| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 4.97 MB | Adobe PDF |
Autores
Resumo(s)
Os movimentos de vertente representam o perigo natural mais frequente no arquipélago dos Açores, devido à sua origem e geomorfologia vulcânicas, à localização geográfica e ao enquadramento climático, caracterizado por elevados níveis de precipitação. Neste contexto, a compreensão dos fatores que controlam a ocorrência destes fenómenos e a sua evolução temporal constitui um elemento fundamental para a avaliação e mitigação do risco associado.
Este estudo baseia-se na base de dados documental NATHA (Natural Hazards in Azores), construída a partir da análise sistemática de fontes históricas e de arquivos de imprensa regional. Através desta base de dados, foi possível identificar e caracterizar os eventos de instabilidade geomorfológica ocorridos na ilha de São Miguel entre 1900 e 2020.
Foram catalogados 236 eventos, responsáveis por 82 vítimas mortais, 41 feridos, 305 desalojados e pela destruição total ou parcial de 66 edifícios. A análise da distribuição espacial revelou diferenças significativas entre os municípios da ilha, sendo o município da Povoação mais afetado, registando cerca de 59% do total de vítimas mortais. A precipitação foi responsável por cerca de 70% dos eventos catalogados, que ocorrem na sua maioria durante os meses mais húmidos do ano, entre novembro e março. A análise temporal revelou uma maior concentração de eventos após 1996, associada a alterações no regime de precipitação, demonstrando, pela primeira vez nos Açores, a influência das alterações climáticas na ocorrência de movimentos de vertente.
O estudo dos padrões e tendências da precipitação na ilha de São Miguel recorreu à análise de clustering hierárquico e a testes não-paramétricos de tendência. Foram identificados quatro clusters de precipitação homogéneos, que confirmam a existência de um forte efeito orográfico, em que a altitude desempenha o papel de principal
regulador da precipitação. Foi observada uma forte correlação positiva entre a altitude e a precipitação, com um aumento de aproximadamente 180 mm de Precipitação Média
Anual (PMA) por cada 100 m de altitude. A análise temporal mostrou reduções significativas na precipitação anual e sazonal, embora a frequência de dias com precipitação mais intensa tenha aumentado. Esta redução é particularmente crítica durante o outono e o inverno, com quebras máximas na precipitação entre 41% e 55%.
O estudo sugere que a fase predominantemente positiva da Oscilação do Atlântico Norte (NAO) no período mais recente contribuiu, pelo menos em parte, para esta tendência de decréscimo da precipitação. O estudo propõe uma abordagem inovadora para o desenvolvimento de limiares empíricos combinados de precipitação, integrando um limiar preparatório, relacionado com a saturação prévia do solo, e um limiar de desencadeamento, associado à precipitação no dia do evento. A validação demonstrou que a utilização deste limiar combinado apresenta um desempenho superior ao limiar preparatório isolado, nomeadamente na redução de falsos positivos e na melhoria da precisão preditiva. As
áreas com materiais vulcânicos mais friáveis e "jovens", como os vulcões das Sete Cidades e das Furnas, são mais suscetíveis, evidenciando limiares de precipitação normalizados mais baixos. Os valores mais elevados das probabilidades mensais de excedência dos limiares, estão associados aos meses de dezembro e janeiro. Verifica-se, igualmente, que tanto os eventos de precipitação intensa de curta duração como os de longa duração estão associados a períodos de retorno inferiores a cinco anos.
Os resultados fornecem uma caraterização integrada, detalhada e inovadora do perigo de movimentos de vertente e da sua relação com a precipitação na ilha de São Miguel,
estabelecendo uma base de conhecimento robusta para a otimização de estratégias de prevenção e gestão do risco no contexto das ameaças geoclimáticas atuais e futuras.
