger role due to climate change, resulting in more frequent droughts, higher temperatures, heatwaves, and landcover changes in temperate latitudes. Despite that importance, only a few studies have been published concerning soil water repellency in temperate European forests, and relatively little is known about soil hydrophobicity associated with so far rare forest fires in Central European spruce and beech forests.

In this study, we examine the impact of different burn severities on soil hydrophobicity down to 15 cm below the surface in the National Park “Sächsische Schweiz” after the forest fire in summer 2022, using the Water Drop Penetration Time (WDPT) test. Measurements were limited to the conductivity or non-conductivity of the water, with the test terminated at a time of 900 s. Various parameters, that could control water conductivity were examined, including burn severity from drone data, ground vegetation, duff layer, slope angle, slope aspect, and elevation of the site. In addition, soil properties such as soil type, carbon and nitrogen contents were analysed. We find a high spatial variability of hydrophobic plots in the studied area. The most hydrophobic plots were found in low severity sites rather than in moderate-high to high severity sites. Plots lacking a duff layer were more likely to exhibit hydrophobic layers. Soil water repellency was also found in unburnt sites. No distinct correlation was found between slope angle, slope aspect, elevation and the occurrence of hydrophobic plots. Plots located in coniferous forests exhibited higher frequencies of hydrophobicity compared to deciduous forests. That large variability and non-agreement with typically formulated relationships argue for a need to rethink the transferability of assumptions from traditional fire regions such as the Mediterranean or the boreal zone to the emerging fire regimes of temperate forests under climate change, requiring more empirical data.

How to cite: Weis, L. E., Dietze, M., Eltner, A., Schwindt, D., Beetz, K., Wolf, D., Busse, A., and Dietze, E.: Impact of burn severity of the July 2022 forest fire on soil hydrophobicity in the Elbe Sandstone temperate forest, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-12598, https://doi.org/10.5194/egusphere-egu24-12598, 2024.

X4.65
|
EGU24-18220
|
NH7.1
|
ECS
Carolina Gallo, Jonathan Eden, Bastien Dieppois, Peter Fulé, Jesús San-Miguel-Ayanz, Valentina Bacciu, Christophe Besacier, and Matthew Blackett

The Mediterranean region has historically been prone to wildfire activity. However, many Mediterranean countries have been particularly impacted in recent years by an increase in fire intensity and fire season length, with hundreds of thousands of hectares burned both north and south of the basin. Larger and more frequent fires are anticipated across the Mediterranean region in the future, a key driver of which is the projected increase in so-called fire weather (the meteorological conditions conducive to fire ignition and spread) associated with a warming world. In view of the loss or degradation of forest areas due to wildfires, and in the context of the ongoing UN Decade on Ecosystem Restoration (2021-2030), Mediterranean countries are actively engaging in post-fire restoration actions. Developing new insights into the evolution of fire weather across Mediterranean ecosystems is crucial for effective forest management and restoration planning.  

For the Mediterranean, fire weather projections under climate change have typically been extrapolated from global-scale studies or otherwise focused predominantly on Southern European countries. By contrast, far less attention has been given to countries in North Africa and the Middle East. Here, we generate high-resolution fire weather projections for the entire Mediterranean region, using the latest generation of global climate models. We calculate the Canadian Fire Weather Index (FWI) following a multivariate bias correction and downscaling of the FWI’s underpinning meteorological variables (namely, maximum daily temperature, minimum daily relative humidity, mean daily wind speed and daily precipitation). 

Results show changes in the magnitude of FWI seasonal means, maxima and fire season length in different scenarios and areas of the Mediterranean region where fire danger is projected to increase in the forthcoming decades. We discuss potential implications for future land management and restoration activities, as current preventive and restorative strategies should consider these future scenarios to ensure their success. The high-resolution fire weather projections generated here will help to better target areas of intervention and types of measures to be implemented.

How to cite: Gallo, C., Eden, J., Dieppois, B., Fulé, P., San-Miguel-Ayanz, J., Bacciu, V., Besacier, C., and Blackett, M.: High-resolution fire weather projections for effective forest management and restoration across the Mediterranean region , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-18220, https://doi.org/10.5194/egusphere-egu24-18220, 2024.

X4.66
|
EGU24-19169
|
NH7.1
|
ECS
|
Highlight
Robert Jackisch, Birgitta Putzenlechner, Simon Drollinger, and Elisabeth Dietze

Anthropogenic climate change increases the risk of forest fire following drought periods in temperate forests of Central Europe. Areas with an increased proportion of standing deadwood are often considered to be at risk. Especially in national parks, deadwood is not removed, forming an essential part of the local ecosystem.

