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 ParenteECSECS | Co-conveners: Mário Pereira, Nikos Koutsias, Andrea Trucchia, Marj Tonini
Orals
| Tue, 25 Apr, 08:30–12:30 (CEST)
 
Room 1.31/32
Posters on site
| Attendance Tue, 25 Apr, 16:15–18:00 (CEST)
 
Hall X4
Posters virtual
| Attendance Tue, 25 Apr, 16:15–18:00 (CEST)
 
vHall NH
Orals |
Tue, 08:30
Tue, 16:15
Tue, 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 governments to develop spatio-temporal datasets and to produce risk and prognostic maps. A key tool in this respect is to investigate the spatial and temporal distribution of wildfires and to understand their relationships with the surrounding environmental, climatological and socio-economic factors.
Innovative algorithms and methodologies developed in the geocomputational science field have proved to be useful in analysing spatially 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, especially for large-scale analysis. A new exciting challenge is to convert available datasets into meaningful and valuable information.
This session aims to bring together wildfire 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 wildfire 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: wildfire incidence mapping and spatial distribution; wildfire severity and damages; wildfire risk management;
• long-term wildfires patterns and trends: relation between wildfires and global changes such as climate and land use/ land cover changes;
• wildfire spread models, ranging from case studies to long-term climatological assessments;
• post-fire vegetation recovery and vegetation phenology.

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 governments to develop spatio-temporal datasets and to produce risk and prognostic maps. A key tool in this respect is to investigate the spatial and temporal distribution of wildfires and to understand their relationships with the surrounding environmental, climatological and socio-economic factors.
In this session, we aim to improve the understanding of the fire regime and to promote new strategies to mitigate the disastrous effects of wildfires. We welcome all the interesting and relevant oral and poster presentations we received, and we are glad to announce you two solicited authors’ work on fire management and on impacts of wildfires.  
We hope that our session attracts wildfire scientists, researchers of various geo-environmental disciplines, economists, managers, people responsible for territorial and urban planning, and policymakers. And that together we can contribute with knowledge in a world where regions are burning with epic breadth and intensity.

Orals: Tue, 25 Apr | Room 1.31/32

Chairpersons: Marj Tonini, Andrea Trucchia, Joana Parente
08:30–08:33
Wildfire mapping and main drivers (convener: Marj Tonini + Joana Parente)
08:33–08:43
|
EGU23-1317
|
On-site presentation
M. Lucrecia Pettinari, Joshua Lizundia-Loiola, Amin Khairoun, Ekhi Roteta, Thomas Storm, Martin Boettcher, Olaf Danne, Carsten Brockmann, and Emilio Chuvieco

The FireCCI project, as part of the ESA Climate Change Initiative (CCI), has developed and validated burned area (BA) algorithms and products with the objective to meet, as far as possible, GCOS (Global Climate Observing System) Essential Climate Variable requirements for global satellite data products from multi-sensor data archives.

The current suite of global products include FireCCI51, whose algorithm uses as input MODIS NIR surface reflectance at 250 m and 1-km-resolution active fires, and currently covers a 20-year time series. An evolution of this algorithm uses the SWIR bands of the Sentinel-3 SLSTR sensor, provided at 300 m resolution by the Synergy products developed by ESA. This input is complemented by VIIRS active fire information at 375 m resolution. The resulting BA product, called FireCCIS310, takes advantage of the improved BA detection capacity of the SWIR bands and the higher resolution of the VIIRS thermal information, apart from upgrades in the algorithm itself. This product is currently available for 2019, and it is being further processed for the subsequent years. FireCCIS310 is capable of detecting 28% more BA than FireCCI51 for the same year.

Complementary, a specific dataset has been created for sub-Saharan Africa, where more than 70% of the total global burned area occurs. This product, called FireCCISFD (SFD standing for Small Fire Dataset), uses surface reflectance from the Sentinel-2 MSI sensor at 20 m spatial resolution, supplemented by active fire information. Version 1.1 of this dataset (FireCCISFD11) covers the year 2016 and is based on Sentinel-2A data plus MODIS active fires, while the newer version (FireCCISFD20) has been processed for the year 2019, and takes advantage of the additional data provided by Sentinel-2B, duplicating the input data amount and temporal resolution, and the improved spatial resolution of VIIRS active fires. Due to the much higher spatial resolution of the input data, this product detected 58% more BA than FireCCI51 in 2016, and 82% more in 2019, mostly due to the enhanced detection of small burned patches, not detectable with coarser resolution sensors.

All these datasets provide very valuable information regarding land cover change dynamics due to fires, and their associated aerosol and greenhouse gasses emitted to the atmosphere. Particularly, the SFD datasets show that current estimations of fire emissions have been underestimated, and that they should be re-assessed taking into account the capabilities of the information provided by medium to high-resolution sensors.

How to cite: Pettinari, M. L., Lizundia-Loiola, J., Khairoun, A., Roteta, E., Storm, T., Boettcher, M., Danne, O., Brockmann, C., and Chuvieco, E.: Global and continental burned area detection from remote sensing: the FireCCI products, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1317, https://doi.org/10.5194/egusphere-egu23-1317, 2023.

08:43–08:53
|
EGU23-8373
|
ECS
|
On-site presentation
Shuhan Lou, Magí Franquesa, Yuqi Bai, and Emilio Chuvieco

Previous studies have found that global burned area products based on coarse resolution sensors tend to miss small fires (<100 ha), particularly in those regions where burnings tend to be human-controlled.

We aimed to analyse the accuracy of small fire detection of current global burned area products in China. Most regions in China tend to have small fires due to strict forest and grassland fire suppression policies, as well as straw burning ban policies. According to Chinese government statistics from 2010 to 2019, there are more than 2000 forest fires every year, with an average size of 10.9 ha per fire. However, most of the studies on fire regimes and fire impacts in China used global burned area products with relatively coarser spatial resolutions, most likely leading to considerable uncertainty in the results.

Our assessment included burned area products with coarse spatial resolution sensors (MODIS, Sentinel-3) with relatively high temporal resolution (1 day) and those with medium resolution sensors (Landsat, Sentinel-2) with relatively low temporal resolution (> 5 days). A burned area reference dataset from Sentinel-2 (S2) images was built to validate those burned area products in China. The extent to which the difference in the spatial and temporal resolutions affecting the total burned area was measured based on the spatial and temporal intercomparison of burned area products.

Accuracy metrics, including both omission (undetected burned pixels) and commission errors (unburned pixels classified as burned), were used in this study. The preliminary results indicate that the ability of global burned area products to detect small fires needs to be improved. This work imposes the essential requirement of relatively high spatial and temporal resolution burned area products for small fires.

How to cite: Lou, S., Franquesa, M., Bai, Y., and Chuvieco, E.: Preliminary assessment of burned area products in the detection of small fires: A case study of China, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8373, https://doi.org/10.5194/egusphere-egu23-8373, 2023.

08:53–09:03
|
EGU23-11914
|
ECS
|
On-site presentation
Angel Caroline Johnsy, Roberto Llop Cardenal, Penelope Kourkouli, and Qiaoping Zhang

There has been a significant increase in wildfires in recent years due to climate change, which is having an increasing impact on human settlements, infrastructure, buildings and the environment [1]. The statistics published by the California Department of Forestry and Fire Protection states that the area affected by fires in 2021 (2.6 million acres) is approximately 47% higher than the average over the last five years of 1.6 million acres [2].  Mitigating and preventing wildland fires is a crucial task that involves strategic planning and detailed monitoring of fire activity.

Many remote sensing techniques have been established to aid emergency responders  with the immediate planning and actions to be taken in the event of wildfire. Synthetic Aperture Radar’s (SAR) capability of penetrating the clouds and smoke offers a unique advantage for monitoring the progression of wildfire, which can be quite challenging to contain [3]. The capability of coherent change detection applied to interferometric SAR images for describing changes associated with wildfire is presented here. The coherence from interferometric SAR has been shown in many case studies to be an effective tool for monitoring deforestation and forest fires [4].  An increase in time interval between images leads to increased variation in the scatterer’s distribution which results in lower coherence values, which can reduce the usefulness of this method. The value of coherence is much more evident in forest areas when a shorter interval between subsequent images can be utilized. .

The Rum Creek fire burned 21,347 acres in southwest Oregon, USA, which was ignited by lightning strikes on August 17, 2022 and continued for nearly a month. Two ICEYE stripmap SAR images (August 26, 2022 and August 28, 2022) captured the scene during the fire event and demonstrated the effectiveness of using interferometric SAR methods in detecting the burn scars. The results show good agreement with the fire perimeter released by the Wildland Fire Interagency Geospatial Services (WFIGS) Group [5].

