CASA Seminar Series 2024-25

This talk provides a walkthrough over some applications and best practices for utilising Google Earth Engine (GEE) in geospatial analysis. GEE offers a powerful cloud-based platform for processing vast geospatial datasets and conducting complex analyses, but effectively harnessing its capabilities requires understanding key strategies and techniques. We will cover topics such as: effective exploration of image collections, sampling and conditioning when working with noisy optical and radar satellite imagery at scale, how to use advanced AI models with GEE, and a few more to leverage GEE for advanced geospatial research. 

Gennadii Donchyts is a Cloud Geographer at Google. He helps public and private sector customers to use Earth Engine and Google Cloud platforms across multiple industries in the EMEA region and beyond. With expertise in remote sensing and software engineering, he previously led research at Deltares as a principal scientist, focusing on Earth Observation and Earth Engine for water-related applications. Gennadii holds a PhD in remote sensing from Delft University of Technology, where he studied surface water mapping using open satellite imagery. He is a passionate coder and advocate for Earth Engine, geospatial technology, and machine learning. 

Register to Attend

With the recent rise in studies focused on characterising spatial non-stationarity in processes that vary across space, Geographically Weighted (GW) models have emerged as the de facto statistical toolkit for addressing such challenges. This has attracted a diverse range of users and developers from across various scientific disciplines. In this talk, Dr Lu will introduce several newly developed GW models and demonstrate their applications through a series of case studies from different fields. To support the widespread adoption of GW models in research, he will also present the R package ‘GWmodel’ alongside a newly developed standalone software, ‘GWmodelS,’ which is built upon the GWmodel framework. The standalone software offers several key advantages, including a user-friendly graphical interface, enhanced operational efficiency, and improved accessibility. These features are designed to encourage usage across a broad spectrum of users, regardless of their technical expertise. 

Dr. Binbin Lu holds a Ph.D. in Geocomputation from the National University of Ireland, Maynooth, where he was affiliated with the National Centre for Geocomputation. He is currently an Associate Professor at the School of Remote Sensing and Information Engineering, Wuhan University, China. Dr. Lu's research expertise spans geocomputation, spatial statistics, geographically weighted (GW) modelling, open-source GIS, R programming, and the analysis of spatio-temporal big data. As the main developer and maintainer of the GWmodel R package, he has integrated a wide range of localised techniques, such as GW regression and GW principal component analysis, providing valuable tools for spatial data analysis. 

Register to Attend

Nearly 110 million people are forcibly displaced people worldwide. However, estimating the scale and patterns of internally displaced persons in real time, and developing appropriate policy responses, remain hindered by traditional data streams. They are infrequently updated, costly and slow. Mobile phone location data can overcome these limitations, but only represent a population segment. Drawing on an anonymised largescale, high-frequency dataset of locations from 25 million mobile devices, we propose an approach to leverage mobile phone data and produce population-level estimates of internal displacement. We use this approach to quantify the extent, pace and geographic patterns of internal displacement in Ukraine during the early stages of the Russian invasion in 2022. Our results produce reliable population-level estimates, enabling real-time monitoring of internal displacement at detailed spatio-temporal resolutions. Accurate estimations are crucial to support timely and effective humanitarian and disaster management responses, prioritising resources where they are most needed. 

Francisco Rowe is Professor in Population Data Science and Lead of the Geographic Data Science Lab at the University of Liverpool. His areas of expertise are: human mobility and migration, spatial inequalities, and geographic data science. Francisco is interested in the use of data from digital sources to understand population movements in real time at high spatial and temporal resolutions and how the resulting insights can be used in decision making processes. Francisco is part of the United Nations (UN) Committee of Experts on Big Data and Data Science for Official Statistics. He works closely with the UN International Organization for Migration, particularly the Global Data Analysis Centre (GMDAC) and the Displacement Tracking Matrix (DTM) unit, and UN the Economic Commission for Latin America (CEPAL). 

Register to Attend

Population nowcasting is an emerging field that aims to estimate current population sizes and demographics for small areas. This is particularly impactful in otherwise data-scarce contexts such as crisis response or supporting official statistics in the absence of a census. In our digital world, there is an increasingly dizzying array of potential data sources for population nowcasting ranging from satellite remote sensing to digital traces generated from our internet and mobile phone activity. This seminar will present several papers highlighting innovative applications of population nowcasting from around the world including census support in Nigera, DRC, and Colombia as well as rapid-response humanitarian action in Ukraine and Gaza. Discussing future directions and opportunities for collaboration. 