In the Harz National Park, we aim at a comprehensive impact assessment following a fire in a spruce forest that was already disturbed after a massive bark beetle infection to understand deadwood breakdown, vegetation succession, surface erosion and changes in soil properties. The Quesenbank fire of August 2022 burned an area of approx. 13 ha within four days. We scanned 10 ha of burned compared to unburned areas using unoccupied aerial vehicles (UAVs) equipped with multispectral, thermal, high-resolution RGB and light-detection and ranging (LiDAR) sensors. Derived orthoimages, 3D point clouds and canopy height models (CHM) are employed to estimate standing deadwood, fractional cover and succession indicators, thermal ground regime alterations and small-scale morphological changes. To capture the gradual breakdown of deadwood, we collected ground truth on vegetation biophysical parameters, such as fractional cover, plant area index (PAI) and fraction of absorbed photosynthetically active radiation (FAPAR) from upward-directed digital hemispherical photos. The surveys were conducted 2, 9, 11 and 12 months post-fire together with the UAV campaigns in diffuse or near-dusk light conditions.

The analysis of the digital CHM and ground models reveal a decline in the detection rate of tree crowns (tree height ≥ 2 m) by 15 %, crown area by 74 %, and a corresponding loss of surface material affecting at least 0.9 ha between October 2022 and October 2023, respectively.

The ground reference data confirmed considerably lower fractional cover on burned areas. PAI and FAPAR in burned standing deadwood was lower in unburned stands, altering light, soil moisture and temperature regimes. This is reflected in the occurrence of typical post-fire and light-demanding species such as Epilobium spec. on burned areas, though in lower coverage compared to an unburned, logged site. As the variation in reference data was relatively low over the observation period, we suggest that the main dynamics of the breakdown of standing deadwood had already happened several weeks after the fire. Interestingly, we found a very heterogeneous microtopography due to granite boulders, with subsurface tunneling and unstable ground, influencing post-fire recovery. Upcoming analysis will include analyses of fire-influenced soil properties, morphodynamics and biogeochemical cycling in a region that still shows traces of past land use associated with the mining history of the Harz.


We acknowledge the collaboration with the Harz National Park Authority. A preliminary data set from two months after the fire can be accessed via Zenodo: Jackisch, R., Putzenlechner, B., & Dietze, E. (2023). UAV data of post fire dynamics, Quesenbank, Harz, 2022 (orthomosaics, topography, point clouds) (1.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.7554598

How to cite: Jackisch, R., Putzenlechner, B., Drollinger, S., and Dietze, E.: Quantifying post-fire effects and recovery in a disturbed landscape: Quesenbank fire, Harz National Park, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19169, https://doi.org/10.5194/egusphere-egu24-19169, 2024.

X4.67
|
EGU24-22445
|
NH7.1
Maria Prodromou, Stella Girtsou, George Leventis, Dimitris Koumoulidis, Marios Tzouvaras, Christodoulos Mettas, Alexis Apostolakis, Mariza Kaskara, Haris Kontoes, and Diofantos Hadjimitsis

This study presents the actions that are currently been conducted through a demonstration project in the framework of the EXCELSIOR funded project, entitled “Capitalizing on the ERATOSTHENES Data Cube to support the development of the Fire Risk Prediction Model” between the ERATOSTHENES Centre of Excellence and the National Observatory of Athens. Wildfires detection is a major issue for authorities. There are various causes of fire events with the most common being human influence. A fire risk prediction model through the analysis of geo-environmental and climate data is important for early warning and fire management. An effective wildfire risk prediction and management depend on the up-to-date, spatial explicit representation of the environment, mainly focusing on the biomass and characteristics of live and dead vegetation, which is the primary factor influencing fire behaviour and risk. In this work, a dataset from multiple modalities, including road density, travellers, forest-agriculture interface, burned areas from historical fire events, metrological data, land cover, vegetation indices from data cube, is generated. These factors are selected based on their potential correlation with the unique characteristics of the area investigated, the historical fire events, and the availability of relevant data. Artificial intelligence and machine learning models can use this multimodal dataset to improve forest fire management. Specifically, the combination of data cubes, machine learning, and geospatial ontology-based data access (OBDA) technologies, allows for effective harmonization of diverse data sources, enhancing the accuracy and efficiency of fire risk computations.


ACKNOWLEDGEMENT
The authors acknowledge the 'EXCELSIOR': ERATOSTHENES: Excellence Research Centre for Earth Surveillance and Space-Based Monitoring of the Environment H2020 Widespread Teaming project (www.excelsior2020.eu). The 'EXCELSIOR' project has received funding from the European Union's Horizon 2020 research and innovation programme under Grant Agreement No 857510, from the Government of the Republic of Cyprus through the Directorate General for the European Programmes, Coordination and Development and the Cyprus University of Technology.

How to cite: Prodromou, M., Girtsou, S., Leventis, G., Koumoulidis, D., Tzouvaras, M., Mettas, C., Apostolakis, A., Kaskara, M., Kontoes, H., and Hadjimitsis, D.: Creation of data cube for the analysis of wildfires in Cyprus using open access data , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-22445, https://doi.org/10.5194/egusphere-egu24-22445, 2024.