 

References:

   [1] A. Borunda, “The science connecting wildfires to climate change”, https://www.nationalgeographic.com/science/article/climate-change-increases-risk-fires-western-us

   [2] California Department of Forestry and Fire Protection  [online]  https://www.fire.ca.gov/stats-events/

  [3] A. Moreira, P. Prats-Iraola, M. Younis, G. Krieger, I. Hajnsek and K. P. Papathanassiou, "A tutorial on synthetic aperture radar," in IEEE Geoscience and Remote Sensing Magazine, vol. 1, no. 1, pp. 6-43, March 2013, doi: 10.1109/MGRS.2013.2248301.

  [4] S. Takeuchi and S. Yamada, "Monitoring of forest fire damage by using JERS-1 InSAR," IEEE International Geoscience and Remote Sensing Symposium, 2002, pp. 3290-3292 vol.6, doi: 10.1109/IGARSS.2002.1027159.

 [5] The Wildland Fire Interagency Geospatial Services [online] https://data-nifc.opendata.arcgis.com/datasets/nifc::wfigs-current-wildland-fire-perimeters/about

How to cite: Johnsy, A. C., Cardenal, R. L., Kourkouli, P., and Zhang, Q.: Wildfire mapping with Interferometric ICEYE SAR data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11914, https://doi.org/10.5194/egusphere-egu23-11914, 2023.

09:03–09:13
|
EGU23-11882
|
ECS
|
On-site presentation
Aleksandra Kolanek and Mariusz Szymanowski

Forest fires in Poland are a serious disturbance of forest ecosystems, generating large natural, social and economic losses. For this reason, it is very important to identify and quantify the factors that may affect the occurrence of fires in the forest environment.

In this paper, it was hypothesized that the structure of the forest landscape significantly affects the occurrence of forest fire. In order to verify this thesis, basic landscape metrics were calculated, reflecting the features of the structure of forest areas in Poland (number of patches, average and median area of patches, length and density of patch edges, patch shape) in relation to three characteristics of forest stands: age class, moisture type and trophic type (data from Poland National State Forest). The analyzes were carried out both at the landscape level and at the level of selected classes. Two approaches were tested to statistically test the impact of landscape structure on forest fires:

1. The metrics were calculated with reference to the test set (A), which was the areas of buffer zones with a radius of 1000 m established around the occurrences of fires from 2015 in one of the Polish voivodeships (Lubelskie voivodeship), registered by the National Forest Fire Information System (NFFIS). The control sets were buffer zones of the same radius, established around randomly placed points: allowing (set B) and not allowing (set C) the overlapping of buffer zones. It was assumed that finding a statistically significant difference between set A and sets B or C allows the acceptance of a research hypothesis in relation to the features of the landscape structure expressed by the analyzed metrics. Therefore, statistical analysis was performed using the Kruskal-Wallis test and the Wilcoxon test for paired observations.

2. The indicators were calculated in relation to the test set (A), which was the areas of buffer zones with a radius of 1000 m established around the occurrences of fires from 2007-2017 in Poland registered by the National Forest Fire Information System (NFFIS). The control set consisted of buffer zones of identical radius established around randomly selected points (selected by stratified random sampling). It was assumed that finding a statistically significant difference between the two groups allows for the positive verification of a research hypothesis in relation to the feature of the landscape structure expressed by the analyzed indicator. For this purpose, statistical analysis was performed using the Mann-Whitney U test.

Preliminary results allowed to identify statistically significant differences between the features of the forest structure in the vicinity of fire sites and the surroundings of control groups, regardless of the method of selection of control groups. This may indicate the important role of the forest structure in shaping the fire risk. Further studies are planned to confirm this conclusion. The research is carried out as part of the PRELUDIUM project no. 2019/35/N/ST10/00279, funding by the National Science Centre Poland.

How to cite: Kolanek, A. and Szymanowski, M.: The structure of the forest landscape - a potential determinant of forest fires?, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-11882, https://doi.org/10.5194/egusphere-egu23-11882, 2023.

09:13–09:23
|
EGU23-15247
|
On-site presentation
José Maria Costa-Saura, Valentina Bacciu, Donatella Spano, Pierpaolo Duce, Giovanni Sanesi, and Mario Elia

Heat waves (HWs) and wildfires are potentially considered highly correlated hazards with dramatic impacts on ecosystems and society. However, previous studies focused only on one single hazard leaving out potential compounding and cascading effects. The aim of this work is thus to investigate spatio-temporal patterns for both hazards in Mediterranean Europe and assess the potential influence of HWs on fire regime.

The historical wildfire dataset was derived from the Global Wildfire Information System (GWIS). This dataset corresponds to georeferenced polygons detected between 2000 and 2021 by MODIS. To understand the relationship between fires and HWs, we extracted several fire regime metrics accounting for density, seasonality and severity. All the metrics were computed at the scale of 12 km square grid.

Then, we characterised the HWs in terms of frequency, duration, seasonality and intensity for the same period based on a high resolution meteorological dataset (ERA5-land from Copernicus Climate Data Store). A HW was defined as a period of at least 6 days with temperatures greater than the 90th percentile whereas intensity was calculated as the total amount of degrees exceeding that percentile during the HW. Finally, we compared HWs occurrence with information about fire events. Compound and cascading events were analysed through hazard maps to identify simultaneous occurrences of the two hazards and look at different combinations of hazard sequences. Finally, multiple linear regression models were applied to explore the complex influence of HWs characteristics on fire metrics.

The results highlighted the hotspot areas in Mediterranean countries where the common hazard occurrences were identified, and the role of HWs in the compound and cascading fire hazard events. The findings of this study could support the assessment of hazard patterns and, in turn, prevention and monitoring activities to support disaster risk reduction.

How to cite: Costa-Saura, J. M., Bacciu, V., Spano, D., Duce, P., Sanesi, G., and Elia, M.: Exploring the influence of heat waves on wildfire occurrence in Mediterranean countries of southern Europe, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15247, https://doi.org/10.5194/egusphere-egu23-15247, 2023.

09:23–09:33
|
EGU23-405
|
ECS
|
On-site presentation
Patrícia Páscoa, Célia Gouveia, Ana Russo, and Andreia Ribeiro

The occurrence of large bushfires in southeastern Australia has been linked to the occurrence of extreme climate conditions, such as droughts and extreme temperatures. Several extreme bushfires have occurred following or during severe droughts and heatwaves, namely the Black Saturday bushfires in 2009 and the Black Summer of 2019-2020. Fire-prone weather conditions have become more severe in this region during the past years and are expected to worsen as the frequency of compound drought and extreme temperature events is expected to increase in the future, even under lower emission scenarios. This is particularly important as the impacts of compound climate events are usually larger when compared to the impacts resulting from an individual event. Therefore, compound drought and temperature extremes are likely to increase the probability of large bushfires, compared to the occurrence of an individual event.

In this work, the trivariate relationship between burned area in the months of December to February, drought conditions, and temperature extremes was analyzed using copulas for the period 2001-2020. Burned area across forests were computed using the GlobFire dataset. Drought conditions were assessed using the Standardized Precipitation Evapotranspiration Index (SPEI), computed with monthly precipitation and temperature data from the CRU TS4.05 dataset. The indices Number of Hot Days (NHD) and Number of Hot Nights (NHN) were used to identify conditions of extreme temperature and were computed using hourly temperature data from ERA5. The influence of concurrent and previous climate conditions on the burned area was assessed, for up to 3 months before the beginning of the fire event.

The results show a clear influence on the probability of occurrence of large fires under conditions of drought and extreme temperature. Drought conditions in the months before the fire event had a larger effect than temperature extremes. Moreover, the probability of occurrence of large fires is higher when compound drought and hot events are present than given only one individual extreme event.

 

Acknowledgements: This study was supported by the H2020 FirEUrisk project (EU H2020, Grant Agreement 101003890) and by FCT (Fundação para a Ciência e Tecnologia, Portugal) through national funds (PIDDAC) – UIDB/50019/2020 and project Floresta Limpa (PCIF/MOG/0161/2019).

How to cite: Páscoa, P., Gouveia, C., Russo, A., and Ribeiro, A.: Large bushfires under drought and extreme temperature conditions in southeastern Australia: a probabilistic assessment using copulas, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-405, https://doi.org/10.5194/egusphere-egu23-405, 2023.