Doug Leasure is a Senior Researcher and Data Scientist at the Leverhulme Centre for Demographic Science in the Department of Population Health at the University of Oxford. Doug's interdisciplinary background brings together concepts and methods from demography, ecology, remote sensing and GIS, underpinned by methods in Bayesian statistics and machine learning. He currently leads the Oxford NowPop Project which develops methods in Bayesian Statistics for population nowcasting to support targeted humanitarian action in Ukraine, Gaza, and beyond. Prior to joining the University of Oxford, Doug led the Spatial Statistical Population Modelling team in the WorldPop Research Group at the University of Southampton. After completing his PhD in Population Ecology at the University of Arkansas, he held a postdoctoral research position at the Odum School of Ecology working on a NASA ecological forecasting project. 

Register to Attend

Adama is the second most populous city in Ethiopia and experiences frequent flash floods that have a detrimental impact on the community’s livelihood. To this effect, this study emphasizes the significance of conducting a comprehensive investigation to identify flood-resilient neighbourhoods in Adama City.

By considering the existing spatial pattern and morphology that are resilient to urban flooding, the city can enhance its flood management strategies in the future. The findings of this study demonstrate varying levels of spatial connectivity within Adama City. Detailed examination of two neighbourhoods identified through space syntax analysis with high and low spatial connectivity revealed important insights into flood resilience.

The high-connective neighbourhoods exhibited well interconnected street systems with manageable street spacing, facilitating efficient runoff flow and effective flood management during flooding events. These neighbourhoods also had shorter block sizes with frequent intersections, promoting better water drainage and reducing the risk of flooding during heavy rain events. The grid pattern observed in these areas allowed for efficient water runoff through multiple drainage paths, including the street surfaces. On the contrary, neighbourhoods with low spatial connectivity exacerbated urban flooding. The lack of connectivity and abundance of dead-end streets posed challenges for flood evacuation during emergencies. Irregular block arrangements disrupted the natural drainage system, aggravating the potential for urban flooding. These findings have implications for other flood-prone areas of neighbourhoods in Adama City and similar urban areas in the global south on how human settlements are arranged spatially to mitigate urban flood vulnerability. 

Bikila Merga Leta is an Assistant Professor in the Architecture Department at Addis Ababa Science and Technology University in Ethiopia. He is a practicing landscape architect and urban planner with a strong academic background. He completed his PhD in Urban and Regional Planning from the Technical University of Dortmund (Germany) and Addis Ababa University’s Ethiopian Institute of Architecture, Building Construction and City Development (AAU, EiABC). In addition to his doctorate, he holds an MSc in Landscape Architecture from the Free University of Brussels and Addis Ababa University, as well as a BSc in Urban Planning from Addis Ababa University. His research focuses on spatial analytics, urban development, urban flood resilience, urban sustainability, green infrastructure, and the application of GIS, remote sensing, and satellite imagery. 

Register to Attend

In her talk, Bea Taylor will take us on a journey of how she came to the Centre of Advanced Spatial Analysis, UCL. The journey begins with her PhD work, developing and applying machine learning models for understanding disease progression in dementia. Whilst here she will *attempt* to elucidate the challenges of applying AI in the healthcare domain. The next step takes us to a policy placement, where she researched regional variation in dementia diagnosis. Arriving at CASA, Bea will discuss her initial plans for her postdoctoral work, where she will be exploring how London is evolving – in particular, changing patterns of residential developments across the city using the recently launched London Planning Datahub. Be warned – in the spirit of collective intelligence she is expecting some audience participation. 

Bea Taylor recently joined CASA, UCL as a research fellow, working on the Smart Cities project as part of the new AI for Collective Intelligence hub. She has just completed a PhD at the Centre for Medical Image Computing, in the Computer Science department here at UCL. She has an MMath in Mathematics from Warwick University.