Fire behavior/modelling
X4.68
|
EGU24-16369
|
NH7.1
|
ECS
|
Highlight
Johanna San Pedro, Andrey Krasovskiy, Shelby Corning, Pavel Kiparisov, and Florian Kraxner

The increasing frequency of wildfires caused by climate change poses a significant threat globally, particularly in Latin America – a region known for its critical ecosystems. Its vulnerability to climate change-induced wildfire threats, resulting from increasing temperatures and changing precipitation patterns, is uncertain, highlighting the need for comprehensive strategies such as incorporating advanced modeling and proactive measures to understand, manage, and conserve its ecological state  in the face of  threats posed by climate change, such as wildfires. This study utilizes the wildFire cLimate impacts and Adaptation Model (FLAM) by IIASA to provide a comprehensive analysis of past and projected wildfire dynamics in Latin America. FLAM is a process-based fire parameterization algorithm used to assess the impacts of climate, fuel availability, topography, and anthropogenic factors on wildfire characteristics. It is highly adjustable and adaptable, making it suitable to analyze past and future wildfire trends in diverse regions such as Latin America. We analyzed spatial and temporal wildfire patterns using MODIS satellite data alongside historical climate and anthropogenic data to calibrate FLAM. We generated projections of burned areas until 2100 under 3 RCP scenarios for Latin American as a whole, as well as for distinct sub-regions to better assess regional wildfire dynamics and climate change impacts. Moreover, we developed a scenario to explore the impacts of increased fire suppression efficiency on projected burned area and highlight the impacts of focusing mitigation and management efforts on areas identified as hotspots (high risk of wildfire).

The study shows FLAM’s effectiveness in modeling historical wildfires and its sensitivity to the RCP scenarios in predicting wildfire trends in Latin America. Our analysis and results show how FLAM helps in evaluating the potential future changes in wildfire intensity, and geographic spread under various climatic scenarios.  FLAM projected a dramatic rise in burned area until the end of the century across Latin America in line with observed trends, especially under severe climate change scenarios. Regions with the highest temperature rises are also prone to reduced precipitation, which further increase  wildfire risks. The spatially-explicit projections highlight  areas at higher risk of wildfire, enabling targeted and efficient fire management and mitigation strategies. Our study further showed the potential impact of adaptive measures, such as enhanced fire suppression efficiency in identified hotspots, in reducing annual mean burned area. Overall, this study provides critical insights into the relationship between climate change and wildfire dynamics using a state of the art model. It sets the foundation for further research on fires in Latin America and efficient management strategies which can be modelled by FLAM.

How to cite: San Pedro, J., Krasovskiy, A., Corning, S., Kiparisov, P., and Kraxner, F.: Modeling Wildfire Dynamics in Latin America Using the FLAM Framework, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16369, https://doi.org/10.5194/egusphere-egu24-16369, 2024.

X4.69
|
EGU24-885
|
NH7.1
|
ECS
Krzysztof Szewczyk, Dominika Łuców, Boris Vanniere, Milena Obremska, and Michał Słowiński

Fire is one of the fundamental factors that governs and shapes the functioning of many ecosystems and directly influences changes in vegetation, biodiversity and ancient societies. Recent decades have shown how wildfires and their associated global relationships contribute to accelerate climate change through changes in vegetation, permafrost conditions, and the release of residues, greenhouse gases and aerosols. The last IPCC report shows that the European region is characterized by an increase in fires directly caused by current climate change. In addition, forecasts for the next century indicate a continuous increase in air temperature, which could lead to an increase in the frequency and intensity of wildfire events. In this regard, the questions arise: What are the social and environmental impacts of future unexpected climate events? And how could an increase in wildfires accelerate climate change?

During a fire, smoke, particles and various chemical compounds are released into the atmosphere, which can have a harmful effect on human health. Because of these consequences, it is important to understand what affects the occurrence and severity of fires. Paleofire reconstructions are useful for studying the effects of climate change and vegetation on fires at a time when human influence was less than today. Archived charcoal particles  in peat and lake sediments have been successfully used as geographic patterns in changing fire conditions. However, as many publications show, charcoal data can only provide partial estimates of changes in biomass burning. Therefore, the aim of the project is to indicate and verify the relationship between fire and its record in peat and lake sediments. To do this, burned areas within 40 km of 10 test sites (lakes and peatlands) are identified, and the intensity of each fire is estimated using fire data (i.e. fire type: ground, surface or crown), fire indicators (burnt area, weather conditions, wind speed and direction), fuel information (ecosystem type, forest age and species structure), obtained from the State Forests. Past fires and regional vegetation will be reconstructed using cores collected from lakes and peatlands based on pollen and charcoal analysis  with morphotypes in six fractions (100, 150, 200, 300, 400, 500 μm) with a high sampling resolution (0.5-1 cm). μ-XRF scanning will also be utilised to detect erosion and redeposition processes. The chronology will based on radiocarbon dating (AMS) and cesium-137 dating. Finally, the model of the spread of charcoal from the burnt area will be created. We assume that the amount of carbon accumulated on the lake and marsh surface is directly proportional to fire distance, burned area, and fire intensity. In this project, we want to advance the interpretation of reconstructed fires. This study could be the next step to better understand the fire signals preserved in our archives and improve the interpretation of paleo fires. This research is funded by the Polish National Science (No. 2023/49/N/ST10/04035).