09:33–09:43
|
EGU23-16160
|
ECS
|
On-site presentation
Bruno Barbosa, Ana Gonçalves, Sandra Oliveira, Jorge Rocha, and Mario Caetano

Wildfires occur unevenly in the territory, driven by different biophysical and social factors. Understanding the spatial and temporal distribution of wildfires can help identifying common characteristics and/or dissimilarities between regions. In this research, we use specific fire metrics, from historical fire data, to explore the possibility to identify groups of municipalities based on their pyrosimilarities. We apply a clustering model based on the method k means to identify and compare groups of municipalities (defining pyroregions) of mainland Portugal (n=277), using fire data from the last 22 years (between 2000 and 2021). The fire metrics used were: (a) cumulative percentage of total burned area, (b) cumulative percentage of burned area in the summer months, (c) mean annual number of fires and (d) GINI index applied for burned area over time. We used tools available in Geographic Information Systems (ArcGIS Pro) linked with python programming, to apply the cluster method and map the results. Our preliminary results divided the mainland in 5 clusters. CL1 (n=66) is seen in the west coast and is characterised by a burned area concentrated in a few years (high Gini index), but the fires occur mainly outside the summer months; CL2 (n=50) cover municipalities in the northeast and is characterised by a high mean number of fires dispersed over the years (low GINI index); CL3 (n=26) located in the central Portugal has a high percentage of cumulative burned area throughout the years, but with low number of fires, concentrated in time; CL4 (n=63) covers the municipalities in the southwest and south and shows a low mean number of fires but  these occur mainly in the summer season and, CL5 (n=55) appears throughout the country, but is more concentrated in the west and is characterised by intermediate values in all analysed metrics. The extreme wildfires that occurred in 2017 in Portugal influence the clustering; for example, CL1 occurs on the west central coast, where the consolidated maritime pine forest has burned in October 2017, outside the summer months. The next steps of this analysis are: (i) apply other clustering methods to compare with these clusters identified with k means and their characteristics and (ii) analyse the explanatory variables that influence these fire patterns.

 

This work was funded by FCT, I.P.: BB and AG in the scope of PhD projects [2022.12095.BD], [2020.07651.BD], SO under the contract ‘2020.03873.CEECIND′, Centre for Geographical Studies—University of Lisbon and FCT under Grant number [UIDB/00295/2020 + UIDP/ 00295/2020].

How to cite: Barbosa, B., Gonçalves, A., Oliveira, S., Rocha, J., and Caetano, M.: Identifying pyroregions in mainland Portugal with clustering methods using GIS, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-16160, https://doi.org/10.5194/egusphere-egu23-16160, 2023.

09:43–09:45
09:45–09:55
|
EGU23-14472
|
ECS
|
solicited
|
Highlight
|
On-site presentation
Sílvia A. Nunes, Carlos C. DaCamara, Ricardo M. Trigo, Isabel F. Trigo, Célia M. Gouveia, Renata Libonati, and Liz B. C. Belém

Forest fires are a global phenomenon with severe and destructive impacts at the ecological, environmental, socio-economic and health levels. Portugal, like all Mediterranean Europe, is recurrently affected by large wildfires, particularly intensive during the summers of 2003, 2005 and 2017. The latter is remembered because of the 500 000 hectares of burned area and the at least 116 deaths in the fire season.

CeaseFire is a website that aims at converting the scientific knowledge produced at universities and research institutes into useful and user-friendly tools that provide information on fire activity and meteorological fire danger tailored to the needs of the fire community in Portugal and assist in decision making on fire management and combat and on fire damage mitigation. The website, relying on information from the LSA SAF project, provides maps and data of (1) components of the Fire Weather Index System;(2) Fire Radiative Power (FRP) released by wildfires (3) classes of meteorological fire danger, ignition potential and aftermath; (4) outlooks of the fire season severity; and (5) prescribed burned classes. The site has been sponsored by The Navigator Company, a leading force in the global pulp and paper market since the operational start of the website in 2016. The number of registered users has increased up to more than 1 600 within the fire community, comprising firemen, civil protection officers, municipalities, academic researchers and private owners. In recent years E-Redes, a national electricity distribution utility service, has also been supporting the development of the website.

CeaseFire is sufficiently flexible to include other tools and be implemented in other regions. In this context, the feedback provided by users and companies has been decisive to improve the tools and products, and extend the site to other geographical areas, namely the Zambezia province in Mozambique and, more recently, the Brazilian Pantanal.

Research work was funded by the Portuguese Fundação para a Ciência e a Tecnologia (FCT) I.P./MCTES through national funds (PIDDAC) IDL (UIDB/50019/2020), project FIRECAST (PCIF/GRF/0204/2017) and by EUMETSAT Satellite Application Facility on Land Surface Analysis (LSA SAF).

How to cite: Nunes, S. A., DaCamara, C. C., Trigo, R. M., Trigo, I. F., Gouveia, C. M., Libonati, R., and Belém, L. B. C.: CeaseFire: from scientific knowledge to operational tools for fire combat and management, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14472, https://doi.org/10.5194/egusphere-egu23-14472, 2023.

09:55–10:05
|
EGU23-4619
|
ECS
|
On-site presentation
Douglas Radford, Holger Maier, Hedwig van Delden, Aaron Zecchin, and Amelie Jeanneau

Wildfires can be dangerous phenomena, creating risks for communities that are likely to be exposed to wildfire. The likelihood of community exposure to a wildfire is influenced by the interaction of fire behaviour factors (weather, fuel and topography) across multiple spatial scales.

Our objective is to develop an index that measures the connectivity of our communities to the multi-scaled interactions of fire behaviour factors in a computationally efficient manner. The index serves as a proxy for relative wildfire likelihood and represents temporally and spatially variable patterns in wildfire likelihood. The index will support wildfire risk assessments, including exploring problems such as optimising landscape treatment placements.

Here, we introduce the connectivity index as a multi-scaled, process-informed spatial aggregation of wildfire hazard properties across a landscape. We use a case study landscape to compare the connectivity index against simulated burn probability and historical burnt areas. Using a historically-informed parameterisation, we find a high correlation (0.83) to simulated burn probability with a fraction of the computational effort (0.3% of the runtime). The connectivity index also demonstrates an improved ability to explain historical burnt areas. We identify opportunities to further improve performance by incorporating the index into data-driven model structures.

Our findings demonstrate that the connectivity index captures structural patterns in wildfire likelihood, as influenced by the interaction of fire behaviour factors across multiple scales. By achieving this in a computationally efficient manner, we believe that the connectivity index can work alongside other measures of wildfire likelihood to inform and plan wildfire risk reduction activities, including in large-scale analysis.

How to cite: Radford, D., Maier, H., van Delden, H., Zecchin, A., and Jeanneau, A.: Efficiently Estimating Patterns in Wildfire Burn Probability, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4619, https://doi.org/10.5194/egusphere-egu23-4619, 2023.

10:05–10:15
|
EGU23-14932
|
ECS
|
On-site presentation
|
Luke Oberhagemann, Maik Billing, Markus Drüke, Boris Sakschewski, and Kirsten Thonicke

The SPITFIRE fire model is used with several Dynamic Global Vegetation Models to model fire-vegetation interactions on a global scale. Since its development in 2010 it has been used for multiple studies in this field. The model consists of components that calculate ignitions, fire spread, post-fire mortality, and emissions. We find that the fire spread component of this model contains errors that introduce significant biases to its results. In particular, errors in the application of the Rothermel equation result in fires that are significantly too large and intense. Further, unphysically low live grass moistures in the model result in excessively fire-prone grasslands, and therefore a strong link between the presence of grasslands and the presence of fire. This combination results in areas where the SPITFIRE model calculates excessive tree mortality and consequent grassland formation. We perform a detailed analysis of these errors and examine the impact that corrections to them have on SPITFIRE model results.

How to cite: Oberhagemann, L., Billing, M., Drüke, M., Sakschewski, B., and Thonicke, K.: Correcting errors in the SPITFIRE fire model that result in excessively large and intense fires, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-14932, https://doi.org/10.5194/egusphere-egu23-14932, 2023.

Coffee break
Chairpersons: Andrea Trucchia, Joana Parente, Marj Tonini
Convener introduction (convener: Andrea Trucchia + Joana Parente)
10:45–10:55
|
EGU23-9310
|
ECS
|
On-site presentation
Andrina Gincheva, Alberto Moreno, Sonia Jerez, Juan Pedro Montávez, and Marco Turco

This contribution seeks to better understand recent changes in synchronous fire danger across Europe that can overwhelm fire suppression capacity. We analyze the spatio-temporal synchronicity of fire danger in Europe over the period 1979-2021 based on the Canadian Fire Weather Index, one of the most commonly used fire indices globally (FWI; Vitolo et al. 2020). The daily synchronicity index indicates the total area with a level of FWI above 50, that represents the extreme fire danger threshold as classified by the European Forest Fire Information System (EFFIS). The annual mean surface affected by synchronicity extreme fire danger increased by about 100000 km2 over the 42-year-long study period (i.e. 57% of the mean historical value). The expansion of synchronized fire potential can compromise fire management efforts.

References

Vitolo, C., Di Giuseppe, F., Barnard, C., Coughlan, R., San-Miguel-Ayanz, J., Libertá, G., & Krzeminski, B. (2020). ERA5-based global meteorological wildfire danger maps. Scientific data, 7(1), 1-11.