Register to Attend

A macroeconomic agent-based model is presented, featuring a community of profit-maximizing firms facing a heterogeneous labour force whose behavioural attitudes are influenced by workers' educational backgrounds. Two educational systems are analysed: a general system, which provides broad skills, and a "vocational" one, focused on specific job-related skills. Additionally, the model is extended to explore how the economy responds to technological shocks. 

Laura Mazzarino has a background in Economics and Finance and holds a PhD in Complex Systems from the University of Catania (Italy). Her research primarily focuses on the impact of economic agents' skills on outcomes in financial and labour markets. She uses network science, data-driven models, and simulated agent-based models to analyse complex market interactions at both microeconomic and macroeconomic levels.

Register to Attend

Maui County in Hawai‘i and conflict-affected regions like Sudan face significant food security challenges. This seminar highlights how Earth observation (EO), machine learning (ML), and adaptive approaches address these challenges with context-specific solutions. 

In Maui County, where 90% of food is imported, our project integrates satellite data, community-based data collection, and local knowledge to support sustainable agriculture. Using design thinking—an iterative, creative, human-centered approach to innovation—we developed tools such as the Maui Nui: Crop Monitor. This monthly newsletter provides agricultural insights, supports cultural practices, and aggregates data using traditional land divisions to amplify Indigenous knowledge and protect farmer privacy. 

In Sudan, where conflict restricts ground-truth data collection, our analysis uses Dynamic World land cover maps and uncertainty assessments to quantify changes in crop area in agricultural regions. This case study underscores the value of remote sensing in complementing ground-based survey estimates and the importance of accuracy assessments in map-based area estimates. 

These examples demonstrate how EO, ML, and tailored approaches—whether community-driven or focused on quantifying uncertainty—can strengthen agricultural resilience and inform responses to environmental and conflict-driven crises. 

Dr. Ana M. Tárano is an Assistant Research Professor in the School of Computing and Augmented Intelligence at Arizona State University. Her research focuses on developing human-centered machine learning solutions to address food security challenges. Previously, as a Senior Research Associate III at the University of Miami, she used ML to map marine habitats using Earth Observation data. She also served as Project Manager for a public-facing, real-time conversational AI installation led by interdisciplinary artist Rashaad Newsome and funded by the Stanford Institute for Human-Centered Artificial Intelligence. Dr. Tárano earned her PhD in Aeronautics and Astronautics from Stanford University, where she advanced AI methods for NASA’s Asteroid Threat Assessment Project. 

Register to Attend

Cities offer unique opportunities to address the pressing global challenges of our time. With two-thirds of the world's population expected to live in urban areas by 2050, urban design will play a pivotal role in fostering sustainable futures. This seminar will focus on the use of digital tools and analytical methods to inform sustainable design strategies that tackle these critical challenges. 

We will particularly highlight the use of mobility data to model travel flows in London, showcasing how urban design can significantly reduce carbon emissions. By leveraging mobile phone data, we have developed models that estimate travel patterns and their associated carbon impacts under various planning scenarios. 

Our findings reveal that non-commuting trips and public transport-centric policies hold significant potential for carbon reduction. High-density developments, mixed-use areas, and active travel networks are identified as crucial elements for reducing urban carbon emissions. While some of these scenarios remain theoretical, the data supports the adoption of sustainable mobility and urban design practices. 

This research underscores the importance of robust digital tools in analysing and implementing effective urban policies, ultimately paving the way for more carbon-efficient cities. 

Mateo Neira is an Associate Partner and Data Scientist at Foster and Partners. He completed his PhD at the Centre for Advanced Spatial Analysis, UCL, as part of the Alan Turing Institute doctoral program, where he focused on quantifying urban segregation and human behaviour in cities through network science and information theory. 

At Foster and Partners, Mateo leads data-driven projects that inform urban design and masterplanning. His current work revolves around complex networks, data science, and their application to the study and modelling of social and urban systems. He is particularly interested in leveraging AI to address complex urban challenges. 

Book to Attend

This event has been cancelled due to unforeseen circumstances. Thank you for your understanding.

This talk explores a non-traditional path to data leadership in healthcare. From accidental beginnings in data entry to spearheading advanced analytics in a global federated network, this journey highlights the transformative power of data across diverse healthcare settings. Discover how passion, collaboration, and a commitment to data-driven solutions can unlock valuable insights and shape the future of healthcare. 