How to cite: Szewczyk, K., Łuców, D., Vanniere, B., Obremska, M., and Słowiński, M.: The fire, burned areas and charcoal - charcoal-data modelling of burned areas, cross-validation of the fires and charcoal signal, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-885, https://doi.org/10.5194/egusphere-egu24-885, 2024.

X4.70
|
EGU24-769
|
NH7.1
|
ECS
|
Dina Jahanianfard, Joana Parente, Oscar González-Pelayo, and Akli Ait Benali

Wildfires have been known as one of the most disturbing phenomena in Portugal during last decades with increasing frequency, annual number of ignition and affected area. However, the extent of wildfire-induced changes on soil and vegetation, or burn severity, of these historical wildfires is unclear. To contribute to a better knowledge of post-fire impacts, this study presents a long-term burn severity atlas of historical wildfires in Portugal from 1984 to 2022 using satellite data.

Burn perimeters and start/end dates for large wildfires (>=100ha) were gathered and necessary corrections were manually applied on them. Due to the availability of satellite images, different imagery from Landsat sensors were used for different years: Landsat-5 (TM) for 1984 to 2011, Landsat-7 (ETM+) for 2002, and Landsat8 (OLI) for 2013 to 2022. The time lag between wildfire occurrence and satellite image acquisition dates was quantified and used to determine the suitability of each satellite image to estimate burn severity. Then, using Google Earth Engine API (JavaScript) and through a semi-automated process, the burn severity of each wildfire was calculated via difference normalized burn ratio (dNBR) derived indices (dNBR, relative dNBR (RdNBR), Relativized Burn Ratio (RBR), dNBR – Enhanced Vegetation index (dNBR-EVI)). These maps were created by the application of a pair of pre- and post-fire images with the highest suitability values.

The analysis performed on the time lag quantification showed a decrease in dNBR accuracy with the increase of both pre- and post-fire time lags. Over 3.7 million ha of land burned in Portugal from 1984 to 2022 in all vegetation types, around 3.2 million were associated with wildfires equal or larger than 100ha with known start and end dates (86.2%). Among these wildfires, 3.1 million ha had dNBR estimates (83.72% of all wildfires and 97.05% of wildfires>=100ha).

To the best of our knowledge, a long-term burn severity atlas has never been developed for an entire European country before. Another noteworthy advancement provided by this atlas is that the imageries from Landsat family of sensors were utilized for development of burn severity maps, offering the resolution of 30m over the manually corrected historical wildfire data (perimeters, locations, and dates). Also, a semi-automated process has been provided, equipped with the capacity to develop burn severity atlas for historical wildfires of any other region in the world with the prerequisite of wildfire data. Such datasets can be used by both scientific and management communities to improve current knowledge on post-fire impacts and develop better pre- and post-fire management plans to mitigate wildfire impacts. Moreover, this multidecadal burn severity dataset can be used by other research communities in the fields related to water, soil, and air quality which are potentially at risk due to wildfire occurrences.

Acknowledgements

We acknowledge CESAM by the Portuguese Foundation for Science and Technology FCT/MCTES (UIDP/50017/2020+UIDB/50017/2020+LA/P/0094/2020). D. Jahanianfard is supported by the Portuguese Foundation for Science and Technology (FCT-Fundação para a Ciência e Tecnologia) with a PhD grant reference (2021.08094.BD). O. Gonzalez-Pelayo further acknowledges FCT for the funding of FRISCO (PCIF/MPG/0044/2018) and SOILCOMBAT (PTDC/EAM-AMB/0474/2020) projects.

How to cite: Jahanianfard, D., Parente, J., González-Pelayo, O., and Ait Benali, A.: A multidecadal satellite-derived burn severity atlas for Portugal (1984 – 2022), EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-769, https://doi.org/10.5194/egusphere-egu24-769, 2024.

X4.71
|
EGU24-11261
|
NH7.1
|
ECS
Katrin Kuhnen, Maria Isabel Asensio, José Manuel Cascón, José Manuel Iglesias, Mariana Silva Andrade, Tatiana Klisho, Herbert Formayer, and Harald Vacik

Forest fires are becoming an important hazard in the mountain forests of European alpine areas due to changing environmental and socio-economic conditions. Understanding the fire behaviour is critical for all phases within the disaster management cycle – from prevention and preparedness to response and recovery. PhyFire is a simplified physical model which simulates the fire propagation and allows considering fire suppression measurements to see their effect on the fire behaviour. The model has been used for Mediterranean countries so far but has been now adapted to central European alpine mountain forests within this case study. The model requires the following input data: a digital height model, fuel data, and meteorological data. Besides the adaption process itself the effect of different qualities of input data in terms of temporal and spatial manner were investigated. In this contribution we show the challenges of the adaption process of the PhyFire model to the characteristics of the Austrian case study area and analyse the effect of different resolutions of input data resolution on the overall quality of the fire propagation simulation.