Acknowledgments

We acknowledge funding through the project ONFIRE, grant PID2021-123193OB-I00, funded by MCIN/AEI/ 10.13039/501100011033. A.G. thanks the Ministerio de Ciencia, Innovación y Universidades of Spain for PhD contract FPU19/06536.

How to cite: Gincheva, A., Moreno, A., Jerez, S., Montávez, J. P., and Turco, M.: Assessing fire danger synchronicity in Europe, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9310, https://doi.org/10.5194/egusphere-egu23-9310, 2023.

10:55–11:05
|
EGU23-8876
|
ECS
|
On-site presentation
Ruxandra Zotta, Stefan Schlaffer, Markus Hollaus, Alena Dostalova, Harald Vacik, Mortimer Müller, Clement Atzberger, Markus Immitzer, Gergö Dioszegi, and Wouter Dorigo

The frequency and severity of wildfires in the Alpine region will likely increase due to climate change. Most fire danger forecasts currently adopted in this region are based on meteorological data, such as the Canadian Fire Weather Index (FWI). They are typically only available at relatively coarse spatial resolutions (up to ca. 1 km) and, therefore, are of limited use in mountain regions with complex topography. Other factors, such as vegetation type and structure and the role of humans causing ignitions, are typically not considered.  

We address this gap by presenting a novel, high-resolution, satellite-supported integrated forest fire danger system (IFDS) for Austria. For this purpose, we use radar and optical satellite data from the Copernicus Sentinel-1 and Sentinel-2 missions, airborne laser scanning (ALS), socio-economic data, and topographic properties next to meteorological data. Two independent methods were investigated: (i) an expert-based approach that allows combining various data layers with different weightings assigned by experts and (ii) a machine-learning approach. Here, we focus on the results of the machine learning approach for a study area covering the federal state of Styria in Austria (ca. 16 400 km²). We use several data layers computed within our study as predictors in random forest models. Moisture indicators and tree species maps were derived from satellite data from the Copernicus Earth observation programme. Vegetation structure parameters, solar potential and a digital surface model (DSM) were derived from ALS data. In addition to the remote sensing data, we used meteorological variables, fire weather indices (FWI) and socio-economic data. We trained the model using forest fire events from the Austrian fire database.  

The cross-validation showed that the best-performing model predicts high fire danger for most fire events (87%). By integrating all the information layers compared to a baseline model using only FWI, the overall accuracy improved from 68% to 87%. The feature importance showed that the vegetation structure parameters, tree species, socio-economic parameters and DSM are essential for the model in addition to the meteorological predictors. Using this data-driven approach allowed us to learn from past fire occurrences and improved the spatial representation of fire ignition drivers, their importance and interactions. Also, this method permitted the identification of areas with higher danger risk, typically located in the vicinity of densely populated settlements. 

This study has been performed within the CONFIRM project with funding from the Austrian Research Promotion Agency (FFG). 

How to cite: Zotta, R., Schlaffer, S., Hollaus, M., Dostalova, A., Vacik, H., Müller, M., Atzberger, C., Immitzer, M., Dioszegi, G., and Dorigo, W.: Using satellite, airborne laser scanning and socio-economic data in a machine learning framework for improved fire danger modelling in the Alps, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8876, https://doi.org/10.5194/egusphere-egu23-8876, 2023.

11:05–11:15
|
EGU23-9137
|
Highlight
|
On-site presentation
Marco Turco, Andrina Gincheva-Norcheva, Miguel Ángel Torres-Vázquez, and Sonia Jerez

As societal exposure to large fires increases, there is a growing concern about possible shifts in the fire regimes due to climate change. Understanding the response of fires to climatic variations is essential to adapt fire management systems and to design future prevention strategies. However, many aspects on this topic remain to be revealed.

ONFIRE, a Spanish national research project supported by the State Investigation Agency and by Ministry of Ministry of Science and Innovation, aims to push the state-of-the-art knowledge of climate impacts on fires beyond its current limitations and applications. Moreover, this project is open to any contributions. Specifically, we encourage any researcher/fire agency to joint this initiative.

In this contribution we show some preliminary results related to the main objectives of the ONFIRE project:

  • The creation of a first-of-its-kind unified, open-access and user-friendly database comprised of all available burned area records from national inventories.
  • A better understanding of past trends in fire series and their attribution to the anthropogenic component of climate change.
  • An assessment of the spatio-temporal synchronicity of fire danger.
  • The design and implementation of a public operational prototype system to perform global seasonal predictions of climate-driven fire risk for decision-making applications.

Acknowledgments
We acknowledge funding through the project ONFIRE, grant PID2021-123193OB-I00, funded by MCIN/AEI/ 10.13039/501100011033. A.G. thanks the Ministerio de Ciencia, Innovación y Universidades of Spain for PhD contract FPU19/06536.

How to cite: Turco, M., Gincheva-Norcheva, A., Torres-Vázquez, M. Á., and Jerez, S.: The ONFIRE project - on the response of fires to climate variability and change, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9137, https://doi.org/10.5194/egusphere-egu23-9137, 2023.

11:15–11:25
|
EGU23-1758
|
ECS
|
On-site presentation
Motti Zohar, Bar Gennosar, Ronnen Avny, Naama Tessler, and Avigdor Gal

For the past decade, Twitter has become a robust platform for distributing messages (tweets) among numerous subscribers worldwide. During and around the occurrence of natural hazards, tweet volumes increase significantly. While Twitter is used for near real-time alerts, processes for extracting reported damage from tweets and resolving their geographical spread in high resolution are still under development. In this study we examine the spatio-temporal distribution of tweets associated with the November 2016 fire, which lasted in Haifa (Isreal) for nearly two days. The acquired tweets were classified and filtered using topic modeling procedure, a portion of them were accurately georeferenced by the Open Street Map and GeoNames gazetteers, and their hyperlocal spatio-temporal patterns were examined. It was found that the tweets’ sentiment (peaks and lows) corresponds to the fire’s occurring cascading events while their spatial distribution can be aligned with most of the actual (true) reports. Despite large uncertainties in the process, results show Twitter can serve in the future as another layer of information to assist decision makers and emergency agencies during and after cascading catastrophes.

How to cite: Zohar, M., Gennosar, B., Avny, R., Tessler, N., and Gal, A.: Spatio-temporal analysis in high resolution of curated tweets associated with the November 2016 wildfire in Haifa (Israel), EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-1758, https://doi.org/10.5194/egusphere-egu23-1758, 2023.

11:25–11:27
11:27–11:37
|
EGU23-8114
|
solicited
|
Highlight
|
On-site presentation
|
Joao Pedro Nunes, Joana Parente, Akli Benali, and Luis Filipe Dias

Wildfires can change vegetation cover and soil properties, often enhancing surface runoff and sediment transport processes. The ash produced by these fires can also be mobilized and contaminate downstream water bodies with fine sediments, heavy metals, nutrients and organic carbons. Urban water supplies are usually taken from watersheds with natural vegetation cover to limit agricultural contamination; this makes these supply sources vulnerable to disruption after wildfires, an occurrence which might be infrequent but carry large consequences such as supply disruptions. Moreover, mobilized ashes can deposit in stream and reservoir beds and be resuspended for years after the fire, prolonging the disruption in time.

Forest and water managers can take some steps to manage these risks, including preventive forest management planning and contingency planning for emergency interventions in the burnt areas themselves and at the treatment plants. However, the elements to quantify these risks are generally poorly quantified in most fire-prone watersheds. Fire regimes might be known, but the relationships between fire characteristics and impacts on water quality are difficult to assess without good datasets; and the costs and benefits of different mitigation approaches are usually not well understood. To further complicate matters, the impacts of wildfires on hydrology and sediment processes tend to vary significantly across climatic regions, making it difficult to transfer knowledge.

This presentation will provide an overview of the issues surrounding the assessment of the risk of water quality contamination after wildfires. It will also provide an example on how this is being done in Portugal, through project FRISCO: Managing Fire-Induced Risks of Water Quality Contamination (FCT, ref. PCIF/MPG/0044/2018). The project, now in its fourth and final year, has (i) determined the most important fire and post-fire conditions leading to fire-induced contamination events, through a detailed analysis of a 20-year water quality database for over a hundred water supply reservoirs, linked with a concurrent atlas of fire severity; (ii) developed, together with water managers, a risk assessment index that can be used after a fire to inform managers on the need for further action; and (iii) is assessing multiple post-fire intervention options, from the biophysical and socioeconomic perspectives, to help inform managers on which actions they can take to address the issue. This project provides a blueprint on how these issues might be addressed by water managers in other fire-prone watersheds.

