Search

Fire weather indices are used to assess the effect of weather on wildfire behaviour and to support fire management. Previous studies identified the high daily severity rating percentile (DSRp) as being strongly related to the total burnt area (BA) in Portugal, but it is still poorly understood how this knowledge can support fire management at a smaller spatial scale. The aims of this study were to (1) assess whether the 90th DSRp (DSR90p) threshold is adequate for estimating most of the BA in mainland Portugal; (2) analyse the spatial variability of the DSRp threshold that explains a large part of BA, at higher resolution; and, (3) analyse whether vegetation cover can justify the DSRp spatial variability. We used weather reanalysis data from ERA5-Land, wildfire and land use data from Portuguese land management departments for an extended summer period (15 May to 31 October) from 2001 to 2019. We computed and related DSRp to large wildfires (BA > 100 ha) and land use to clarify the effectiveness of the DSRp for estimating BA in Portugal and assess how vegetation influences it. Results revealed that the DSR90p is an adequate indicator of extreme fire weather days and BA in Portugal. In addition, the spatial pattern of the DSRp associated with most of the total BA shows variability at the municipality scale. Municipalities where large wildfires occur with more extreme weather conditions have most of the BAs in forests and are in coastal areas. By contrast, municipalities where large wild fires occur with less extreme weather conditions are predominantly covered by shrublands and are situated in eastern and inland regions. These findings are a novelty for fire science in Portugal and should be considered by fire managers and fire risk assessors.

The parish of Alvares, in the municipality of Góis, was strongly affected by the fires of June 2017, with 70% of its area burned. To improve the protection of local communities and increase their response capacity, it is critical to assess the exposure of settlements to large wildfires. In this sense, the exposure level of the settlements located in the parish was evaluated, combining the density of the exposed elements, with probability to burn. Exposure level was calculated for two dimensions: structural (based on the number of times burned since 1980) and dynamic (based on simulations of the probability to burn considering different fuel management strategies). Social vulnerability was also estimated, taking into account the sociodemographic characteristics of the residents, as well as the characteristics and conditions of the buildings. The results suggest that: 1) the settlements located in the north and northeast of the parish are more exposed to large wildfires; 2) almost half of these have high or very high vulnerability; 3) the fraction of settlements with very high exposure decreases up to 80% with the implementation of a fuel break network. These results, at the level of settlements, can contribute to a better hierarchy of protection and emergency actions, adjusting the fire prevention and mitigation strategies to the characteristics of the population and landscape., La parroquia de Alvares, en el municipio de Góis, se vió muy afectada por los incendios de junio de 2017, ya que quemó el 70% de su área. Para mejorar la protección de las comunidades locales y aumentar su capacidad de respuesta, es esencial evaluar el nivel de exposición de las aldeas a grandes incendios. En este sentido, se evaluó el nivel de exposición de las aldeas parroquiales, combinando la densidad de los elementos expuestos, con una probabilidad de arder. El nivel de exposición se calculó en dos líneas: estructural (basado en el número de veces quemadas desde 1980) y dinámico (basado en simulaciones de probabilidad de combustión para diferentes escenarios de gestión de combustible). También se estimó la vulnerabilidad social, teniendo en cuenta las características sociodemográficas de los residentes, así como las características y condiciones de mantenimiento de los edificios. Los resultados sugieren que: 1) las aldeas ubicadas en el norte y noreste de la parroquia están más expuestas a incendios rurales; 2) casi la mitad de las aldeas tienen una vulnerabilidad alta o muy alta; 3) la fracción de las aldeas con exposición muy alta disminuye hasta un 80% con la implementación de carriles de gestión de combustible. Estos resultados en toda la parroquia pueden contribuir a una mejor jerarquía de protección y acciones de emergencia, ajustando las estrategias y medidas de prevención y mitigación de incendios a las características de la población y del paisaje., A freguesia de Alvares, no concelho de Góis, foi muito afetada pelos incêndios de junho de 2017, tendo ardido 70% da sua área. Para melhorar a proteção das comunidades locais e aumentar a sua capacidade de resposta, é fundamental avaliar o nível de exposição das povoações a grandes incêndios. Neste sentido, foi avaliado o nível de exposição das povoações da freguesia, combinando a densidade dos elementos expostos com probabilidade de arder. O nível de exposição foi calculado em duas vertentes: estrutural (baseado no número de vezes ardido desde 1980) e dinâmico (baseado em simulações de probabilidade de arder para diferentes cenários de gestão de combustíveis). Também foi estimada a vulnerabilidade social, tendo em conta características sociodemográficas dos residentes, assim como as características e condições de manutenção dos edifícios. Os resultados sugerem que: 1) as povoações localizadas a norte e nordeste da freguesia estão mais expostas a incêndios rurais; 2) quase metade das povoações apresentam vulnerabilidade elevada ou muito elevada; 3) a fração das povoações com exposição muito elevada diminui até 80% com a implementação de faixas de gestão de combustível. Estes resultados, à escala da povoação, podem contribuir para uma melhor hierarquização de ações de proteção e emergência, ajustando as estratégias e medidas de prevenção e mitigação de incêndios às características da população e da paisagem.

