GIS Colloquium – Talks (Winter Term 2016/2017)
Dr. Jochen Wendel
Mon, Oct 24, 2016, 2.15 pm, INF 348, Lecture Hall (Room 015)
Research on smart city approaches is of high complex multidisciplinary nature and involves multiple stakeholders. Data used in city wide research projects usually differs by domain and by means of data gathering that requires intensive data integration efforts. A major obstacle in this process is the current proprietary nature of major commercial technologies. The lack of interoperability makes it difficult to share and further process research results. This presentation focuses on the development of an open data infrastructure that allows data storage, data exchange, data analysis, as well as data visualization across projects and research domains. The application of the proposed infrastructure will be discussed in regards to different smart city approaches. Examples will be demonstrated for three different geographic regions that includes the development of a socio-demographic data mining approach for energy usage for the city of Karlsruhe, an analysis of electric car charging behavior using a Neural Network classification model for the island of La Reunion, and the usage of open-data in smart city approaches demonstrated on New York City.
Dr. Michael Bauder
Mon, Oct 31, 2016, 2.15 pm, INF 348, Lecture Hall (Room 015)
The data shared on online photo platforms like Flickr and Tumblr are an example for so-called Ambient Geospatial Information (AGI). Opposing Volunteered Geographic Information (VGI), AGI data is not made and distributed with the purpose of an usage beyond the personal level (i.e. it is “ambient” of other uses), and the geographic information - namely the location - contained in the AGI is merely a byproduct. Under the umbrella of Big Data analytics AGI are often seen with the mindset of solutionism - believing that these data sets offer new insights, characterized by objectivity, truth, and accuracy (boyd/Crawford, 2013) - and “proto-naïve neo-positivism” (Boeckler, 2014), which is based on the belief that in the age of Big Data there is no need for a hypothesis anymore and no need for causal correlations. Both can be seen as the nightmare of (not only) any critical social scientist, fearing a new age of apophenia and so many AGI based studies show significant flaws. But AGI still hold great potentials, if we are aware of their origin and characteristics, as AGI link two types of spaces, what we might call a material and a virtual space. The material space is the space that the person actually encounters and interacts with. The virtual space is the space represented on the platforms later. I argue that these are not the same spaces, as the process of making photos, as well as the process of uploading the photos, is a process of selection following distinct rules. So, online platforms provide a source of Geospatial Information of a space which can be considered as a deviant reproduction of the material space originally encountered by the photographer. However, this dialectic of space provides manifold research opportunities. Based on the field of tourism research, I demonstrate the opportunities discussed in current research projects with a special focus on a GPS tracking/AGI combination connecting measured stopping times of tourists with the amount of photos taken.
Dr. Franz-Benjamin Mocnik
Mon, Nov 7, 2016, 2.15 pm, INF 348, Lecture Hall (Room 015)
Spatial information inherits properties, in particular structural properties, from space. Most prominent Tobler's law claims that “everything is related to everything else, but near things are more related than distant things”. After a short discussion of such structures of spatial information, I will contrast the structure of the map and text media in order to understand their differing affordances of representing spatial information. Maps are, for example, good in representing spatial relations, while texts and other media is known to be far better in telling stories, which is a major reason behind the embedding of multimedia and text elements in multimedia maps. Based on this comparison of the map and text media, a new paradigm of map use is proposed to improve the ability to tell stories happening in space and time by using “conventional” maps (and not multimedia maps).
Dr. Jiaoyan Chen
Mon, Nov 14, 2016, 2.15 pm, INF 348, Lecture Hall (Room 015)
Predictive analytics like time-series forecasting plays an important role in big spatial data analysis. This report first presents a case study of big spatial data analysis about China smog (big air pollution event), which adopts multiple physical sensor data streams and social media (e.g., Weibo) data streams. Different data mining techniques (e.g., feature extraction, ensemble learning, artificial neural network, text analysis) are adopted and evaluated for the prediction problems in smog and smog-related health analysis. Good results can be achieved with a suitable combination of these algorithms. However, such traditional data mining still has big improvement space for spatio-temporal prediction with big data. Machine learning models may encounter the problems in solving concept shift, providing explanations, fusing big data, etc. This report also presents my study of consistent prediction method, which incorporates knowledge reasoning for the concept shift problem in supervised stream learning with the case study of air quality forecasting. In this part, my personal ideas of why we need knowledge graph (i.e., linked data) besides raw data records and how integrating learning and reasoning benefits predictive analytics of spatial data will be briefly presented.
Prof. Dr. Sven Lautenbach
Mon, Dec 12, 2016, 2.15 pm, INF 348, Lecture Hall (Room 015)
Land use decisions have wanted and unwanted consequences with respect to the goods and services produced by land systems. Identifying the trade-offs and synergies that come along with land use decisions is therefore important for informing policy makers. The aim of the talk is to provide an overview about methods to assess the different environmental goods and services produced by land systems and to quantify trade-offs. It will touch upon a series of case studies from regional to global scale that use spatial analysis, remote sensing, simulation models, multi-objective optimization and machine-learning approaches to quantify trade-offs in different land systems.