ABSTRACT: Landslides are the most frequent natural hazard in the Azores archipelago, due to its volcanic origin and geomorphology, geographical location, and climatic setting characterised by high levels of rainfall. In this context, understanding the factors that control the occurrence of these phenomena, and their temporal evolution is fundamental for the assessment and mitigation of the associated risk. This study is based on the documentary database NATHA (Natural Hazards in Azores), compiled from a systematic analysis of historical sources and regional press archives, to identify and characterise landslide events on São Miguel Island between 1900 and 2020. A total of 236 events were catalogued, responsible for 82 fatalities, 41 injuries, 305 displaced people, and the total or partial destruction of 66 buildings. Spatial analysis revealed significant differences among municipalities, with Povoação being the most affected, accounting for about 59% of the total fatalities. Rainfall was identified as the main triggering factor in approximately 70% of the events, which occur mostly during the wettest months of the year, between November and March. Temporal analysis revealed a greater concentration of events after 1996, associated with a change in the rainfall regime, demonstrating, for the first time in the Azores, the influence of climate change on landslides occurrence. Rainfall patterns and trends on São Miguel Island were further analysed using hierarchical clustering and non-parametric trend tests, identifying four homogeneous rainfall clusters and a strong orographic effect, with altitude as the main regulator of rainfall. A strong positive correlation between altitude and Mean Annual Precipitation (MAP) was observed, with an increase of approximately 180 mm per 100 m of elevation. Despite an increased frequency of intense rainfall days, total annual and seasonal rainfall display a marked decreasing trend, particularly during autumn and winter, with reductions between 41% and 55%. The study suggests that the predominantly positive phase of the North Atlantic Oscillation (NAO) observed in the most recent period has contributed, at least in part, to this decreasing trend in rainfall. The study proposes a new approach for the development of combined empirical rainfall thresholds. An innovative combined empirical rainfall threshold approach was proposed, integrating a preparatory threshold related to prior soil saturation with a triggering threshold associated with rainfall on the day of the event. Validation demonstrated that the use of this combined threshold provides better performance than the preparatory threshold alone, particularly in reducing false positives and improving predictive accuracy. The study shows that areas with more friable and “younger” volcanic materials, such as Sete Cidades and Furnas volcanoes, are the most susceptible to landslides, exhibiting lower normalised rainfall thresholds. The highest values of the monthly probabilities of threshold exceedance are associated with December and January. It is also observed that both short-duration intense rainfall events and long-duration rainfall events are associated with return periods of less than five years. Overall, this work provides a comprehensive, integrated, detailed, and innovative characterisation of the landslide hazard and its relationship with rainfall on the São Miguel Island, establishing a robust knowledge base for optimising risk prevention and management strategies under current and future geoclimatic threats.
ABSTRACT: Landslides are the most frequent natural hazard in the Azores archipelago, due to its volcanic origin and geomorphology, geographical location, and climatic setting characterised by high levels of rainfall. In this context, understanding the factors that control the occurrence of these phenomena, and their temporal evolution is fundamental for the assessment and mitigation of the associated risk. This study is based on the documentary database NATHA (Natural Hazards in Azores), compiled from a systematic analysis of historical sources and regional press archives, to identify and characterise landslide events on São Miguel Island between 1900 and 2020. A total of 236 events were catalogued, responsible for 82 fatalities, 41 injuries, 305 displaced people, and the total or partial destruction of 66 buildings. Spatial analysis revealed significant differences among municipalities, with Povoação being the most affected, accounting for about 59% of the total fatalities. Rainfall was identified as the main triggering factor in approximately 70% of the events, which occur mostly during the wettest months of the year, between November and March. Temporal analysis revealed a greater concentration of events after 1996, associated with a change in the rainfall regime, demonstrating, for the first time in the Azores, the influence of climate change on landslides occurrence. Rainfall patterns and trends on São Miguel Island were further analysed using hierarchical clustering and non-parametric trend tests, identifying four homogeneous rainfall clusters and a strong orographic effect, with altitude as the main regulator of rainfall. A strong positive correlation between altitude and Mean Annual Precipitation (MAP) was observed, with an increase of approximately 180 mm per 100 m of elevation. Despite an increased frequency of intense rainfall days, total annual and seasonal rainfall display a marked decreasing trend, particularly during autumn and winter, with reductions between 41% and 55%. The study suggests that the predominantly positive phase of the North Atlantic Oscillation (NAO) observed in the most recent period has contributed, at least in part, to this decreasing trend in rainfall. The study proposes a new approach for the development of combined empirical rainfall thresholds. An innovative combined empirical rainfall threshold approach was proposed, integrating a preparatory threshold related to prior soil saturation with a triggering threshold associated with rainfall on the day of the event. Validation demonstrated that the use of this combined threshold provides better performance than the preparatory threshold alone, particularly in reducing false positives and improving predictive accuracy. The study shows that areas with more friable and “younger” volcanic materials, such as Sete Cidades and Furnas volcanoes, are the most susceptible to landslides, exhibiting lower normalised rainfall thresholds. The highest values of the monthly probabilities of threshold exceedance are associated with December and January. It is also observed that both short-duration intense rainfall events and long-duration rainfall events are associated with return periods of less than five years. Overall, this work provides a comprehensive, integrated, detailed, and innovative characterisation of the landslide hazard and its relationship with rainfall on the São Miguel Island, establishing a robust knowledge base for optimising risk prevention and management strategies under current and future geoclimatic threats.
Descrição
Tese de Doutoramento, Geologia, 17 de julho de 2026, Universidade dos Açores.
Palavras-chave
Precipitação Parâmetros atmosféricos
Contexto Educativo
Citação
SILVA, Rui Filipe Fagundes. (2026). Spatio-temporal analysis of rainfall and its effect for landslide triggering in São Miguel Island (Azores Archipelago). Ponta Delgada: Universidade dos Açores, 2026. 135 p. Tese de Doutoramento em Geologia. Disponível em http://hdl.handle.net/10400.3/9031