Home > Research > Publications & Outputs > Bayesian spatial modelling of environmental and...

Electronic data

  • 2019chaconphd

    Final published version, 5.7 MB, PDF document

    Available under license: CC BY-ND: Creative Commons Attribution-NoDerivatives 4.0 International License

Text available via DOI:

View graph of relations

Bayesian spatial modelling of environmental and health data with applications in Brazilian Amazonia: linking environment and health in disadvantaged groups

Research output: ThesisDoctoral Thesis

Publication date2019
Number of pages163
Awarding Institution
  • Lancaster University
<mark>Original language</mark>English


Impacts of climate change on human health are a major concern for public health. Increase in frequency and intensity of extreme hydro-climatic events (floods and droughts) is one of the main characteristics of climate change. The occurrence of these events can drastically affect the lives of the population through different pathways. For example, by affecting accessibility to sufficient, safe and nutritious food (food security), increasing levels of malnutrition or increasing disease incidence. We hypothesize that nutrition might be a relevant pathway through which extreme hydro-climatic events affect human health and that the impacts are worse for vulnerable groups where they exacerbate existing vulnerabilities. Then, to understand and evaluate the effects of extreme hydro-climatic events on human health, we developed three studies. First, we propose a model-based standardized index to identify and quantify extreme temporal events and compared it against the classical standardized precipitation index (SPI). We found that our index holds the properties of the SPI, but improves on the methodology by tackling some of its limitations. Second, we used the model-based standardized index to evaluate the effects of exposure to extreme hydro-climatic events during pregnancy on birth weight. We controlled for other social and placed-based factors that could influence birth weight and found out that floods could significantly reduce birth weight. We also detected characteristics of vulnerable groups where birthweight is expected to be lower. Finally, we proposed our denominated spatial item factor analysis to model and predict spatially structured latent factors. With our application on predicting food insecurity in a roadless city of the Brazilian Amazonia, we discover that severely food insecure areas were related to flood-prone, poor and marginalized neighbourhoods. In general, our results highlight the importance of policies to reduce the effects of extreme hydro-climatic events on vulnerable populations of the Brazilian Amazonia. Although our methods were motivated by the study of the impacts of extreme hydro-climatic events, they can be applied in more general cases.