Sarah Seager brings over 25 years of dedication to the health sector, spearheading advanced analytic strategies, reflecting a deep-seated passion for data science and its transformative potential in healthcare and the life sciences.  As a budding fine artist, her career in healthcare analytics began purely by chance and evolved from data entry into centralised government analytics within the Department of Health, public health intelligence in the NHS, Head Statistician for the General Medical Council, and then further into data management and strategy across various roles; advancing into life sciences consulting and the use of federated networks to leverage global health data. 

Owning land opens access to powerful wealth creation and accumulation tools. Long valued in England for political clout and the preservation of enormous wealth disparity, land ownership today continues to perpetuate extreme inequalities. Susannah’s current research engages with histories of land ownership in England and potentials for upending its current inequitable distribution. It does so through historic, contemporary, and speculative narratives of how land has or may change hands in response to disruptive events. The talk will give a brief overview of several previous projects before focussing specifically on who owns land in London (and why it is so hard to find that out). 

Susannah Cramer-Greenbaum joined CASA, UCL in April 2024 on a mobility fellowship from the Swiss National Science Foundation. She recently worked on the Open City Project, a collaboration between four universities which explored how London accommodates new forms of urban life. She completed her PhD at ETH Zurich in architecture and urban development. In her previous life she worked as an architect in the US on large scale urban redevelopments, regional healthcare systems, and university libraries. 

Book to Attend

Professor Mahesh Kumar Jat will present his latest work using remote-sensing data and cutting-edge machine-learning to extract land use and settlement typologies in Indian cities to support scenario development for urban planning. 

Professor Dr. Mahesh Kumar Jat is a distinguished academic and researcher in the field of Geospatial Information Systems (GIS), remote sensing, image processing, and land use/land cover change modeling, with a focus on their implications for water resources and environmental management. He currently serves as a Professor in the Department of Civil Engineering at the Malaviya National Institute of Technology Jaipur (India). Professor Jat contributes to national programs as a board member and regularly reviews for various funding agencies and academic journals. He also serves as an Editor and Editorial Board Member for over four ISI-indexed journals. His dedication extends to capacity building, where he has played a key role in enhancing the application of geospatial technologies for diverse stakeholders. 

Register to Attend

This seminar examines why U.S. firms shifted toward intangible assets since the 1970s. Guido Pialli will explain how stricter employment protection laws prompted companies to invest in R&D and innovation while reducing physical capital investments. Since intangible assets are harder to protect, retaining key talent, especially inventors, became critical. Using state-level legal changes as a natural experiment, he will show how these shifts increased knowledge intensity and polarized employment toward top inventors. 

Guido Pialli is a Postdoctoral Research Fellow at UCL, based at CASA. He obtained a PhD in Economics from the University of Torino and Maastricht University in December 2022. He is an applied economist, and his research broadly falls under the area of the economics of knowledge and innovation. He also works on topics related to labour, economics of culture and applied economic history. Guido is currently working on: i) the drivers and direction of firm and regional innovation; ii) the drivers and consequences of income inequality; iii) the role of intangible capital in the knowledge economy. 

Register to Attend

Can crises caused by government negligence trigger a rally-round-the-flag effect? On the 04 August 2020, a massive explosion occurred in Beirut, Lebanon, resulting in hundreds of deaths, thousands of injuries, and extensive damage to infrastructure and buildings. The explosion was caused by the detonation of 2,750 tons of ammonium nitrate that had been stored unsafely in a port warehouse for several years. In this study, we investigate whether exposure to the blast caused an increase in institutional trust at both regional and neighbourhood levels.

We analyse both pooled cross-sectional regional data and original geo-localised micro-data from surveys conducted before and after the blast using a difference-in-differences approach. In our spatially disaggregated model, we exploit the fact that that the blast caused far more damage in Beirut’s eastern neighbourhoods than in the city’s western half. This enables us to disentangle distinct channels of influence. We can determine whether actual exposure to the bomb, as expressed by increased building damage and bodily injuries, has any detectable influence on trust in government. Satellite imagery is employed as a robustness check to assess the extent of building damage, serving as an additional measure to validate reported physical exposure.