Spectroscopy and hyperspectral imagery for applications in agriculture – from in-situ measurements to spatial assessment
Dr. Thomas Jarmer
Mon, Dec 19, 2016, 2.15 pm, INF 348, Lecture Hall (Room 015)
Remote sensing is of essential importance in environmental monitoring and assessment. In this context, reflectance spectrometry is accepted as a cost-efficient and fast screening tool to assess soil- and vegetation parameters. However, even with reflectance spectroscopy a spatial assessment of surface properties exclusively by terrestrial investigation is not feasible for larger regions or areas with limited accessibility. Airborne- or spaceborne multi- and hyperspectral remote sensing systems provide objective and repetitive data which allow broad spatial coverage and hence, are especially useful for environmental monitoring. But without ground truthing these data only facilitate qualitative statements. Combining quantitative field measurements with remote sensing imagery offers the real potential of remote sensing: the quantitative assessment of parameters on regional scale. In this talk, the potential of lab- and field spectroscopy as well as remote sensing imagery will be presented by different examples for the quantitative assessment of surface properties. Empirical-statistical models to predict soil and crop properties from lab- and field spectra will be introduced and adopted to image data to assess the spatial variability. The derived information will be included in approaches for soil regionalization and yield estimation. Finally, the talk will be closed by an outlook for potential usage and extension of the presented approaches and future research questions.
Dr. Wei Huang
Mon, Jan 9, 2017, 2.15 pm, INF 348, Lecture Hall (Room 015)
The way people live in cities forms human activity patterns, which affects how urban systems work and plays a key role in a variety of urban applications, such as urban planning, emergency response, transportation planning and epidemic prevention. Therefore, it is essential to understand human activity patterns, where precise prediction of human movements and mechanistic modelling of human activity patterns are the two keys. Most of existing work on prediction of human movements cannot deal with activity changes, leading to a negative impact on the predictive accuracy. Furthermore, the majority of current work on modelling human activity patterns are mainly researched from spatiotemporal perspectives, but the motivation behind is usually being neglected, which is crucial to understanding activity changes. In this presentation, I first introduce two of my PhD work: 1) a novel human mobility predictive model with consideration of human activity changes; and 2) a new approach of modeling human activity patterns, which enables to uncover human activity patterns not only from spatiotemporal dimensions but also considering the motivation behind. Then some ideas I am looking to further explore are proposed and discussed.
Dr. Yingwei Yan
Mon, Jan 16, 2017, 2.15 pm, INF 348, Lecture Hall (Room 015)
Our project provides volunteer-based pest management solutions that enable farmers to collectively share information regarding pest conditions in agricultural fields. It gathers scattered pieces of pest observation information from individual farmers, uses sense making algorithms to leverage on the information, and produces forecasts of pest infestations which will be available to the community. As a result, the solutions will have the potential to improve agricultural productivity, and by extension, global food security. The talk will first give an overview of the project, and subsequently focus on two aspects: (1) quality assurance of the data provided by farmers; (2) sense making of the data provided by farmers.
Prof. Shunichi Koshimura
Mon, Jan 30, 2017, 2.15 pm, INF 348, Lecture Hall (Room 015)
In the aftermath of catastrophic natural disasters, such as earthquakes and tsunamis, our society has experienced significant difficulties in assessing disaster impact in the limited amount of time. In recent years, the quality of satellite sensors and access to and use of satellite imagery and services has greatly improved. More and more space agencies have embraced data-sharing policies that facilitate access to archived and up-to-date imagery. Tremendous progress has been achieved through the continuous development of powerful algorithms and software packages to manage and process geospatial data and to disseminate imagery and geospatial datasets in near-real time via geo-web-services, which can be used in disaster-risk management and emergency response efforts. Satellite Earth observations now offer consistent coverage and scope to provide a synoptic overview of large areas, repeated regularly. These can be used to compare risk across different countries, day and night, in all weather conditions, and in trans-boundary areas. On the other hand, with use of modern computing power and advanced sensor networks, the great advances of real-time simulation have been achieved. The data and information derived from satellite Earth observations, integrated with in situ information and simulation modeling provides unique value and the necessary complement to socio-economic data. Emphasis also needs to be placed on ensuring space-based data and information are used in existing and planned national and local disaster risk management systems, together with other data and information sources as a way to strengthen the resilience of communities.Through the case studies of the 2011 Great East Japan earthquake and tsunami disaster., we aim to provide evidence regarding how Earth observations, in combination with local, in situ data and information sources, can support the decision-making process before, during and after a disaster strikes. We also provide evidence regarding how such space-based applications integrated with real-time simulation can contribute to the aims of the post-2015 DRR framework, which has a strong focus on disaster-risk reduction and on avoiding the generation of new risks.