How to cite: Nunes, J. P., Parente, J., Benali, A., and Dias, L. F.: Assessing and managing the risk of water quality contamination after wildfires: an example approach for Portugal, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8114, https://doi.org/10.5194/egusphere-egu23-8114, 2023.

11:37–11:47
|
EGU23-7392
|
ECS
|
On-site presentation
Michaela Flegrova and Helen Brindley

Wildfires can have significant impacts on the Earth's surface albedo, with effects that can be long-lasting. In Africa, the frequency and severity of wildfires are changing due to a combination of factors including drought, land use change, and human activity. However, the impact of wildfires on surface albedo in Africa is not well understood and previous research has produced conflicting results.

In this study, we are using Moderate Resolution Imaging Spectroradiometer (MODIS) data to investigate the potential changes in surface albedo following wildfires in Africa. Our preliminary results suggest that wildfires may have a complex and variable impact on albedo, with some regions potentially experiencing a decrease in albedo and others an increase or no significant change following a wildfire.

We are also exploring the use of different approaches to analyse the MODIS data to better understand how the choice of method may impact the results and looking how the effect varies for different land cover types. In addition, we are using Fire Radiative Power data from the Spinning Enhanced Visible and Infrared Imager (SEVIRI) to examine the role of fire intensity in driving albedo changes.

This research aims to provide new insights into the impact of wildfires on surface albedo in Africa, the underlying mechanisms driving these changes and how different analysis methods affect the conclusions. These results may have important implications for land management strategies in the region, and for understanding the impact of wildfires on local climate and ecosystem processes.

How to cite: Flegrova, M. and Brindley, H.: Investigating the Impact of Wildfires on Surface Albedo in Africa Using MODIS Data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-7392, https://doi.org/10.5194/egusphere-egu23-7392, 2023.

11:47–11:57
|
EGU23-8151
|
ECS
|
On-site presentation
Amin Khairoun, Florent Mouillot, Wentao Chen, Philippe Ciais, and Emilio Chuvieco

Deforestation represents one of the major challenges of the current era as it constitutes an imminent threat to the forest carbon sink offsetting the different sources of greenhouse gas (GHG) emissions. Fire is highly contributing to forest loss and degradation as a driver (i.e. wildfires) and a clearing tool (e.g. slash-and-burn agriculture). However, fire impacts were mainly analysed at coarse spatial resolutions, and therefore estimations are deemed to be very conservative because of the high omission errors in global BA products. In this study, we focus on sub-Saharan Africa (SSA), which represents the region most affected by fires globally. We analysed fire-related forest loss at 20 m resolution based on FireCCISFD datasets available for 2016 and 2019, in combination with the Global Forest Cover (GFC) maps derived from medium-resolution sensors (Sentinel 2 and Landsat, respectively). These estimations were compared to the ones of two global BA products derived from the MODIS sensor, namely MCD64A1 (Collection 6) and FireCCI51. We found that fires are a precursor of forest loss, as burned areas had more than twice the chance to be lost than unburned ones during the two study years, and that on average, fires were directly involved in almost half of forest losses in SSA (46 ± 3.80% in 2019 and 47 ± 4.21% in 2016). Depending on biomes and the year of study, fire-related forest cover loss ranged from 32 ± 1.83% to 75 ± 3.65%. In general, the subtropics dominated by Tropical Savanna and Dry Tropical Forest exhibit the highest contributions of fire, whereas the tropical zone, where Moist Tropical Forest is prevailing, showed lower contributions due to the lower fire activity. Fragmentation, as well as fire season, were found to be drivers of forest loss as the majority of these losses occur in fragmented areas close to forest edge (< 320 m) and in late fire season. We conclude that the use of medium-resolution BA products to assess fire impacts in tropical ecosystems is crucial.

How to cite: Khairoun, A., Mouillot, F., Chen, W., Ciais, P., and Chuvieco, E.: Assessment of fire contribution to forest loss in sub-Saharan Africa using medium-resolution BA, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8151, https://doi.org/10.5194/egusphere-egu23-8151, 2023.

11:57–12:07
|
EGU23-17073
|
On-site presentation
Adrien Guyot, Kathryn Turner, Jordan Brook, Joshua Soderholm, Alain Protat, and Hamish McGowan

The 2020 worldwide bushfire activity was the most intense and widespread since the existence of satellite-based observational capabilities. The economic, societal, and ecological consequences have been immense: in Australia alone, the 2019-2020 Black Summer bushfires resulted in an economic cost of more than $100 billion, a burnt area of more than 18 M ha, 10,000 destroyed buildings, 34 direct deaths and more than 400 deaths due to smoke exposure. On the Australian East Coast, these intense wildfires lasting for almost two months produced very large smoke plumes and often fire-triggered thunderstorms - pyrocumulonimbus. These plumes and storms were predominantly within the range of operational weather radars, enabling observations of the plume thermodynamics, kinetics, and their composition. Here, we present two months of observations from a dual pol weather radar located near Sydney: a newly developed texture- and machine learning-based method enables us to extract smoke plumes and associated clouds from complex weather radar scenes including clear air and sea clutter. The characteristics of these smoke plumes are quantified including cloud top heights, volumes, projected areas, horizontal extends and daily dynamics. Using dual polarisation data, in-depth insights can be gained on the plumes’ microphysics and the transition zone from smoke to pyrocumulus and pyrocumulonimbus. These high-resolution observations contribute to a better understanding of smoke plume dynamics and provide the foundations to develop nowcasting tools to predict associated hazards such as fire-triggered storms such as downbursts, plume collapse, and ember transport.

How to cite: Guyot, A., Turner, K., Brook, J., Soderholm, J., Protat, A., and McGowan, H.: Properties of smoke plumes and associated clouds from the Australian 2019/2022 wildfiresidentified using dual polarisation weather radar observations, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-17073, https://doi.org/10.5194/egusphere-egu23-17073, 2023.

12:07–12:17
|
EGU23-15418
|
ECS
|
Highlight
|
On-site presentation
Ana Gonçalves, Bruno Barbosa, Sandra Oliveira, and José Luís Zêzere

After the extreme wildfire events occurred in Portugal in 2017, which burned about 500.000 hectares of land and killed more than 100 people, several initiatives were established aiming to improve the resilience and safety of communities, among which the programs "Safe Villages” (SV) and “Safe People" (SP), coordinated by the Portuguese Civil Protection (ANEPC). The present study aimed to identify the territorial and social characteristics of the villages where the SV program is being implemented. For this study, a database was built with the location of the SV already implemented in three different regions: i) Algarve, the municipalities of Alcoutim, Monchique, São Brás de Alportel; ii) Pinhal Interior, the municipalities of Alvaiázere, Figueiró dos Vinhos and Oliveira do Hospital; iii) Caramulo mountain region, the municipalities of Tondela and Mortágua. The location of the villages where the SV program is implemented was obtained via the overlap of a point representing each SV with the corresponding built-up area. The analysis was focused on the surrounding area of each village and a buffer of 100 m was drawn, as defined by law (DL 82/2021). For each protective buffer, the following parameters were calculated: percentage of critical area (high and very high structural hazard); percentage of wildland area (forest-shrubland-herbaceous); percentage of slopes above 20°; the number of times burned between 1975 and 2021; and the population density within the built-up area. Preliminary results show that, in total, 166 SV were implemented in 6 municipalities, while 2 municipalities do not have any SV implemented, showing that the implementation of the program differs largely between municipalities and regions. The characteristics of the surrounding area of the SV also vary; in Algarve, 97% of SV have 75% of the buffer with critical area, whereas in Pinhal Interior and Caramulo, there are 89% and 78%, respectively. The SV with the highest population density is in the Caramulo region with 58.6 hab/ha, then the Pinhal Interior with 56.9 hab/ha and lastly the Algarve region with 44.9 hab/ha. The municipality of Alcoutim has the highest implementation of SV (84 villages), although the maximum critical area found in their surroundings is only 18%, in one village, and in 19% of the villages, no critical area is found. These results indicate that the implementation of the SV program does not depend only on the physical factors of the territory, but also on the involvement of the population, namely the existence of a volunteer Safety Officer. This person must know the local context, existing structures at the local level and the actions required in case a wildfire is expected. The efficiency of this program and the protection of local communities can be improved by combining different criteria in selecting the villages, such as the proportion of critical area and population density.

This work was funded by FCT, I.P.: JLZ in the framework of the project People&Fire [PCIF/AGT/0136/2017], AG and BB in the scope of PhD projects [2020.07651.BD], [2022.12095.BD], SO under the contract ‘2020.03873.CEECIND′. 

How to cite: Gonçalves, A., Barbosa, B., Oliveira, S., and Zêzere, J. L.: Assessing the implementation of the “Safe Villages” program for wildfire mitigation, in three regions of Portugal, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15418, https://doi.org/10.5194/egusphere-egu23-15418, 2023.

