Open Research Exeter - University of Exeter
Not a member yet
41213 research outputs found
Sort by
Operational modal analysis of a ten-storey building featuring vertical modes
No full textAs a contribution to addressing challenges and improving technology in field implementation and practice of ‘operational modal analysis’ (OMA), a vibration test was carried out on a 10-storey rectangular building, featuring modal identification for both horizontal and vertical axes over a grid of 80 locations. The study used six triaxial accelerometers in 20 setups and the particular challenges with data are with close/buried modes, significant noise disparity, and contamination with harmonic signals in two frequency bands around 2 Hz and 8 Hz, which are addressed using recently developed techniques in Bayesian OMA. A particular feature of the building dynamic behaviour in the higher band of modes revealed by the study was a set of global vertical vibration modes. In addition to the short-term multi-setup test, two months of monitoring data were collected using a single accelerometer on the top floor to study amplitude dependence and ensemble statistics of modal properties, particularly for the first horizontal mode. The results reveal that the natural frequencies of the first three modes decrease with the vibration level while the damping of the first mode shows an increasing trend. The field data used in this study are archived in an open repository for public access.tion supplied</p
Tropical cyclone impact data in the Philippines: implications for disaster risk research
No full textNatural hazards such as tropical cyclones (TCs) cause widespread destruction. Historical impact data provides a resource for understanding TC impacts and associated societal vulnerabilities which is essential for building resilience. However, characteristics of impact data such as resolution and coverage can influence its utility for disaster risk reduction (DRR) applications. With this in mind, we present a province-level impact dataset for TCs in the Philippines between 2010 and 2020 for deaths, affected population, housing damage and economic loss curated for DRR applications. Specifically, we evaluate the effect of the dataset’s spatial resolution and its coverage of hazard intensities, impact magnitudes and impact types and discuss the implications for DRR applications. Considering the utility of impact data within the context of DRR is crucial, and a dataset with comprehensive coverage of impact and hazard magnitudes and appropriate spatial resolution is pivotal for DRR applications. The research presents a guide for others using this dataset and data more generally in DRR applications.</p
The Girandole: a framework to aid equality, diversity and inclusion-focussed reflection in nursing education
No full textPurposeThe need for authentic equality, diversity and inclusion (EDI) in nursing education and clinical practice is well documented, with many institutional initiatives developed over the past decade. Yet, despite these efforts, discriminatory behaviours toward nursing students and healthcare practitioners continue to surface—undermining institutional policies, professional codes and fundamental human values. This discursive paper presents a conceptual framework—the “Girandole” (French for the spinning child’s toy windmill or pinwheel)—designed to identify enablers and barriers to EDI in clinical learning environments. The paper also explores its application to current nursing placement experiences.Design/methodology/approachAn interdisciplinary, multi-cultural team of academics, clinicians and researchers, EDI leads, spontaneously united for a framework lab. The team activities and numerous EDI discussions led to co-designing the Girandole framework, which maps the domains influencing EDI in clinical education.FindingsThe Girandole framework symbolises the rotating forces that influence EDI in nursing placements. The direction of the rotation of the vanes of the Girandole indicates whether these are enablers or obstacles for culturally and inclusively competent care and education. It provides a lens to examine the hidden curriculum, cultural competence and lived experiences, while promoting active bystandership and organisational accountability.Originality/valueThis discursive paper offers a novel, visually engaging framework grounded in lived experience. The Girandole offers educators, mentors, students and institutions a practical tool to reflect on and improve EDI culture. Though developed in nursing, the Girandole is transferable to other healthcare professions, supporting inclusive and ethically grounded education.</p
The Colonial Origins of Israel’s Carceral Regime: Examining Colonial and Settler Colonial Applications of Carcerality in Palestine
