Professor of Geograhical Information Science in the Department of Geography, University of Manchester
Bio: After gaining an MSc in GIS in 1993, Sarah applied GIS to the task of developing urban atmospheric emissions inventories. Since then she has worked primarily in multi-disciplinary teams bringing a spatial perspective to application areas which range from air pollution epidemiology and climate science, through to vulnerability assessment and the health and wellbeing benefits of urban ecosystems
Beyond the buffer? A short history of spatial analysis in exposure assessment. Spatial methods are used to help support both deductive and inductive approaches to understanding disease incidence and their causal factors. A review in 2012 identified that proximity analysis remained the most prominent spatial method used in epidemiology in the previous decade (Auchincloss et al., 2012), but how far does the field still rely on the basic buffer? This talk will review the role of the humble buffer and consider whether it is destined to remain the primary basis for exposure assessment in the decades to come
Assistant Professor in the Department of Geography, University of California, Santa Barbara
Bio: Somayeh's research focuses on understanding and modelling of movement in human and ecological systems. She develops computational data analysis, knowledge discovery, modelling, and visualization techniques to study how movement patterns are formed in dynamic systems across spatial and temporal scales
Data-driven movement analytics for pandemic responses. This talk discusses various techniques to develop data-driven insights on human movement patterns both at individual and aggregate levels to inform non-pharmaceutical intervention actions in a pandemic such as the ongoing COVID-19 pandemic. First, I review our systematic comparative analysis of exiting mobility indices that are used to inform COVID-19 responses across the US. Second, I will discuss a new time-geographic based approach to tracing space-time contacts and social interactions that might be critical in a pandemic situation
Distinguished Professor of Spatial Data Science in the Lancaster Environment Centre, University of Lancaster
Bio: Peter's research involves the application of space-time statistics and geostatistics, machine learning and AI, and dynamic numerical modeling, to Earth observation (EO) and other spatio-temporal data, to answer a wide range of science and social science questions.
Joint Deep Learning – a novel paradigm for simultaneous land use and land cover classification.It is well known that land cover classes, which represent the land state, can be predicted directly from remotely sensed imagery whereas land use, which represents a higher-order function, must generally be classified based on spatial context. It is no wonder then that deep learning has found application for the prediction of land use. However, land use is commonly built on land cover building blocks at a lower ontological and spatial level than land use, opening the possibility for their joint prediction. This talk presents ‘Joint Deep Learning’ which is the first algorithm specifically to exploit directly the ontological interdependence between land cover and land use. Moreover, it extends statistical joint distribution modelling to include a variable that is defined as a higher-order representation (aka. predicted by deep learning). This new realization and concept has the potential to completely revolutionize how land use and land cover classification is done in remote sensing.
All participants are invited to register for FREE using the button below. Registration link will remain open until the 14th April 2021
If you have any enquiries, please contact the conference chair, Prof. Scott Orford, at . You can keep up to date with conference information by checking back to this website, signing up to our mailing list, and following us on Twitter.
GISRUK Organizing Committee and Prof. Scott Orford (Cardiff University)