How to cite: Kuhnen, K., Asensio, M. I., Cascón, J. M., Iglesias, J. M., Andrade, M. S., Klisho, T., Formayer, H., and Vacik, H.: Forests on fire: Effects of input data resolution on forest fire behaviour modelling – A case study from Lower Austria, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-11261, https://doi.org/10.5194/egusphere-egu24-11261, 2024.

X4.72
|
EGU24-708
|
NH7.1
|
ECS
|
Luana Santos, Miguel M. Lima, Pedro M. M. Soares, Ricardo M. Trigo, and Rita M. Cardoso

Wildland fire spread and behaviour are complex phenomena owing to both the number of involved Physicochemical factors and the non-linear relationship between variables. In Portugal, one of the European countries most affected by wildfires, forest and bushfires occur every summer and are often exacerbated when extremely dry weather sets along with high temperatures. On the 17th of June 2017, an extreme heatwave associated with a severe drought and compounded by unusual levels of atmospheric instability led to a multiplicity of wildfires with many active fronts, and the formation of pyro-cumulus with explosive fire behaviour. All these factors contributed to the catastrophic fires that occurred in Pedrogão Grande on that day, with more than 100 fatalities and heavy impacts on livelihoods and assets.

The June 2017 extreme fire event in Pedrogão Grande is simulated with the WRF- Fire and Sfire model using a nested framework with increasing spatial resolution, including high-resolution regional scale (2km), local (0.4km) and Large Eddy Simulation (0.08km) resolutions. In this simulation 68 hybrid vertical levels are used, the model top is fixed as 20hPa, the first level is set at approximately 15m from the ground. Initial and boundary conditions for the outer domain were extracted from the ECMWF operational analyses, at 6-hourly intervals. Three microphysics schemes and three boundary layer parameterisations were employed to evaluate the best combination that suits robust reproduction of this complex event. The fire module is a simple 2D model of a surface fire, where the fire spreads through fuels on the ground. In every time step, the fire model inputs surface wind, which drives the fire, and outputs the heat flux from the fire into the atmosphere, which in turn influences the atmosphere. Among the different unusual features, we were particularly interested in assessing the model’s ability to reproduce a series of downbursts that occurred prior to and during the event and that have contributed decisively to atmospheric instability.

It was found that WRF can simulate those features, as well as the pyro-cumulus formation, yet their development is highly dependent on the interaction between the chosen microphysics and the boundary layer schemes. As in the observed event, the fire spread is accelerated westwards in association with the pyrocumulus. The initial simulated fire spread is faster than the observed in all simulations while the extent of the pyro-cumulus is shorter. The FWI (Fire Weather Index), the CHI (Continuous Haines Index) and the FWIe index (blending of FWI and CHI) were high prior and during the fire, observed in all domains, indicating extreme fire hazard and the presence of large instability conditions that can enhance fires that might become out of control, and with erratic behaviour.

Acknowledgements: This work was funded by the Portuguese Fundação para a Ciência e a Tecnologia (FCT) I.P./MCTES through national funds (PIDDAC) – UIDB/50019/2020. L.C. Santos is supported by the EarthSystems Doctoral School, at University of Lisbon, supported by FCT project UIDP/50019/2020-2023, University of Lisbon. M.M. Lima was supported through the PhD FCT programme grant PRT/BD/154680/2023.

How to cite: Santos, L., M. Lima, M., M. M. Soares, P., M. Trigo, R., and M. Cardoso, R.: High-Resolution Simulation of the Extreme Fire Event in Central Portugal, Pedrogão Grande (2017), EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-708, https://doi.org/10.5194/egusphere-egu24-708, 2024.

Fire danger/risk assessment
X4.73
|
EGU24-7211
|
NH7.1
|
ECS
Dasom Lee, Kwanchul Kim, Seong-min Kim, Jeong-Min Park, Gahye Lee, and Young J. Kim

Recently, several huge wildfires in South Korea caused record-breaking damage and casualties. In addition, wildfire occurrence and the amount of damages showed an increasing trend from the 1970s to the 2020s. Moreover, wildfires have become a major concern for the public and key ministries. Thus, the Korean government has been operating various wildfire observation systems using watch towers, CCTV, sensors, observers, etc. However, these systems have spatiotemporal and technical limitations such as short effective distances and discontinuous monitoring. Despite remarkable advances in wildfire detection, it remains challenging to early detect rapidly long-range wildfire events. Here, we developed equipment for early wildfire detection based on scanning Light Detection and Ranging (LiDAR) as a proof of concept to fill that void. Existing scanning LiDAR is used to track and monitor aerosol plumes providing multi-dimensional views of atmospheric layers. An early wildfire detection system using scanning LiDAR has improved for detection of wildfire smoke within 15 minutes with an enhanced spatial distance over a 10km radius and contained both eye safety function and trajectory tracking for point of ignition using HYSPLIT-based Emissions Inverse Modeling System for wildfires (HEIMS-fire).  We showed that the enhanced system continuously detected fire and smoke in rural areas during the day and night. The developed scanning LiDAR system can likely be used for early wildfire detection to prevent large-scale disasters.