No full textThe establishment of the state of Israel and its subsequent occupation of the Palestinian West Bank and Gaza Strip brought a long history of British colonial policy and practice in Palestine back to life. This resurrection, achieved through policy transfer, primarily involved carceral policy pertaining to police roles, the administration of prisons, the application of detention, traditions of torture, along with other punitive administrative measures, such as curfews and deportation. However, the (limited) scholarship frames this transfer merely as a continuation, with the result that it is predisposed to overlook the distinctions between the carceral practices of colonial and settler regimes. In addition, this neglects the role of Zionism within the Mandate framework, with the result that it is instead presented inversely, namely as the impact of colonialism on Israel and its policy in the oPt. This thesis, in examining the nuances of Mandate policies in Palestine and Israel’s adoption of colonial carceral policy and practice, analyses the eliminatory nature of carcerality within a Zionist settler programme in Palestine that spans over 140 years. It contends that, rather than existing as strictly separate categories, coloniality and settler coloniality were, in the British–Israeli case, also shaped by the Mandate’s commitments to Zionism, resulting in a more nuanced and complex mission in Palestine. While British carceral policy was indeed adopted by Israel and its occupation regime, this thesis proposes to instead focus on the intensification of these practices under a fully realised settler project which had nearly an additional five decades to develop. Further, building on Kelly Lytle Hernández’s work in the US, this thesis argues that carcerality functioned not only to suppress resistance but also as a mechanism of native elimination in Palestine, with this becoming especially evident during periods of popular uprising, such as the Great Revolt (1936–1939) and the First Intifada (1987-1993). Though distinctly situated, both uprisings were fundamentally struggles against Zionist erasure in its varying forms. A significant portion of this thesis therefore focuses on these two pivotal periods, which precipitated escalations in carceral violence and the expansion of carceral infrastructure. While this thesis acknowledges a slight shift in trajectory in British carceral policy following the rise of anti-Mandate Zionist militancy, (however still not at all comparable to the carceral subjugation inflicted upon Palestinians), it primarly focuses on the era when Mandate and Zionist coordination was intact and flourishing. This thesis advances our understanding of the expanse of carceral violence in its settler colonial application, and its utility as a tool for native elimination in Palestine, while also illuminating how carcerality has facilitated the advancement of both colonial and settler colonial programmes in Palestine. Crucially, it highlights overlooked nuances between these classifications in the Palestinian context. By engaging with the growing (yet still limited) scholarship on this subject in other colonial contexts – particularly the United States – it seeks to challenge and expand the discourse around this subject. Although this thesis centres its attention on the Mandate period (1920-–1948) and the early years of the Israeli occupation, its findings provide a framework for understanding contemporary Israeli carceral regime across occupied Palestine, particularly in light of the current, and ongoing, genocide in Gaza, much of which has taken place within settler spaces of captivity
Unravelling the Function of the Unusual Antioxidants Ergothioneine and Ovothiol in Plants and Photosynthetic Protists
No full textErgothioneine and ovothiol are histidine derived thiols, predicted to provide
antioxidant and cytoprotective roles. They have been identified in a small number
of photosynthetic organisms. However, their biosynthesis and function in vivo has
not been extensively explored. Metabolite profiling combined with genomic
surveys indicated that the specificity of the SAM-methyltransferase is an
important determinant of the biosynthetic pathway present, where the
streptophyta exclusively synthesize ergothioneine, and green and red algae
produce both ergothioneine and ovothiol. It has also demonstrated differences in
the biosynthetic pathways between major lineages of the Archaeplastida and
highlighted the loss of this pathway in the angiosperms. Knockout of the mpegt1
gene in the liverwort Marchantia polymorpha Tak-1 resulted in a reduction of
ergothioneine and high light-induced stress showed an increase in the rate of
development in the mutant strains. Knockout of the SAM-methyltransferase
domain in the diatom Phaeodactylum tricornutum resulted in a loss of ovothiol B,
but an accumulation of ovothiol A, an ovothiol B like compound, and ascorbate.