12:17–12:27
|
EGU23-8650
|
ECS
|
On-site presentation
|
Maria Zubkova, Michael Humber, and Louis Giglio

Globally, the amount of fire activity detected by satellite observations seems to have changed substantially in the last two decades. Discussions regarding the main force behind the current trends have dominated research in recent years, with several studies attributing the global decline in wildfires to the expansion of agricultural areas. Nevertheless, most studies have failed to acknowledge that the regions where these changing patterns have been observed are known to be data-poor.

Here we discuss the uncertainties and limitations of remotely sensed data used to determine global trends in burned area and changes in their potential drivers. Specifically, we i) quantify changes in the amount of burned area and cropland area and illustrate discrepancies between commonly used datasets, ii) state the limitations of remote-sensed fire and land cover products, and iii) highlight recent fire-trend studies and hypothesize likely effects of the choice of datasets on their conclusions. We argue that to legitimately conclude that the reported global decline in fire activity is driven by cropland expansion, three conditions must be met. First, negative trends should be global in scope, not localized to a specific continent or region. Second, the trends in both fire activity and agricultural areas must be supported by data that definitively demonstrate such trends over time. And third, the decline in the amount of burned area must be documented within or proximal to the areas of cropland expansion.

We demonstrate that a drastic decline in fire activity in the last 20 years was only observed within regions highly sensitive to coarse resolution fire product biases, while most regions/continents did not experience significant changes in the amount of burned area. Additionally, the analysis of several global land cover products reveals the lack of consistency in the direction and magnitude of the trend in cropland land cover type. And finally, based on the available data, no clear spatial relationship can be detected between areas experiencing cropland expansion and areas where the amount of burned area is declining.

Therefore, our knowledge of anthropogenic effects on fire, while growing, remains incomplete, particularly the effects of cropland expansion on wildfires, due to the low confidence in estimated fire trends within human-managed land and lack of understanding of the spatial distribution of cropland expansion/loss.

How to cite: Zubkova, M., Humber, M., and Giglio, L.: Is global burned area declining due to cropland expansion? How much do we know based on remotely sensed data?, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-8650, https://doi.org/10.5194/egusphere-egu23-8650, 2023.

12:27–12:30

Posters on site: Tue, 25 Apr, 16:15–18:00 | Hall X4

Chairpersons: Marj Tonini, Joana Parente, Andrea Trucchia
Wildfires mapping and main drivers (convener: Marj Tonini + Joana Parente)
X4.30
|
EGU23-13976
|
ECS
Tichaona Mukunga, Matthias Forkel, Matthew Forrest, Ruxandra-Maria Zotta, Stefan Schlaffer, and Wouter Dorigo

Wildfires are a pervasive feature of the terrestrial biosphere and contribute large carbon emissions within the Earth system. Humans are responsible for the majority of wildfire ignitions. Physical and empirical models are used to estimate the effects of fires on vegetation dynamics and the Earth system. However, there is no consensus on how human-caused fire ignitions should be represented in such models. 

 We aim to identify which globally available predictors of human activity explain global fire ignitions as observed by satellites. We apply a random forest machine learning framework to state-of-the-art global climate, vegetation, and land cover datasets to predict global fire ignition density remote sensing data. We establish a baseline against which influences of socioeconomic data (cropland fraction, gross domestic product (GDP), road density, livestock density, and grazed lands) are evaluated to determine their effects on fire ignition occurrence predictions. Our results show that a baseline random forest without human predictors captures the spatial patterns of fire ignitions globally but overestimates ignition occurrence, with the highest predictions over Sub-Saharan Africa and South East Asia. Adding human predictors one by one to the baseline model reveals that human variables vary in their effects on fire ignitions, and GDP is the most vital driver of fire ignitions. We also find that high road density leads to decreased ignitions globally despite improved environmental access, mainly because these regions are primarily human settlements and will have fewer fires. A combined model with all human predictors shows that the human variables improve the ignition predictions in most areas of the world; still, some regions, e.g., East Africa, have worse predictions than the baseline model.  

We conclude that an ensemble of human predictors can add value to physical and empirical models. There are complex relationships between the variables, as evidenced by the improvement in bias in the combined model compared to the individual models. 

Furthermore, the variables tested have complex relationships that random forests may struggle to disentangle.  Further work is needed to reach concrete conclusions at a global scale, but rather introduce the need for additional work to detangle the complex regional relationships between these variables, particularly across Central and Eastern Africa, where the full model performs poorly despite the high availability of fire ignition data. 

How to cite: Mukunga, T., Forkel, M., Forrest, M., Zotta, R.-M., Schlaffer, S., and Dorigo, W.: Effect of socioeconomic variables in predicting global wildfire ignition occurrence, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-13976, https://doi.org/10.5194/egusphere-egu23-13976, 2023.

X4.31
|
EGU23-5159
Marj Tonini, Joana Parente, Nikos Koutsias, and Mário Pereira

Wildfires play a fundamental role in Land Use/Land Cover (LULC) change dynamics by burning vegetation in forested and rural areas and by affecting human infrastructures. Conversely, LULC can disturb fire regime by altering vegetation cover, conditioning subsequent transitions, and acting on fuel loads and continuity. Though there is an evident mutual influence between wildfires and LULC changes, a rigorous quantification of their reciprocal effects in Europe has never been performed before. To fill this gap, in the present study we developed a methodology allowing the evaluation of different indicators for the quantitative assessment and a better understanding of the transitions among LULC classes and Burnt Areas (BA) that occurred in Europe within the last two decades (2000 – 2020).

Our analyses revealed that the two LULC classes which had experienced major changes were Forests (44%), and Scrubs and/or herbaceous vegetation associations (32%). As a general trend, within the five European Mediterranean Countries more prone to wildfires (Portugal, Spain, France, Italy, and Greece) we found a decrease in the classes Forests and Arable land, and an increase in Scrubs and/or herbaceous vegetation associations, suggesting the impact of wildfires in shaping the natural and rural landscape. This assumption was better evaluated and confirmed by the following analyses, performed at both the European and national levels. Results showed that most of the BA have occurred in Forests (42% for the entire Europe), with a predominance in Coniferous forests; the subsequent transitions from BA were generally to Transitional woodland/shrub or again to BA. This last indicates a high frequency of wildfires in a given area, while the first transition can be partially due to the regeneration/recolonization of the vegetation after a wildfire event. Outcomes for the single countries followed almost the same trend.

Overall, our results confirm the existence of a strong relationship between wildfires and LULC changes in Europe, which have been quantified in the present study. These findings are in line with previous research and provide a deep insight into the process at the global and local levels, paving the way for further analyses on fire intensity and frequency with coupled environmental elements of land cover and climate changes.

How to cite: Tonini, M., Parente, J., Koutsias, N., and Pereira, M.: Transitions between use/land cover classes and burnt areas in Europe , EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-5159, https://doi.org/10.5194/egusphere-egu23-5159, 2023.

X4.32
|
EGU23-11565
Analysis of Wildfire Hotspots in Using Copula for Climate Indicators in Kalimantan, Indonesia.
(withdrawn)
Ardhasena Sopaheluwakan, Mohamad Khoirun Najib, and Sri Nurdiati
Fire weather (convener: Andrea Trucchia + Joana Parente)
X4.33
|
EGU23-15341
|
ECS
Carolina Gallo, Jonathan Eden, Bastien Dieppois, Igor Drobyshev, Peter Fulé, Jesús San-Miguel-Ayanz, and Matthew Blackett

Weather and climate play an important role in shaping global fire regimes and geographical distributions of burnable areas. At the global scale, fire danger is likely to increase in the near future due to warmer temperatures and changes in precipitation patterns, as projected by the Sixth Assessment Report (AR6) of the Intergovernmental Panel on Climate Change (IPCC). There is a need to develop the most reliable projections of future climate-driven fire danger to enable decision makers and forest managers to respond to future fire events.

Climate change projections generated by general circulation models, especially those that contribute to the 6th Coupled Model Intercomparison Project (CMIP6), are the most important basis in understanding future changes in fire-conducive weather and climate associated with a warming world. However, errors and biases inherent to such models are rarely taken into account when generating climate change projections. For fire weather in particular, projections have typically been expressed by a single model or through a multi-model mean. This approach can be misleading, as it explains little about the consensus among different models and their uncertainties. Here, following a comprehensive evaluation of the performance of 16 different CMIP6 climate model ensembles, we present new scenarios for detecting changes in fire-prone conditions based on a statistical weighting approach that accounts for both model skill and independence. We demonstrate the value of a weighted approach in accounting for and reducing model uncertainties, and more generally in the development of fire weather scenarios that ultimately as useful as possible. In conclusion, we make recommendations for how the new set of scenarios can benefit end users in decision-making and forest management.