Acknowledgement: This research was supported by a grant (2023-MOIS-20024324) of Ministry-Cooperation R&D Program of Disaster-Safety funded by Ministry of Interior and Safety (MOIS, Korea).

How to cite: Lee, D., Kim, K., Kim, S., Park, J.-M., Lee, G., and Kim, Y. J.: Development of wildfire detection and trajectory tracking system using scanning LiDAR, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-7211, https://doi.org/10.5194/egusphere-egu24-7211, 2024.

X4.74
|
EGU24-5196
|
NH7.1
|
Highlight
Francesca Di Giuseppe

The European Centre for Medium range weather forecast (ECMWF) on behalf of the Copernicus Emergency Management Service (CEMS) has recently widened the fire danger data offering in the Climate Data Store (CDS) to include a set of fire danger forecasts with lead times up to 7 months. The dataset incorporates fire danger indices for three different models developed in Canada, United States and Australia. The indices are calculated using ECMWF Seasonal Forecasting System 5 (SEAS5) and verified against the relevant reanalysis of fire danger based on the ECMWF Re-Analysis (ERA5). The data set is made openly available for the period 1981 to 2023 and will be updated regularly providing a resource to assess the  predictability of fire weather at the seasonal time scale. The data set complements the availability of seasonal forecast provided by the Copernicus Emergency Management Service in real time. 

A preliminary analysis shows that globally anomalous conditions for fire weather can be predicted with confidence 1 month ahead. In some regions the prediction can extend to 2 months ahead. In most situations beyond this horizon, forecasts do not show more skill than climatology. However an extended predictability window, up to 6-7 months ahead is possible when anomalous fire weather is the results of large scale phenomena such as the El Ni\~no Southern Oscillation and the Indian Ocean Dipole, often conducive of extensive fire burning in regions such as Indonesia and Australia.

How to cite: Di Giuseppe, F.: A new publically available dataset of global seasonal prediction of fire danger, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5196, https://doi.org/10.5194/egusphere-egu24-5196, 2024.

X4.75
|
EGU24-17032
|
NH7.1
Mafalda Canelas da Silva, Catarina Alonso, Rita Durão, and Célia M. Gouveia

The Mediterranean countries are largely affected by wildfires, and appropriate monitoring of daily fire danger is crucial to contribute to quick decision-making and mitigate destructive wildfire disturbances. In operational wildfire monitoring, the Canadian Forest Fire Weather Index System (CFFWIS) is one of the most used fire danger indices, particularly over the  Mediterranean region, which rates relative danger of wildfire occurrence by combining six components: three fire behavior indices and three fuel moisture codes.

Since FWI results from all components, it is evident that the short and long-term variations of meteorological variables will be reflected and the components will have different influences on FWI values. The main purpose of this work is to contribute to the definition of a new class associated with Exceptional fire weather danger, based on a statistical analysis of FWI, FFMC, and ISI indices for different Mediterranean countries, and for the months between June and October of 2010-2023 period and information of Fire Radiative Power, from the Land Surface Analysis Satellite Applications Facility project (LSA-SAF).

Results show, on one hand, that the extreme values of FWI (given by the 99th percentile) of the Mediterranean region are higher in Italy and Greece, in contrast with Portugal and Spain. On the other hand, regarding FFMC and ISI values, higher values can be seen in North African regions for FFMC, and in Italy and Greece for ISI. This is clear evidence of the variations in fire activity in the different Mediterranean regions. A new Exceptional class of fire danger is defined based on the extreme classes of FWI, FFMC, and ISI, jointly based on the occurrence of the recent Megafires in the Mediterranean region, such as the fires in Portugal in 2018 and Greece in 2023. The new approach to define the Exceptional class revealed to be an extremely important tool for fire danger assessment and for the definition of planning activities and suppression measures in the present context of climate warming.

Acknowledgements: This study is partially supported by the European Union’s Horizon 2020 research project FirEUrisk (Grant Agreement no. 101003890) and by the Portuguese Fundação para a Ciência e a Tecnologia (FCT) I.P./MCTES through national funds (PIDDAC) – UIDB/50019/2020- IDL and DHEFEUS - 2022.09185.PTDC.

How to cite: Canelas da Silva, M., Alonso, C., Durão, R., and M. Gouveia, C.: On the definition of an Exceptional fire danger rating over Mediterranean Countries, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17032, https://doi.org/10.5194/egusphere-egu24-17032, 2024.

X4.76
|
EGU24-9001
|
NH7.1
|
ECS
|
Highlight
Nicolò Perello, Andrea Trucchia, Giorgio Meschi, Mirko D'Andrea, Silvia Degli Esposti, and Paolo Fiorucci

The socio-economic changes over recent decades, marked by rural abandonment and fuel accumulation, coupled with the impact of climate change altering spatio-temporal weather patterns, have created conditions conducive to potential extreme wildfire events. Numerous wildfire management systems have thus faced significant challenges, leading to an additional push to develop or improve decision-support tools. Forest Fire Danger Rating models have been widely used by wildfire management systems in recent decades, aiding in daily operations planning and the production of fire bulletins.