P. tricornutum ovothiol mutant strains showed higher specific growth rates under
high light compared to the wild type. This study has provided the first steps in the
characterisation of these sulphur-containing histidine derivatives and developed
tools including metabolite profiling techniques. It has also resulted in the
generation of knockout mutants in two model organisms. The mutant strains have
shown the involvement of histidine derived thiols in response to high light and
provides two new systems which can be utilised to allow their wide roles and
cellular functions to be more fully understood
Deterministic Extremes and Climate Tipping Points
No full textExtreme events are becoming an increasingly important field of study, which encom-
passes the study of several different phenomena. As such, different mathematical
formulations and frameworks exist in which extreme events can be defined and stud-
ied. In this thesis we focus on two different interpretations. The first is Extreme
Value Theory applied to dynamical systems, which understands extremes as large
fluctuations of a chaotic dynamical system. This framework allows to compute prob-
ability distributions that relate the probability of an event with its magnitude, and
encode in its parameters dynamical and geometrical quantities related to properties
of the dynamical system. The second is stochastically perturbed differential equa-
tions, in which the stochastic component eventually forces the system to transition
between stable states. This gives rise to noise-induced tipping, and to the study of
the statistical properties of the transitions.
An important application of Extreme Value Theory is to estimate two dy-
namical indicators, the local dimension and the extremal index, used to quantify
persistence in phase space. However, the mathematical theory underpinning these
estimations is insufficiently developed. Here we describe various algorithms used
for estimating the local dimension and study in detail the asymptotic laws corres-
ponding to Hitting a ball, and Recurring to a ball, in terms of their potential to
estimate the local dimension. While the theoretical studies up until now are mostly
concerned with Hitting, Recurrence is used in applications when only one realisa-
tion of the dynamics is available. We show that when the dynamical system under
study has an absolutely continuous invariant measure the algorithms work well and
are interchangeable. Also, error bounds are found for the algorithm based in ex-
ceedances over a threshold, when sufficiently fast decay of correlations is imposed
on the system.
The two dynamical indicators are typically computed using the exceedances
over a threshold, which turn to form a Generalized Pareto Distribution in the asymp-
totic limit of high threshold in many cases. Here we examine in detail issues that
arise when estimating these quantities for some known dynamical systems. We focus
2
on how the geometry of an invariant set can affect the regularly varying properties
of the invariant measure. We demonstrate that singular measures supported on sets
of non-integer dimension are typically not regularly varying and that the absence of
regular variation makes the estimates resolution dependent. We show as well that
the most common extremal index estimation method is not well defined for continu-
ous time processes sampled at fixed time steps, which is an underlying assumption
in its application to data.
Many physical and chemical phenomena are governed by stochastic escape
across potential barriers. The escape time depends on the structure of the noise
and the shape of the potential barrier. By applying α-stable noise from the α = 2
Gaussian noise limit to the α < 2 jump processes, we find a continuous transition
of the mean escape time from the usual dependence on the height of the barrier
for Gaussian noise, to a dependence solely on the width of the barrier for α-stable
noise. We consider the exit problem of a process driven by α-stable noise in a double
well potential. We study individually the influences of the width and the height of
the potential barrier in the escape time, and we show through scalings that the
asymptotic laws are described by a universal curve independent of both parameters.
