Spatial and Temporal Assessment of Vulnerability and Resilience in Semi-Arid Landscapes

点击翻译按钮获取中文摘要

Food security, particularly in dryer more climate sensitive areas of the world, is becoming increasingly unstable. Greater understanding of human and biological elements of agricultural and environmental systems, particularly in dryland areas, could help mitigate negative impacts of climate change on food security whilst simultaneously addressing issues of environmental degradation. However, biological systems, agricultural systems and the interplay between the two are complex and non-linear and, at present, available methods are not able to fully and simultaneously capture all dynamic and static elements of these integrated systems (Inwood et al., 2018). The aim of this PhD is to provide the means for landscape scale simultaneous assessment of agricultural and environmental systems, and their interplay, through development of a novel methodology and its application to open-source geo-spatial data. The objectives are three-fold: 1) Assess and adapt methodology for understanding the multivariate spatial-temporal interplay between elements of a complex system; 2) Apply the method to real-world data within two active study-sites and assess its effectiveness; 3) Perform 'what-if?' scenarios to identify leavers of change and 'tipping points' within the system. The geographical focus of this research are two study sites, one in Taita-Taveta hills, south-east Kenya, and the second in Yebelo, southern Ethiopia. Methodological efficacy will be established through mathematical proofs as well as testing in a real world environment via ground-truthing in each of the study sites. Ground-truthing will be informed by site visits and over 1000 RHoMIS household surveys collected and analysed for each study site. 'What-if?' scenarios includes changes in environmental factors such as changes in meteorological activity, increases/decreases in wildlife, changes in forest cover etc, as well as human, such as population change and policy interventions. Simultaneous multivariate spatial-temporal systems analysis is an unsolved problem (Chen et al., 2020). Multiple existing models are available within the latent variable, neural network and Bayesian hierarchical fields. A detailed literature review will be undertaken to identify the most likely path to success.