How to cite: Gallo, C., Eden, J., Dieppois, B., Drobyshev, I., Fulé, P., San-Miguel-Ayanz, J., and Blackett, M.: A model weighting scheme for fire weather projections simulated by CMIP6 climate model ensembles, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-15341, https://doi.org/10.5194/egusphere-egu23-15341, 2023.

X4.34
|
EGU23-12233
|
ECS
Stephanie Bohlmann and Marko Laine
Wildfires caused by extreme weather conditions e.g high temperatures and drought have been increasing both in frequency and magnitude in the recent years. Due to the changing climate also regions with few recorded forest fires in the past are likely to be more frequently affected by wildfires. These wildfires have many social, economic, and environmental impacts. The EU funded SAFERS project, Structured Approaches for Forest Fire Emergencies in Resilient Societies, is creating an open and integrated platform featuring a forest fire decision support system in order to support societies becoming more resilient when acting against forests fires. The platform will, among other information, provide weather forecasts and forest fire indices to assess the risk of forest fires. Using the Canadian Forest Fire Weather Index (FWI) system [Van Wagner, 1974], fuel moisture (FFMC, DMC, DC) and fire behaviour indices (ISI, BUI, FWI) can be calculated using solely meteorological parameters (temperature, relative humidity, wind speed, and 24-hour precipitation). The Canadian FWI has been proved useful for forest fire risk assessment in different regions and has been adapted by multiple meteorological agencies worldwide. In our study, we use high resolution deterministic weather forecasts, as well as medium-range and extended-range weather forecasts provided by the European Centre for Medium-Range Weather Forecasts to calculate the FWI indices for up to 3, 15 and 46 days ahead, respectively. In our contribution we will present preliminary results of FWI verification and calibration methods.

References:
Van Wagner, C. E.: Structure of the Canadian Forest Fire Weather Index, Departmental Publication 1333, Environment Canada, Canadian Forestry Service, Petawawa Forest Experiment Station, Chalk River, Ontario, p. 49, 1974.

How to cite: Bohlmann, S. and Laine, M.: Assessment of fire weather index calculations in Europe, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-12233, https://doi.org/10.5194/egusphere-egu23-12233, 2023.

Fire risk assessment (convener: Joana Parente + Marj Tonini)
X4.35
|
EGU23-9681
|
ECS
|
Highlight
Andrea Trucchia, Hamed Izadgoshasb, Giorgio Meschi, Paolo Fiorucci, and Marj Tonini

Comparison of Different Algorithms and Vegetation Classes’ Importance Ranking in Wildfire Susceptibility Maps. 
Wildfire Susceptibility Maps (WSM) and the analysis of the explanatory variables affecting the model’s predictions are innovative tools to support forest protection and management plans. Namely, WSM identify areas subject to wildfire, in terms of relative spatial likelihood, on the base of the observed past events, stored in spatio-temporal inventories, and on the local environmental and anthropogenic properties of an area. Approaches based on Machine Learning (ML) are particularly suited for WSM since they are capable to make predictions on data by modelling the hidden and non-linear relationships between a set of input variables and the output observations.
In the present work, Authors continue a research framework developed at local scale for Liguria Region, and lately improved at national scale (Italy), consisting in the implementation of a ML-approach, based on the algorithm Random Forest, allowing to assess the susceptibility to wildfires under the influence of different variables (e.g., land cover, vegetation classes, altitude and its derivatives, nearby infrastructures). In the present study the following improvements are introduced: (i) to evaluate which ML-algorithm performs better in terms of prediction capabilities we compared Random Forest (RF), Multi-layer Perceptron (MLP), and Support Vector Machine (SVM); (ii) to evaluate the impact of different classes of local and neighbouring vegetation on wildfires occurrence we used of a more accurate map of vegetation as input local explanatory variable; (iii) to consider both the spatial and the temporal variability of the burning seasons (summer and winter) we improved the selection of the testing dataset, based on a clustering approach. 
The output probabilistic predicted values resulting from the different ML-algorithms (RF, MLP, and SVM) allowed to elaborate the seasonal WSMs. Finally, the spatial distribution of the more susceptible areas will be presented. The performance of the three ML-algorithms was assessed by means of the AUC (Area Under the Curve) ROC (Receiver Operating Characteristics), evaluated over the testing dataset. In addition, the variable importance ranking was estimated as by-product of RF, which can handle both the typical numerical variables and native categorical variables (as for the classes of vegetation at pixel level). Vegetation resulted by far to be the most important explanatory variables; the marginal effect of each single class of vegetation was also assessed and the results will be discussed. 
Reference 
Trucchia, A.; Izadgoshasb, H.; Isnardi, S.; Fiorucci, P.; Tonini, M. Machine-Learning Applications in Geosciences: Comparison of Different Algorithms and Vegetation Classes’ Importance Ranking in Wildfire Susceptibility. Geosciences 2022, 12, 424. https://doi.org/10.3390/geosciences12110424

How to cite: Trucchia, A., Izadgoshasb, H., Meschi, G., Fiorucci, P., and Tonini, M.: Comparison of Different Algorithms and Vegetation Classes’ Importance Ranking in Wildfire Susceptibility Maps., EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9681, https://doi.org/10.5194/egusphere-egu23-9681, 2023.

X4.36
|
EGU23-6457
Maria Papathoma-Koehle, David Hausharter, Matthias Schlögl, and Sven Fuchs

Recent events have clearly shown that wildfires may occur in areas that until now have not experienced large wildfires and the associated consequences (e.g. Scandinavia, Siberia, Austria, etc.). There is a need to understand the risk posed by wildfires and develop tools for the assessment of the vulnerability of assets such as buildings and infrastructure. Building quality and design standards are important not only because building loss is costly but also because robust buildings may offer shelter when evacuation is not possible. Studies aiming at the analysis of wildfire vulnerability for the built environment are limited.  We present a new wildfire vulnerability index for buildings in Austria based on a choice of vulnerability indicators and expert judgement. Vulnerability indicators express characteristics of the buildings and their surroundings that influence their vulnerability to a hazardous process. A list of indicators based on the existing literature has been used together with an expert panel to decide which indicators may be relevant and what is their importance in controlling vulnerability. The indicators and their weights are aggregated into a wildfire vulnerability index which can be assigned for each building located in the wildland urban interface (WUI) zone. This index is related to information such as the structural building type, the roof type, material and shape of the roof, the inclination of the ground, the surrounding vegetation, the material of the shutters and the ground covering. The resulting vulnerability and its spatial pattern may guide decisions, strategies and vulnerability reduction activities that will increase the resilience of communities to this emerging risk. The index may be used by decision-makers, emergency services, homeowners and insurance companies to visualise physical vulnerability to wildfire.

How to cite: Papathoma-Koehle, M., Hausharter, D., Schlögl, M., and Fuchs, S.: A wildfire vulnerability index for buildings in Austria, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-6457, https://doi.org/10.5194/egusphere-egu23-6457, 2023.

Fire management (convener: Andrea Trucchia + Joana Parente)
X4.37
|
EGU23-4815
Inseon Suh, Sungmin Kim, Youngmi Lee, Kyewon Jun, and Byungsik Kim

Spatiotemporal prediction of wildfire spread is very important to minimize damage and respond urgently to these urban forest fires since forest fire damages caused by strong winds such as the Foehn wind are increasing every year, especially along the eastern coastal cities in Korea. Because forest fires spread under the influence of environmental factors such as fuel, topography, and weather, the values of these factors are known as important variables for accurate forest fire spread prediction models. In this study, we developed a forest fire spread prediction model that considers wind speed, wind direction, fuel information, and slope as main factors by analyzing past forest fire damage data in Gangwon-do such as meteorological factors, fuel and terrain characteristics. The wildfire spread prediction model (hereinafter referred to as WINS, Wind field Network for Fire Spread Simulation) produces meteorological information of a numerical forecasting model calibrated with MOS (Model output statistics) as 1km x 1km grid values, and the slope and fuel information between each grid are configured. Land use information in the Gangwon area is divided into artificial grassland, mixed forest, natural measures, coniferous forest, and broad-leaved forest, and the depth of the surface fuel layer and the amount of water removal surface fuel are layered by grid according to Anderson fuel type. As soon as the ignition point information is obtained, the predicted wind speed and wind direction values of the grid are layered by time and GIS-based predicted spatiotemporal information is produced. The WINS model for forest fire cases in the Gangwon region occurred from 2019 to 2021 was verified, and real-time map-based forest fire spread prediction information was utilized by local governments and related stakeholders in the urban forest fire response task and decision-making stage according to the simulated scenario.