When the dependence in the stability parameter is considered, we see that there
are two different diffusive regimes in which diffusion is described either by Kramer’s
time or by the corresponding asymptotic law for α-stable noise. We determine the
regions of the noise parameter space in which each regime prevails and exploit this
result to construct an anomalous example in which a double well potential exhibits
a different diffusion regime in each well for a wide range of parameters
Participatory Design of Citizen-led Remote Sensing Forensics Tools in Mexico
No full textIn Mexico, a state marred by intractable and extreme drug-related violence, the state is either through incompetence or collusion unable to investigate and prosecute the thousands of murders which take place annually and has lost its ability to control significant amounts of its territory. Citizens have taken the practice of investigation and forensics into their own hands, appropriating state-centric forms of vision and technologies of investigation for their own uses, as part of the forensic civism movement. The realities of this appropriation collide in unpredictable and sometimes difficult ways, producing both danger and controversy as the boundaries of expertise, vision, and danger are created and negotiated through the use of new technologies. The politics and power of vision, which is integral to the application and functioning of state power, favour the state and give it the power to construct an exclusionary and abstract view of the world. Participatory design, which contends with the distribution of power through engaging with citizens and stakeholders to create an artefact or other end product, can contest these boundaries of expertise and vision through creating new ways of seeing, and demonstrating the use of these technologies in a politically relevant way.
As a researcher, I have worked with them in a participatory design framework and have encountered opposition and pushback from authorities and scientists who seek to police the boundaries of ethical and scientific research. This paper explores the interaction between my role as a researcher, and different forms of vision as I work with forensic citizens and professionals in my research. The competing forms of vision and the expertise that gatekeeps them reveal different aspects of the mass grave and are related to a politics of measurement which leads the state to not see what it finds inconvenient. When forensic citizens seek to appropriate forms of vision which give them their own abstracting and obscuring powers, they are met with resistance from authorities ostensibly for their own good, with participatory design aiming to respond to this through opening up new forms of vision for forensic citizens, and new possibilities for existing forms of vision to be used in articulating key citizen rights.
In the practice of participatory design, new rights, futures, and possibilities are revealed and articulated in collaboration with forensic citizens. This thesis documents the process from interview to artefact and uses this artefact to understand and interrogate the processes which have produced this research and the challenges that face forensic citizens
Monitoring climate and land use impacts in African rangelands with machine learning and Earth observation
No full textRangelands cover two thirds of Africa and underpin the livelihoods of more than 200 million pastoralists and agro-pastoralists across the continent. African rangelands are under threat from climate change, changing land use patterns and the spread of invasive plant species, degrading their ability to provide vital ecosystem services. Yet significant uncertainties remain about the extent and impact of these changes across African rangelands.
This thesis develops a suite of statistical and machine learning approaches using large spatiotemporal Earth observation datasets to provide new insights into these challenges. Across three research chapters, I examine the relationship between vegetation gross primary productivity and a range of biophysical drivers to illuminate how rangelands are responding to climatic variability and anthropogenic pressures.
Chapter 2 analyses the relationship between vegetation productivity and biophysical conditions across the continent using generalised additive models, with a focus on precipitation variability. A robust prediction of climate modelling is that precipitation will become more variable over much of Africa under climate change, with an increase in drought and extreme rainfall. My research demonstrates that greater variability has a modest negative effect on productivity but that this is outweighed by the effects of total rainfall, temperature and soil conditions. The analysis also suggests a possible limit to how far rangeland vegetation can adapt to rising temperatures but finds that African savannas may have greater resilience than grasslands and shrublands.
Chapter 3 applies the findings of Chapter 2 to control for the effect of climatic variability and other biophysical conditions on rangeland productivity, providing a novel tool for detecting local management impacts on vegetation in dynamic rangeland environments. It presents a new application of quantile regression forests to model potential vegetation primary productivity as a function of biophysical conditions, and defines a new monitoring indicator, the relative productivity index (RPI), to describe the ratio of observed to potential productivity. Using rangelands in Kenya and Tanzania as a test case, it demonstrates that RPI outperforms three widely used existing methods to control for climatic influence on vegetation productivity and provides a common basis for assessing both spatial patterns and temporal trends.
Chapter 4 further refines the RPI method, incorporating new covariates and methodological changes to address the limitations identified in Chapter 3 and significantly improve performance. This chapter then tests the application of RPI to three East African case studies that examine (i) the relationship between RPI and traditional in situ indicators of rangeland condition, (ii) the impact of exotic shrub invasion on observed RPI patterns, and (iii) the value of RPI in improving impact evaluation for a disturbance affecting a national park in Kenya. The results suggest that RPI provides new and valuable information on rangeland vegetation change and has considerable potential for further scale-up and application in rangeland monitoring.
