Tuesday 28 Nov 2017, Nov Icelandic and Chilean Volcanoes anticipating future eruptions
Speaker: Dr Dave McGarvie Affiliation: Senior Lecturer in Volcanology, The Open University in Scotland Abstract: Since the extensive ice sheets in Iceland and Chile melted around 10-15 thousand years ago, eruptions from individual volcanoes have either spread ash across the landscape or produced lavas both of which are valuable time markers. Ash layers are particularly useful, as they are widespread and they cover old land formed before the eruption and underlie new land created afterwards. As carbon dating yields accurate ages, we can build a stratigraphy/timeline of how dynamic volcanic landscapes evolve. However, in the period prior to this, when ice sheets covered the land, it is simply not possible to establish such a good stratigraphy/timeline. Any ash deposited on the ice is simply removed when the ice melts. So, we are left with a huge information gap. To fill this gap, I undertake fieldwork on young volcanic rocks that show unequivocal evidence of erupting into ice, some of which give clear indications of minimum ice thickness. Remarkably, these provide the only physical evidence of the thickness of vanished ice sheets: they are unique. When we then use expensive and difficult techniques to get ages for these eruptions, we can say that at location X at time Y the ice sheet thickness was Z. This provides the hard evidence that is lacking from statistical models of ice sheet growth and decay. Another aspect of my work is exploring poorly-known volcanoes and their eruptions, as a further means of anticipating what might happen in the future. With this in mind I have been working on an unstudied volcano in southern Chile (Quetrupillán the 'headless spirit'), Tindfjallajökull in Iceland (near neighbour of the infamous Eyjafjallajökull), and the 1362 AD eruption of Öraefajökull, that produced Iceland's largest explosive eruption since Iceland was settled c.1100 years ago. Though it will be a challenge, in my talk I will try and cover all of these aspects of my ongoing research.
Tuesday 7 Nov 2017, Monitoring the World - the sustainable development goals and what they mean to statisticians.
Speaker: Phil Crook Affiliation: The secretary of the RSS International Development Section and a consultant development statistician Abstract: In September 2015 the United Nations General Assembly adopted seventeen aspirational "Global Goals”, stretching from “No Poverty” and “Zero Hunger” through “Gender Equality” and “Decent work” to “Climate Action” and “Peace, Justice and Strong institutions”. The goals are associated with 169 targets and 241 global indicators and the annual report on progress towards the targets will cover every country, developed as well as developing. Data disaggregation to ensure that no-one is left behind will be a key feature of reporting. Experience in the UK and globally shows that targets and indicators such as these can have profound effects. The talk will cover the process and politics leading up to global goals, the experience with the Millennium Development Goals which preceded the SDGs, and then the SDGs themselves and the statistical challenges and opportunities they are throwing up.
Event in Oct Tuesday 3 Oct 2017, New Mobility Perspectives in the era of Wearable Sensors
Speaker: Dr. Katarzyna Sila-Nowicka Affiliation: School of Social & Political Sciences, University of Glasgow Abstract: Human mobility is vital for understanding expansion processes in urban areas, the establishment of transportat\ ion services and the spatial distribution of facilities. Until recently, exploring human mobility in detail wa\ s challenging because personal trip data collection methods consisted of expensive and time consuming methods.\ The development of sensors such as GPS trackers or other wearable devices equipped with various sensors that \ capture movement data in real-time and at detailed spatial and temporal scales has transformed our ability to \ collect mobility data. However, even though GPS trackers record an individual's location and movement very acc\ urately, they do not record essential characteristics of travel behaviour such as travel mode or trip purpose.\ What can be done to enhance raw movement trajectories in order to uncover these mobility characteristics? An\ d what can be done to publish the data without breaching individuals’ privacy?
Event in May Tuesday 16 May 2017: Modelling Exposure to Air Pollution at Population scale - challenges in accounting for spatio-temporal variability
Speaker: Dr. Stefan Reis Affiliation: Science Area Lead 'Biosphere-Atmosphere Interactions', NERC Centre for Ecology & Hydrology Abstract: The quantification of population level exposure to ambient air pollution is typically based on static population distributions inferred from the postcode of residence, and annual average concentrations of priority air pollutants at that location. Contemporary atmospheric chemistry transport models (ACTMs) are capable of providing air pollution concentration fields with high spatial and temporal resolution, for instance generating 1 km * 1 km gridded maps at hourly time steps. In addition, the 2011 UK Census has generated a novel population data product accounting for the location of the workplace, the so-called 'workday population' distribution, which accounts for the location of individuals during working hours. This 'workday' or 'daytime' population maps include usual residents, as well as others present in an output area during regular working hours and does complement the regular maps of residents, without double-counting of populations. In my talk, I will discuss how accounting for personal mobility (e.g. commuting to place of work or study, or spending time away from the main residential address) affects the exposure of the UK population to selected priority air pollutants (fine particulate matter, PM2.5; ground level ozone, O3; nitrogen oxides, NOx). The insights gained should inform the discussion on how we can address the challenges of spatial and temporal variability of both pollution fields and the population exposed. Finally, I will introduce further challenges in relation to time spent in different micro-environments, and the potential for integrating modelling and sensor date to advance our understanding of, and improve our ability to upscale from personal to population level.
Event in March Tuesday 28 March 2017: The 2021 Census: what is happening north and south of the border?
Speaker 1: Amy Wilson Affiliation: Census Director, Scotland Census 2021, National Records of Scotland. Speaker 2: Garnett Compton Affiliation: Head of Census Statistical Design and Outputs, Census 2021, Office for National Statistics. Abstract: This joint presentation will provide attendees with an update of work and progress towards the next census in 2021, focussing on Scotland and England & Wales. It will touch on the importance of harmonisation balanced against the need for local flexibility and consider the likely key differences from previous censuses and challenges ahead. The single biggest key difference from previous censuses will be the extent to which digital online technology is used in the census operation; presenting both opportunities and challenges. Amy and Garnett will then provide an update about the work taking place to determine what topics and questions should be included in 2021. They will also provide an update about some of the statistical methodology being considered such as around data processing, disclosure control and their importance and impact on the 2021 Census outputs.
Event in February
Tuesday 14 February 2017: Statistics and Commercial Housing
Speaker 1: Dr. Raffaella Calabrese Affiliation: University of Edinburgh Business School Title: A mortgage scoring model with spatial contagion Abstract: Due to the spatial dependence of house price, mortgage scoring models often display spatial dependence but the vast majority of the models used in the literature assume that mortgage defaults are independent. Only recently Zhu and Pace (2014) find that allowing spatial dependence greatly improved the predi ctive accuracy of mortgage scoring model. The main drawback of the model used by Zhu and Pace is that it is built under the assumption that the number of defaulted and non-defaulted mortgages is almost the same. To overcome this drawback, we propose a spatia l regression model for imbalanced binary data sets. We apply the proposed spatial model to a large data set (almost 80,000 observations) from Las Vegas in Nevada. Speaker 2: Vijay Krishnaswamy Affiliation: True North Partners LLP, a specialist consulting boutique Title: Model risk in unusual corners: the case of home price indices Abstract: This talk will cover the role of models in indices, using the example of home price indices. Such indices play a key role in the financial markets for mortgages and are widely used by banks. The objective of this talk is to create better awareness of the model risk that can arise from the use of home price indices.