| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 4.03 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Em Portugal, as Linhas de Muito Alta Tensão, geridas pela REN – Redes Energéticas
Nacionais, S.A., estão sujeitas a Planos de Monitorização cujo um dos objetivos é o de se
estudar o impacte ambiental associado à mortalidade de aves por colisão com os cabos
aéreos. Os planos incluem a quantificação de cadáveres de aves por baixo das linhas mas
sabe-se que estes números diferem da mortalidade real, uma vez que os cadáveres podem
ser removidos por predadores antes da sua contabilização. Assim, para considerar este
fator de enviesamento, estes planos contemplam a realização de ensaios experimentais,
designados por testes de remoção, que permitem obter dados sobre o tempo de persistência
do cadáver no solo até à sua remoção, podendo estes ser dados censurados. A realização
destes ensaios a cada nova linha elétrica implica, contudo, avultados custos financeiros e
logísticos.
Neste âmbito, a "Cátedra REN em Biodiversidade" compilou informação sobre 30
ensaios, realizados em Linhas de Muito Alta Tensão em Portugal Continental, dando
origem a dados geo-referenciados. Assim, neste estudo pretende-se construir um modelo
que permita estimar,para qualquer ponto geográfico do território nacional,a probabilidade
de persistência de um cadáver num determinado intervalo de tempo, obviando assim a
realização de testes a cada novo projeto.
Para responder a este objetivo, foi usada a abordagem INLA (Integrated Nested Laplace
Approximation) conjugada coma abordagem via equações diferenciais parciais estocásticas
(Stochastic Partial Differential Equations) e com análise de sobrevivência, o que permitiu
modelar a probabilidade de persistência de um cadáver considerando quer efeitos fixos
(tamanho do cadáver e época do ano) quer efeitos aleatórios (localização geográfica e
projeto).
Os resultados permitiram analisar a variação desta probabilidade para todo o território
nacional e criar uma ferramenta para utilizadores comuns que estima a probabilidade de
persistência, para uma localização específica, como função das covariáveis consideradas.
In Portugal, transmission power lines managed by REN - Redes Energéticas Nacionais, S.A., are subject to monitoring programmes whose one of the objectives is to assess the impact of bird mortality caused by collisions with overhead cables. These programmes include surveys to quantify bird carcasses under the lines, but these numbers are known to differ from actual mortality since carcasses can be removed by predators before they are counted. To account for this source of bias, these programmes also typically include the performance of field experiments, called carcass removal trials, which allow to obtain censored data on the persistence time of the carcass in the field until its removal. However, carrying out these trials on each new power line entails significant financial and logistical costs. In this context, the "REN Chair in Biodiversity" compiled information on 30 trials, carried out on transmission lines in mainland Portugal, resulting in geo-referenced data. Therefore, in this study we intend to build a model that can estimate, for any geographical point of the national territory, the persistence probability of a carcass for a given time interval, thus eliminating the need for field trials in every new project. To achieve this goal, the INLA (Integrated Nested Laplace Approximation) approach was used combined with the SPDE (Stochastic Partial Differential Equations) approach and survival analysis, which allowed to model the probability of persistence of a carcass considering both fixed effects (carcass size and season) and random effects (geographic location and project). The results allowed to analyze the variation of this probability for the entire national territory and to create a tool for common users that estimates the persistence probability of a carcass, for a specific location, as a function of the covariates considered.
In Portugal, transmission power lines managed by REN - Redes Energéticas Nacionais, S.A., are subject to monitoring programmes whose one of the objectives is to assess the impact of bird mortality caused by collisions with overhead cables. These programmes include surveys to quantify bird carcasses under the lines, but these numbers are known to differ from actual mortality since carcasses can be removed by predators before they are counted. To account for this source of bias, these programmes also typically include the performance of field experiments, called carcass removal trials, which allow to obtain censored data on the persistence time of the carcass in the field until its removal. However, carrying out these trials on each new power line entails significant financial and logistical costs. In this context, the "REN Chair in Biodiversity" compiled information on 30 trials, carried out on transmission lines in mainland Portugal, resulting in geo-referenced data. Therefore, in this study we intend to build a model that can estimate, for any geographical point of the national territory, the persistence probability of a carcass for a given time interval, thus eliminating the need for field trials in every new project. To achieve this goal, the INLA (Integrated Nested Laplace Approximation) approach was used combined with the SPDE (Stochastic Partial Differential Equations) approach and survival analysis, which allowed to model the probability of persistence of a carcass considering both fixed effects (carcass size and season) and random effects (geographic location and project). The results allowed to analyze the variation of this probability for the entire national territory and to create a tool for common users that estimates the persistence probability of a carcass, for a specific location, as a function of the covariates considered.
Descrição
Palavras-chave
Análise de sobrevivência Dados espaciais INLA Probabilidade de persistência