This thesis represents a significant advance in scientific understanding of African rangeland vegetation response to climatic conditions and offers a new method that addresses a core challenge of vegetation monitoring in these dynamic ecosystems. This demonstrates the potential of emerging Earth observation datasets and computational methods to greatly enhance monitoring of vegetation change in complex and variable rangeland environments
Diversifying the Canon of American First World War Poetry
No full textThe First World War had a profound effect on international literature. However, its poetic canon has remained stubbornly British. American war poetry has been excluded from the canon owing in part to vastly different war experiences. American forces did not begin to arrive in Europe until nearly three full years into the war, and many of those men would never see the front line or trench warfare. AEF involvement in the war was recorded in their poetry, which in many cases is not housed in archives belonging to the poets but is instead scattered in newspaper archives around the country. However difficult the research, this poetry provides illustrations of American experiences in the First World War era, and these poems challenge the dominant narrative of poetry that is British, white, and male.
African American First World War poetry reflects their unique experiences. Reviving a study of African American war poetry is rediscovering the truth of life for African Americans in the war era. For the purpose of reimagining a more historically inclusive canon, this thesis considers the poetry of African American veteran poet Lucian B. Watkins and African American citizen poet and songwriter Andy Razaf. Similarly, American women and LGBTQ soldiers had invisible experiences in the First World War that can be seen in their poetry. Charles Harding Divine, who was prolifically published during and after the war, and John Allan Wyeth, who invented a new form of the sonnet, will be analyzed. Aline Kilmer, widow of American soldier poet Joyce Kilmer, will serve as representative of American women owing to her prolific writing career and her widowhood. Each chapter will have a biographical-historical approach
Time Series Prediction: Novel Deep Learning Approaches for Sustainable Datacentres
No full textDatacentres are the key pillars of modern-day digital services but consume huge amounts of energy and characterise significant carbon emissions, largely due to inefficient resource utilisation that result from over-provisioning of computational resources.
Obtaining a prior knowledge about the execution profiles of workloads could facilitate an optimum provisioning of resources in a way that can reduce resource wastage whilst ensuring a smooth execution of the workloads.
Herein, accurately forecasting the execution profiles of workloads can significantly reduce unnecessary energy consumption.
However, achieving a reliable forecast of the execution profiles remains challenging under the complexity, variability and interdependencies that are inherent in contemporary datacentre workloads.
To promote datacentre sustainability, this thesis proposes novel forecasting methods to accurately predict the resource requirements of datacentre workloads.
First, existing workload forecasting techniques are critically reviewed to identify research gaps.
Based on this, three novel prediction models are proposed: Mixed Channel Time-series Dense Encoder (MC-TiDE) for predicting CPU and memory resource requirements; Series Rearrangement Time-series Dense Encoder (SeinE) tailored for predicting complex GPU workload requirements to compliment AI-driven workloads; and Correlation Matching and Patch Squeeze-and-Excitation Time-series Dense Encoder (ChaSE-TiDE), which is a multi-resource prediction model designed to capture complex workload patterns across different hardware platforms.
Notably, ChaSE-TiDE uses advanced correlation matching and multi-scale feature extraction, thereby extending its applicability to a wide range of time-series application domains.
Experiments on real-world datasets from Alibaba and Google Clouds demonstrate the superior performance of the proposed prediction models (achieving the best average prediction accuracy) against state-of-the-art approaches including TiDE, LSTM, PatchTST etc., in terms of Mean Squared Error and Mean Absolute Error.
These results validate the dependability of the proposed prediction models in contributing to sustainable datacentre operations whilst maintaining performance quality via a proactive management of datacentre workloads, thereby offering a wide-range of industrial and environmental benefits