Doutoramento em Engenharia Florestal e dos Recursos Naturais - Instituto Superior de Agronomia Wildfires pose real threats to life and property. In Portugal, the recent year of 2017 had the largest burnt area extent and number of casualties. A knowledge gap still exists in wildfire research related with better understanding individual wildfires, which has important implications for fire suppression, management, and policies. Wildfire spread models have been used to study individual wildfires, however, associated uncertainties and the lack of systematic evaluation methods hamper their capability for accurately predicting their spread. Understanding how fire spread predictions can be improved is a critical research task, as they will only be deemed useful if they can provide accurate and reliable information to fire managers. The present Thesis proposes to contribute to improve fire spread predictions by: i) Developing a methodology to systematically evaluate fire spread predictions ii) Thoroughly characterizing input data uncertainty and its impact on predictions; iii) Improving predictions using data-driven model calibration. The spread of large historical wildfires were studied by combining satellite data and models. The major findings of the present Thesis were: i) Satellite data accurately contributed to provide accurate fire dates and ignition information for large wildfires. ii) The evaluation metrics were very useful in identifying areas and periods of low/high spatio-temporal agreement, highlighting the strong underprediction bias and poor accuracy of the predictions. iii) Uncertainties in wind speed and direction, fuel model assignment and typology, location and timing of ignitions, had a major impact on prediction accuracy. iv) Predictions iii) Uncertainties in wind speed and direction, fuel model assignment and typology, location and timing of ignitions, had a major impact on prediction accuracy. iv) Predictions were improved by ‘learning’ from past wildfires, significantly reducing the impact of data uncertainty on the accuracy of fire spread predictions Overall, the work contributed to advance the body of knowledge regarding individual wildfires and identified future research steps towards a reliable operational fire system capable of supporting more effective and safer fire management decisions with the aim of reducing the dramatic impacts of wildfires N/A

Soil water repellency can impact soil hydrology, overland flow generation and associated soil losses. However, current hydrological models do not take it into account, which creates a challenge in repellency-prone regions. This work focused on the adaptation for soil water repellency of a daily water balance model. Repellency is estimated from soil moisture content using site-specific empirical relations and used to limit maximum soil moisture. This model was developed and tested using approximately 2 years of data from one long-unburned and two recently burned eucalypt plantations in northern Portugal, all of which showed strong seasonal soil water repellency cycles. Results indicated important improvements for the burned plantations, with the Nash-Sutcliffe efficiency increasing from -0.55 and -0.49 to 0.55 and 0.65. For the unburned site, model performance was already good without the modification and efficiency only improved slightly from 0.71 to 0.74, mostly due to the better simulation of delayed soil wetting after dry periods. Results suggested that even a simple approach to simulate soil water repellency can markedly improve the performance of hydrological models in eucalypt forests, especially after fire.

Soil water repellency can impact soil hydrology, overland flow generation and associated soil losses. However, current hydrological models do not take it into account, which creates a challenge in repellency-prone regions. This work focused on the adaptation for soil water repellency of a daily water balance model. Repellency is estimated from soil moisture content using site-specific empirical relations, and used to limit maximum soil moisture. This model was developed and tested using c. 2 years of data from one long-unburned and two recently burned eucalypt plantations in northern Portugal, all of which showed strong seasonal soil water repellency cycles. Results indicated important improvements for the burned plantations, with the Nash-Sutcliffe efficiency increasing from -0.55 and -0.49 to 0.55 and 0.65. For the unburned site, model performance was already good without the modification and efficiency only improved slightly from 0.71 to 0.74, mostly due to the better simulation of delayed soil wetting after dry periods. Results suggested that even a simple approach to simulate soil water repellency can markedly improve the performance of hydrological models in eucalypt forests, especially after fire.

The dependence on satellites to derive burnt area (BA) maps is unquestionable. High resolution inventories normally result from change detection algorithms applied to pre and post fire season high resolution imagery. But these have no temporal discrimination within the occurring season. Limited to the larger fire scars, coarser resolution imagery based on reflectance or thermal information can help to map the individual fire progression. The Moderate Resolution Imaging Spectroradiometer (MODIS) 250m imagery bands, freely available, can be used to provide quick areal estimates and provide the needed temporal discrimination with four times the standard spatial resolution BA products. The scope of this study is to assess the spatial and temporal accuracy of burnt area maps derived by the MODIS 250m resolution Burnt Area algorithm (M250BA) presented by Mota et al., (2013) on an Mediterranean landscape. The algorithm is an improved adaptation of one of the burnt area algorithms developed within the scope of the Fire_CCI project and was applied to an area covering continental Portugal for the period of 2001-2013. The algorithm comprises a temporal analysis based on change point detections and a spatial analysis based on Markov random fields. We explored the benefits of applying standard optimization techniques to the algorithm and achieved significant performance improvements.. Temporal and spatial accuracy assessments were performed by comparing the results with spatial and temporal distribution of active fire maps and with high resolution burnt area maps, derived by the MCD14ML thermal anomalies dataset and by Landsat BA classifications, respectively. Accuracy results highlight the potential applications for this BA algorithm and the advantages of using 250m spatial resolution images for BA detection. The study also extends the current national burnt area atlas since 2010. Due to the open-access data policy, the algorithm can be easily parameterised and applied to any region of the globe with potential high accuracy. The finer resolution combined with the flexibility of the algorithm is an advantage, not only to detect smaller burnt scars in regions where fuel load can be high, but also to fill the temporal gap between the Fire_CCI project MERIS BA product and future BA products derived from optical sensor imagery onboard the Proba-V and Sentinel-3 satellites.