Since the year 2000, independent research programs conducted by the Liguria Region in Italy, and subsequently by the Italian Civil Protection, have led to the creation of the Forest Fire Danger Rating system known as RISICO. The system incorporates meteorological observations and forecasts from Limited Area Models, utilizing vegetation cover and topography as additional inputs to enhance its capabilities. The system is currently adopted at the national level in Italy by the Civil Protection system (Dipartimento della Protezione Civile), supporting the production of the national daily fire danger bulletin, and by several regional authorities.

Over the past year, significant efforts have been made to upgrade the model. Specifically, a new fuel map based on fire susceptibility obtained through Machine Learning techniques has been proposed. This new approach allows for the structured integration of wildfire susceptibility information within the assessment of wildfire danger. Given the importance RISICO places on information about fuel classes, this approach allows a focus on fuel conditions that, when combined with specific meteorological conditions, can lead to extreme wildfire events. Furthermore, the Fine Fuel Moisture component of RISICO has been modified in its dynamics and calibrated using observed data from fuel sticks. This modification aims to better identify prolonged conditions of dry fuel that facilitate the ignition and spread of fires. Finally, the Rate of Spread model has been enhanced through the integration of the PROPAGATOR wildfire spread model's approach, with the goal of providing a more accurate description of the interaction between wind and topography.

The updated model was subsequently validated using fires that occurred in Italy from 2007 to 2022 and compared with the model's performance before the modifications. The results demonstrate an improvement in the model's ability to identify situations particularly dangerous for fire ignition and spread. The updated model, therefore, enhances the prediction of wildfire danger, providing scientific support in the decision-making process and promoting effective wildfire management.

How to cite: Perello, N., Trucchia, A., Meschi, G., D'Andrea, M., Degli Esposti, S., and Fiorucci, P.: Unveiling RISICO 2024: Enhancing Wildfire Forecasting through Cutting-Edge Updates, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-9001, https://doi.org/10.5194/egusphere-egu24-9001, 2024.

X4.77
|
EGU24-19064
|
NH7.1
|
ECS
Minwoo Roh, Sujong Lee, and Woo-kyun Lee

Forest fires exert a significant impact on Earth's ecological systems, resulting in consequences such as deforestation, habitat degradation, and adverse effects on environmental, economic, and social domains. The restoration of areas affected by forest fires demands substantial time and effort to return them to their original state. Proactive identification of areas prone to forest fires is crucial for minimizing the damage caused by such incidents. In this study, a forest fire diagnostic model was developed to enhance the precision of forest fire risk predictions. The model utilized remote sensing data and human activity maps. To gauge the dryness of the land surface, the Vegetation Temperature Condition Index (VTCI) was employed, and density maps of roads, buildings, and cropland were incorporated for the human activity maps. The algorithm of the model was based on the Random Forest classifier, and it was trained on forest fire occurrence data from 2016 to 2020 across South Korea. To assess the actual performance of forest fire forecasting, short-term forecasts for a 3-day period were conducted from February to May 2023. The model successfully predicted 80% of forest fires during this evaluation period.

How to cite: Roh, M., Lee, S., and Lee, W.: Forecasting Forest Fires in South Korea Using a Diagnostic Forest Fire Model, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19064, https://doi.org/10.5194/egusphere-egu24-19064, 2024.

X4.78
|
EGU24-4332
|
NH7.1
|
ECS
Tina Trautmann and Petra Doell

Over the last decades, the frequency and magnitude of wildfires increased worldwide, posing a risk not only for people and infrastructure, but also for the environment. To enable preventive and protective measures it is crucial to monitor and forecast wildfire hazards. Existing systems derive wildfire indices from meteorological data or include remote-sensing-based observations, such as soil moisture anomalies and the state of vegetation health. However, their ability to forecast wildfire hazards into the future is very limited.

Therefore, we here suggest utilizing the potential of a global hydrological model to not only monitor, but also provide seasonal forecasts of wildfire hazards globally. To do so, we force the global water resources and use model WaterGAP by meteorological data from ERA5 reanalysis and SEAS5 seasonal ensemble forecasts. Model output, including for example soil moisture anomalies, are combined with meteorological data to derive indicators for wildfire hazards at a global scale. We assess the capability of such indicators to reflect spatio-temporal pattern of wildfire hazards during the year 2018 by performing a regional analysis.

Eventually, derived wildfire hazard indicators can be made available for stakeholders via the operational multi-sectoral global drought monitoring and seasonal forecasting system (OUTLAST) on WMO’s HydroSOS web portal.

How to cite: Trautmann, T. and Doell, P.: Towards global monitoring and seasonal forecasting of wildfire hazards based on a global hydrological model, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-4332, https://doi.org/10.5194/egusphere-egu24-4332, 2024.