Findings indicate that individuals exposed to the blast exhibit were, on average, 9-25 percentage more trustworthy of political institutions, when compared to residents residing in unaffected regions. Among residents who experienced a higher degree of physical impact from the blast, trust in political institutions is comparatively lower, albeit still positive when compared to individuals outside Beirut. 

Elisabetta Pietrostefani is a geographic data scientist interested in applying data and technologies to improve the understanding of urban development and inequalities. She has analysed the effects of density policies and urban development and assessed inequalities in contexts of mass displacement. She holds a PhD in Planning Policy and Urban Economics from the London School of Economics and is currently Lecturer in Geographic Data Science at the University of Liverpool. 

Register to Attend

The Elizabeth Line is the most significant addition to London’s transport network in a generation. Running for more than 100km through central London the Elizabeth Line connects Shenfield and Abbey Wood in the East to Reading and Heathrow in the West. But how was the line developed from an idea into an operational railway? This seminar will explore the way the line was modelled and developed during an era of changing travel behaviour and highlight the emerging socio-economic outcomes. The accuracy of demand forecasting will be examined considering post-opening monitoring and what has been learnt from demand forecasting, business case development and user response monitoring will be explored. 

Hannah Donovan works in TfL’s Public Transport Planning Team as a Rail Modeller. She is experienced in developing business cases, forecasting and benefits realisation, most recently for the Elizabeth Line. Hannah has a background in highway modelling both strategic and microsimulation and is currently pursuing a PhD with The Bartlett’s Centre for Advanced Spatial Analysis, UCL, developing an agent-based model examining street users’ behaviours and interactions with e-scooters.  

David Warner works in TfL’s Public Transport Planning Team and has over 20 years’ experience of working within the UK Transport Industry. He is an experienced passenger demand and revenue forecaster with detailed knowledge of economic and financial appraisal and rail strategy development. He was responsible for developing the business cases and demand projections for the development of the London Overground and now undertakes similar tasks for the Elizabeth line, also covering the first-year impacts monitoring. 

Register to Attend

The NYPD used facial recognition technology in 22,000 cases between 2016 and 2019 – half of which were in 2019. Despite mounting evidence that facial recognition technology violates human rights, in most cases, we do not know where, when or why. When Amnesty International and others filed a Freedom of Information Law request, the NYPD refused to provide information. To help bring about a ban on police use of facial recognition technology, in mid-2021 Amnesty International launched an ambitious effort, Decode Surveillance NYC. The effort mobilised thousands of digital volunteers to find and categorise CCTV cameras throughout the city. We then worked with data scientists, geographers, and 3D modellers to analyse the crowdsourced data. Our analysis of the Decode Surveillance NYC data reveals an expansive, invasive and discriminatory surveillance machine at the core of the NYPD’s policing tactics. 

Milena Marin is the Deputy Director for Digital Investigations, Amnesty International. Milena leads the Evidence Lab, Amnesty International’s digital investigations team, which works to advance digital methods in human rights research. With a career spanning fifteen years at the intersection of technology, data and social advocacy, Milena has worked to defend human rights, increase public sector transparency, fight corruption and promote open data. Her work includes leading Amnesty Decoders, a pioneering initiative that uses data science, crowdsourcing and artificial intelligence to sift through and analyse large volumes of data for human rights research. Prior to Amnesty International, Milena led the School of Data, a data literacy programme where she trained and mentored activists and journalists to use data for social impact. She has also worked with Transparency International, enabling the organisation’s global network to use technology in the fight against corruption.

Register to Attend

Mobility datasets have opened up new analytical avenues for the study of urban inequalities which have been traditionally studied from a place-based perspective. Individual-based approaches allow for the investigation of intersectional inequalities and a better understanding of how population subgroups are affected by the uneven distribution of urban resources and infrastructure. The talk will present ongoing research on gender inequalities in Brazil, exploring Origin-Destination Survey data for two large cities: Curitiba and São Paulo. 

Joana Barros is currently a Lecturer in Urban Geography at CASA. Her research focuses on the study of patterns and dynamics of socio-economic inequalities in urban space, contributing to the understanding of phenomena such as urban segregation, housing/informal settlements, deprivation, mobility, and transport. She has expertise in cities of the Global South (Latin America in particular) and a strong interest in comparative studies, including those across the Global North and South. 

Register to Attend