 

"This research was supported by the program of Research Program to Solve Urgent Safety Issues (2022M3E9A1095664), through the National Research Foundation of Korea(NRF), funded by the Korean government. (Ministry of Science and ICT(MSIT), Ministry of the Interior and Safety(MOIS))."

How to cite: Suh, I., Kim, S., Lee, Y., Jun, K., and Kim, B.: Development of real-time forest fire spread prediction model based on big data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4815, https://doi.org/10.5194/egusphere-egu23-4815, 2023.

X4.38
|
EGU23-4969
|
ECS
Taejun Sung, Yoojin Kang, and Jungho Im

Due to the irregular and sporadic nature of wildfires, continuous monitoring of large areas is required. Since geostationary satellite sensors can observe large areas with high temporal resolution, they are suitable for monitoring wildfires in real time. However, the threshold algorithm currently employed for satellite-based active fire detection has poor performance in sensors with low spatial resolution. In addition, the algorithm does not account for environmental conditions that affect wildfire detection, resulting in poor generalization performance for large areas. This study examines the viability of an adaptive active fire detection model by combining satellite and numerical model data with deep learning. A model for active fire detection was developed using commonly employed brightness temperature-related variables (key variables) and local environmental variables (sub variables). Key variables are the cross spectral and spatial differences between the MIR (central wavelength of 3.85 m) and 2 TIR (central wavelengths of 9.63 and 11.20 m) channels of the Advanced Himawari Imager (AHI). Sub variables include Solar zenith angle (SOZ) and satellite zenith angle (SAZ) of AHI, skin temperature (ST) and relative humidity (RH) of European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5)-land data. Four processes (confidence, frequency, land cover, and continuity tests) were used to extract reference fire samples from Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS) active fire products. To consider the different properties of key and sub variables, a 2-way convolutional neural network (CNN) structure was developed. To evaluate the influence of environmental variables, a CNN model without sub variables was adopted as a control model. The 2-way CNN (recall of 0.86, precision of 0.96, and standard deviation of recall of 0.13) was more robust at five focus sites than the control CNN (recall of 0.82, precision of 0.97, and standard deviation of recall of 0.163). Despite having a lower spatial resolution than MODIS/VIIRS, 2-way CNN outperformed other satellite-based active fire products (MODIS, VIIRS, AHI, and Advanced Meteorological Imager) in terms of detection capacity. The control CNN demonstrated poor performance under certain environmental conditions (high RH, high SAZ, and transition time between day and night), but 2-way CNN mitigates this tendency. In particular, the use of RH improved detection sensitivity, and SAZ contributed to the spatial robustness. This study demonstrated the significance of environmental conditions in active fire detection and proposed a suitable CNN structure for this intent. Based on the findings of this study, higher-level adaptive active fire monitoring under diverse environmental conditions will be possible together with explainable artificial intelligence.

How to cite: Sung, T., Kang, Y., and Im, J.: A robust deep learning-based active fire detection model in diverse environments by fusion of satellite and numerical model data, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-4969, https://doi.org/10.5194/egusphere-egu23-4969, 2023.

X4.39
|
EGU23-10656
So Ryeon Park, Sanghun Son, Jaegu Bae, Doi Lee, Minji Ryu, Jeong Min Seo, and Jinsoo Kim

Air pollutants, such as large amounts of carbon dioxide produced by fire, are at risk of promoting global warming, causing more frequent and more wildfires all over the world. Large-scale wildfires cause air pollutants to spread, resulting in a significant increase in fine dust concentration. And human damage, property damage, and natural ecosystem damage are increasing. Rapid and accurate information of fire damaged the forest areas using high-resolution satellite imagery is effective in preparing wildfire prevention measures and monitoring burned areas. In this regard, the many research is underway to estimate fire burn damage using the difference spectral feature between healthy forests and burned forests. There are many spectral indices including well-known indices such as NDVI(Normalized Difference Vegetation Index) and NBR(Normalized Burn Ratio). It is possible to estimate burned area by computing a difference image representing a change in the spectral wavelength of the image between pre- and post-fire. In the case of South Korea, it is difficult to estimate wildfires because the spectral characteristics of vegetation vary from each season according to climate. And Compared to other countries, Forest fires are occurring small scale of forest fires less than 2,000ha. In our study, the accuracy was compared by applying various spectral indices to the estimation and evaluation of burned area in South Korea, using one of the deep learning models U-Net. As a result of the IoU value of 0.80 or more, it was confirmed that it was possible to calculate the forest fire damage site.

Acknowledgments:

This research was supported by a grant (2021-MOIS37-002) of "Intelligent Technology Development Program on Disaster Response and Emergency Management" funded by Ministry of Interior and Safety (MOIS, Korea).
This work was supported by the "Graduate school of Particulate matter specialization" of Korea Environmental Industry & Technology Institute grant funded by the Ministry of Environment, Republic of Korea.

How to cite: Park, S. R., Son, S., Bae, J., Lee, D., Ryu, M., Seo, J. M., and Kim, J.: Wildfire burned area detection using with Sentinel-2 and UNet, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-10656, https://doi.org/10.5194/egusphere-egu23-10656, 2023.

Fire impacts (convener: Andrea Trucchia + Joana Parente)
X4.40
|
EGU23-3087
|
ECS
Understanding the drivers of pyroCb development over southeast Australia
(withdrawn)
Wenyuan Ma, Jason Sharples, and Zlatko Jovanoski

Posters virtual: Tue, 25 Apr, 16:15–18:00 | vHall NH

Chairpersons: Joana Parente, Andrea Trucchia, Marj Tonini
Wildfire mapping and man drivers, and impacts of wildfires (all the conveners)
vNH.4
|
EGU23-9608
|
Joaquin Bedia, Catharina Elisabeth Graafland, Andrina Gincheva, and Marco Turco

In light of the ongoing global climate change, a better understanding of global wildfire activity is the key to anticipate future impacts and minimize, as much as possible, their negative consequences on natural ecosystems and human economy. In this work we present a global-scale analysis of the inter-annual synchronicity of wildfires using two types of probabilistic networks, able to capture the underlying structures in the burned area data from two different approaches: correlation-based and bayesian networks. By studying their properties through centrality measures and community detection for the complex network, and inference from the Bayesian network, we seek to gain a better understanding of the interrelationships between fire activity in different regions of the planet, highlighting the most important teleconnection patterns. We expect this exploratory analysis to aid in the development and testing of plausible hypotheses about the underlying mechanisms supporting these relationships.

How to cite: Bedia, J., Graafland, C. E., Gincheva, A., and Turco, M.: Global wildfire synchronicity patterns as revealed by complex network analysis, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-9608, https://doi.org/10.5194/egusphere-egu23-9608, 2023.

vNH.5
|
EGU23-804
|
ECS
|
Mira Shivani Sankar, Alka Singh, Nagesh K Subbanna, and Raian V Maretto

Wildfires are increasing tremendously in Northeast India mostly due to anthropogenic intervention. Indian states are prone to high-intensity fire events resulting in long-term impact on the forest ecosystem such as changes in the vegetation pattern and life cycle of species, decrease in certain species population, and quality of vegetation. Also, forest fires are known to influence the air pollution rate, and the pollutants generated through forest fire are way more complex when compared with urban air pollutants as the composition of forest fire pollutants (FFP) depend on the type of vegetation in the region. Additionally, volatile organic compounds (VOC’s), soot, ozone and black carbon which are few of the products from forest fire has the ability to travel far away from the source fire affecting different ecosystem at various ways and intensity.

As FFP can cause a decline in local, regional and global terrestrial productivity, a deep learning model will be useful in understanding and assessing the impact on vegetation health. Consequently, it is necessary to model the effects of the forest fires, so that their effects in both nearby and far off areas is understood.  In order to accomplish this, we employ Bi-directional Long Short-Term Memory (BiLSTM). The advantages of using a LSTM model are its ability to learn from Spatio-temporal series of data, avoid vanishing and exploding gradient problem, while being tractable to train. The inputs consist of concentrations of, aerosol, carbon monoxide, ozone, black carbon density, evapotranspiration, leaf area index, soil moisture, temperature and relative humidity obtained from MERRA – 2, MODIS and Sentinel – 5P satellite datasets accessed using google earth engine portal. Utilizing these datasets, the normalized vegetation index can be predicted. Standard techniques (mean squared error, root mean squared error, mean absolute error and mean absolute percentage error) are employed to determine the performance of the algorithm.

Keywords: forest fire, pollutants, BiLSTM, google earth engine, deep learning

How to cite: Sankar, M. S., Singh, A., Subbanna, N. K., and Maretto, R. V.: Assessment of the impact of the forest fire pollutants on vegetation and crop health in Northeast India, EGU General Assembly 2023, Vienna, Austria, 23–28 Apr 2023, EGU23-804, https://doi.org/10.5194/egusphere-egu23-804, 2023.