X4.79
|
EGU24-14409
|
NH7.1
Kwanchul Kim, Dasom Lee, Seong-min Kim, Gahye Lee, Jeong-Min Park, Youngmin Noh, Young J. Kim, Kwon-ho Lee, Sungchul Choi, Changgi Choi, Woosuk Choi, and Chunsang Hong

Wildfires are increasing globally due to climate change. Wildfires can spread rapidly in a short period of time, early detection is important. A CCTV and thermal imaging camera are used for early detection and prevention of wildfires and improved using camera analysis method and AI technology. However, the wildfire detection distance is still shortened depending on weather conditions and air quality, and image processing performance deteriorates due to low light at night. A satellite remote sensing technology is difficult to monitor wild fires in real time and affected by cloud mask, low spatial and temporal resolution. A Drone is also limited in flight time by communication, weather conditions, battery capacity, and payload. In the case of lidar-based long-distanced wildfire monitoring, advanced remote sensing monitoring that can monitor wildfires is possible by classifying the type of aerosol particles and the amount of light backscattered by smoke particles. Our recently developed wildfire scanning lidar technology uses light sources of two wavelengths (532 nm and 1064 nm) and developed a system capable of 360° observation within 30 minutes with an angular resolution of less than 1° in the horizontal direction. In addition, it is the wildfire scanning lidar capable of detectiong a wildfire in the atmosphere using the backscattering coefficient and aerosol optical properties calculated at two wavelengths. A depolarization of smoke aerosol in the air can be used to improve the accuracy of wildfire smoke detection using characterization of particles. Presently wildfire monitoring lidar technology under development is producing commercial products that protect eyesight and monitor forest fire smoke within a radius of 10 km through long-wavelength laser and object analysis.

 Acknowledgement: This research was supported by a grant (2023-MOIS-20024324) of Ministry-Cooperation R&D Program of Disaster-Safety funded by Ministry of Interior and Safety (MOIS, Korea).

How to cite: Kim, K., Lee, D., Kim, S., Lee, G., Park, J.-M., Noh, Y., Kim, Y. J., Lee, K., Choi, S., Choi, C., Choi, W., and Hong, C.: Development of early wildfire detecting system using scanning lidar and image processing, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-14409, https://doi.org/10.5194/egusphere-egu24-14409, 2024.

X4.80
|
EGU24-16616
|
NH7.1
|
ECS
|
Highlight
Mislav Anić, Maša Zorana Ostrogović Sever, Doroteja Bitunjac, and Hrvoje Marjanović

Fire represents one of the major disturbances in natural ecosystems around the planet. By burning ecosystems, fire has a significant role in shaping global biome distribution and influences biogeochemical cycles such as the carbon cycle. Particularly in the coastal part of Croatia during the summer months, wildfires often escalate to catastrophic levels, posing serious threats to human lives, infrastructure, and the natural environment.

Our study aims to estimate the changes in forest fire danger in Croatia between 1981 and 2020 based on the Fire Weather Index (FWI) and seasonal severity rating (SSR) calculated using data from the National Meteorological Observation Network. To estimate the risk of forest fires in this region, the study utilizes the Canadian Forest Fire Weather Index System. The original system consists of six components that solely depend on meteorological conditions. The calculated FWI represents the potential fire intensity and is a very good indicator of fire danger. The initial equations are calibrated for Canadian boreal forests, characterized by distinct differences in vegetation and climate features when compared to forests of the Mediterranean regions. Despite these differences, researches have revealed a noteworthy correlation between the components of the FWI system and fire activity in Spain, Portugal, France, Italy, and Greece.

Measurements of air temperature, wind speed, and relative humidity taken at 14h, along with daily precipitation records from 83 meteorological and 119 rain gauge stations were used in the analysis. Daily severity ratings were calculated from FWI values and averaged over the fire season, spanning from June to September, to obtain SSR. The station-based SSR were spatially interpolated using regression kriging and a 1 x 1 km horizontal grid, resulting in 40 raster maps (one for each year). Results from a trend analysis, aimed at identifying areas with the highest increase in fire risk during the period 1981-2020, indicate an overall increase in SSR across a significant portion of the country. The observed trends align well with the positive trends identified in maximum air temperature and the lengthening of dry periods.

Additionally, to evaluate changes in fire weather extremes in Croatia the seasonal count of days with FWI > 30 (FWI30) and the seasonal 90th percentile of FWI (FWIp90) indices were calculated. A comparison of these indices between the periods 1981-2000 and 2001-2020 revealed an increase both in FWI30 and FWIp90 across a substantial portion of the country. These trends highlight a concerning escalation in fire risk for Croatia.

Keywords: forest fire risk, fire weather index, extreme fire weather, climate change, trends

How to cite: Anić, M., Ostrogović Sever, M. Z., Bitunjac, D., and Marjanović, H.: Forest fire risk under a changing climate in Croatia, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16616, https://doi.org/10.5194/egusphere-egu24-16616, 2024.