GIS Colloquium – Talks (Summer Term 2019)
Dr. Jochen Albrecht
Mon, April 29, 2.15 pm (Venue: INF 348, Room 015)
The common assumption is that cities are the main producers of global greenhouse gas emissions (GHGs). Multiple publications by a research group that I am a member of have shown that cities, even when very generously defined as urbanized areas with a significant extent, produce only 33% to 40% of all GHG emissions. This begs the question where and by whom is the bulk of GHGs produced. It turns out that it is the suburbs that place the biggest burden on our planet's sustainability and that cities are both on a per capita basis as well as in absolute terms rather efficient. Using the spatially disaggregated EDGAR dataset of the Joint Research Centre (Ispra, Italy), I present the results of a spatial analysis of some 20 different emission producers for a range of emission types, all aggregated to CO2-equivalents. The regional scale analysis shows that while the overall role of suburban rings is uncontentious, the actual culprits vary widely from region to region.
Mon, May 6, 2.15 pm (Venue: INF 348, Room 015)
Airborne laser scanning (ALS) has been used for the derivation of forest structure descriptors and forest inventory (FI) parameters on an operational level for more than two decades. Mobilization costs of ALS missions, however, are high and ALS acquisitions typically are performed at intervals of several years only. With the advent of laser scanning systems that are operated from the ground (terrestrial laser scanning, TLS) or mounted on unmanned aerial vehicles (ULS), we have promising means at hand for the update of existing FIs on local scales at regular and shorter time intervals. Apart from lower mobilization costs, these latter systems provide a much higher level of detail than could be achieved through traditional ALS missions. This high level of detail allows for the retrieval of FI-parameters as, e.g., the diameter at breast height or stem volumes directly from the point cloud and, therefore, have the potential to partly substitute traditional in-situ FI-measurements. Compared to in-situ measurements, ULS is able to cover larger plots and transects, respectively. Thus, the question is how this new type of reference data can be linked to traditional forest structure metrics derived from ALS data. The comparability of the retrieved structure metrics, however, is a crucial aspect if data from different laser scanning platforms should be combined across scales. In the presentation, I will discuss the possibilities for the derivation of forest structure metrics from laser scanning data and, in particular, present the opportunities of today’s TLS and ULS systems for forestry applications. Furthermore, I will shed light on the differences between forest structure metrics derived from ULS and ALS and show first results from a comparison of the two systems.
Dr. Massimiliano Pittore
Mon, May 20, 2.15 pm (Venue: INF 348, Room 015)
The assessment of risk arising from the increasingly complex and vulnerable urban areas being exposed to natural hazards is a matter of cross-disciplinarity, patience and creativity, and is more and more a matter of geo-information. In order to match the requirements of practitioners and end-users, risk estimates have to precisely catch the relevant features of the built environment and its complex and vulnerable infrastructure. This requires the harmonisation of heterogeneous data sources, from remote sensing to authoritative and volunteered geoinformation, and calls upon new methodologies to collect, process and aggregate each “atomic” information into dynamic, multi-scale and uncertainty-aware models. The increasing availability of large scale loosely structured geoinformation paves the way to the application of (geo)statistical learning (artificial intelligence) techniques, but also advocates for a stronger cooperation among professionals with very different background. In this talk I will review some of our recent research activities where seismology, engineering and different flavours of GIScience are playing together to address trending issues in exposure and vulnerability modelling and in post-earthquake damage mapping.
Mon, July 1, 2.15 pm (Venue: INF 348, Room 015)
The potential of images to answer research questions in geosciences is vast. Earth observation with image sequences allows for qualitative and quantitative assessment of earth surface processes and their changes. The topography and bathymetry can be reconstructed with very high resolution using image processing approaches. Furthermore, hydrological and geomorphological features can be detected and traced to improve environmental monitoring. Different sensors (e.g. RGB and thermal camera) and platforms (e.g. UAVs and terrestrial systems) can be used and due to their flexibility are especially suitable for fragile landscapes. This contribution introduces some basic principles regarding image processing and discusses potentials focusing on two different disciplines, i.e. geomorphology and hydrology. The first application introduces the potential of multi-temporal UAV data to reconstruct soil surface models to identify erosion forms and their changes over time to better understand the process of soil erosion. The second application in hydrology introduces the processing and analysis of image sequences from various sources to measure water levels and flow velocities. These case studies are chosen to display the variety of applications of image processing in geosciences and to support evaluation of further potentials.