SITCOMImplementierung eines Online-Lernspiels ("Reflectories")

Ziel dieser Arbeit ist die technische Umsetzung eines Lernspiels für den Geographieunterricht, das am Institut für Didaktik der Geographie (IfDG) entwickelt wurde. In diesen sogenannten „Reflectories“ werden User in kurzen Audio-Beiträgen vor komplexe geographische Entscheidungen gestellt, die mithilfe von unterschiedlichen Zusatzmaterialien getroffen werden müssen. Je nach gewählter Entscheidungsoption nimmt die Handlung einen anderen Verlauf, sodass sich die User mit den entsprechenden Konsequenzen ihrer Entscheidungen konfrontiert sehen und aufgefordert sind, über ihr Handeln zu reflektieren.

Aufgabe der Arbeit ist die Umsetzung der erarbeiteten Inhalte (hauptsächlich Audio-Dateien, einige PDF und Bilddateien) in eine systemunabhängige (Web-)App. Dazu gehört u. a. die Erstellung und Evaluation einer für die Zielgruppe geeignete Benutzeroberfläche sowie eine robuste Implementation des Backends.

Die Arbeit ist in Zusammenarbeit mit dem IfDG (Gabriele Schrüfer, Nina Brendel) geplant.

Contact: Christian Kray

Author: Niklas Trzaska

Supervisor: Christian Kray

SILMechanismen zur Erstellung von Anfahrtsskizzen

Anfahrtskizzen sind Karten, die optimiert sind, um die Route aus verschiedenen Richtungen zu einem bestimmten Zielort zu zeigen. Zumeist sind sie manuell erstellt, wobei

  • Informationen selektiv ausgewählt wurden
  • schematisiert dargestellt wurden

Die Arbeit soll Anfahrtsskizzen analysieren und die dahinterliegenden Mechanismen extrahieren, wie Anfahrtsskizzen erfolgreich visualisiert werden können. Die Ergebnisse werden systematisch in einer Nutzerstudie getestet.

Beispiele für Anfahrtsskizzen:

Contact: Angela Schwering

Author: Jens Golze

Supervisor: Angela Schwering

Co-supervisor: Vanessa Joy Anacta

STMLopenEO Hub

openEO develops an open API to connect R, Python and Javascript clients to big earth observation cloud back-ends in a simple and unified way. Back-ends process user-defined algorithms on remote sensing data sets within their cloud infrastructure. Although the communication between clients and back-ends is standardized by the openEO API, each back-end will implement the API to a different extent and will differ with regards to available processes and data sets. Therefore users should be able to search on a central platform for back-ends that fully support the users requirements. This includes the ability to search for back-ends by

  • data sets, e.g. temporal extent, spatial extent, platform, sensor, bands or name,
  • processes, e.g. by a process graph provided by the user,
  • other back-end related metadata, e.g. API version, capabilities or costs.

Additionally, it could be useful for users to publish and share their algorithms as process graphs or user-defined functions (UDFs) on this central platform.

This thesis should explore, implement and evaluate one or multiple of these aspects. The scope of the thesis is designed to fit the requirements of a bachelor thesis. More information can be found in the openEO Hub GitHub repository.


Contact: Matthias Mohr

Author: Christoph Friedrich

Supervisor: Edzer Pebesma

Co-supervisor: Matthias Mohr


SITCOMFacilitating Indoor Navigation via virtual situated displays

Dynamic, personalised signage - public displays that show navigational instructions and wayfinding information - can be very useful to help people to successfully reach their destination in complex indoor environments. However, outfitting a large building with a dense infrastructure of public displays would be costly and resource-intensive. One way to overcome is the idea to use virtual displays that can only be seen using a person's mobile device. This thesis will investigate this idea, and test it using a prototypical implementation.

Contact: C. Kray

Author: Jakob Altenstein

Supervisor: Chris Kray

External“Feature Info” – Visualizing GIS Background Information

External Thesis in co-operation with con terra. A PDF file with all information can be downloaded here.
Usability is an important aspect of modern software quality. In recent years, the users’ expectations of user interfaces raised significantly. Within the last 2 decades, a large body of research thus focused on optimizing the usability of desktop and mobile applications, e.g., [1]. However, despite the general popularity of this topic, the usability of GIS software appears to have gained less interest—some examples are [2, 3].
As con terra is a leading provider for integrated Geo IT solutions on an international level, it seeks to optimize the quality of its products and solutions with a high degree of usability. Usually, GIS software products are complex systems, comprising many intertwining components. Thus, optimizing the usability of such complex systems is usually only manageable by focusing on each component individually. One particular component, which is used in basically every GIS software, is the “feature info”—a component, that provides additional background information to the contents visualized on a map. Figure 1 shows an example of a typical feature info.
Though feature infos are important components of practically every GIS software, the concept has not significantly changed within the last years. Yet, the characteristics of conventional feature infos may not suit, e.g., the requirements of mobile applications, which have gained a significant importance and presence in our everyday lives nowadays. One issue, for instance, may be the very limited screen real estate available on such small devices.
Moreover, in some application scenarios, the background information shown in a feature info may be significantly more complex than in the example shown above. For example, it may be necessary to provide data with a specific temporal extent, such as climate measurements. This thesis thus seeks to find novel approaches to visualize additional background information to users of GIS software. For example, the thesis could analyze the performance of three options to position the additional information, i.e., “in place attached”, “in place detached”, or “isolated” (e.g., showing information in a separate browser window or on another device).
Dr. Morin Ostkamp 
con terra GmbH 
Martin-Luther-King-Weg 24 
48155 Münster 48149 Münster
+49 89 207 005 2200 
Prof. Dr. Christian Kray
Institute for Geoinformatics
Heisenbergstraße 2
+49 251 83 33073
[1] Jakob Nielsen. Usability Engineering. Kaufmann, 1993
[2] Clare Davies and David Medyckyj-Scott. Gis usability: recommendations based on the user’s view. International Journal of Geographical Information Science, 8(2):175–189, 1994.
[3] Max J Egenhofer and James R Richards. Exploratory access to geographic data based on the map-overlay metaphor. Journal of Visual Languages & Computing, 4(2):105–125, 1993.


Contact: Morin Ostkamp

Author: Yevgeniya Litvinova

Supervisor: Christian Kray

Co-supervisor: Morin Ostkamp

SILCognitively plausible model of place

This idea behind this comes from Luescher and Weibel (2010) - "we might for example present a different answer to the question ‘shopping opportunities in the city centre?’ to elderly people than to young people (the former avoiding the night club district because they might perceive it unsafe)."

Despite the notion of place being widespread in natural language, it is still difficult to model it in a GIS (which is more concerned with location). The student can identify a suitable site, and attempt to study how the notion of place differs among participants of different age groups (young vs. old), residence (e.g living in the city vs. suburbs) etc. The analysis could offer insights into whether there are differences in the way a place is perceived based on these factors and has potential implications for contextualized location based services.

Contact: Imad Humayun

Author: Tobias Tresselt

Supervisor: Angela Schwering

Co-supervisor: Imad Humayun

SILCreating schematic maps in mobile devices

Schematic maps are maps that intentionally distort and misrepresent geometries in order to simplify maps. Oftentimes, schematic maps give a better overview of the environment, because they focus only on important aspects. The displays of mobile devices offer only limited space. This thesis topic aims at the development of an algorithm to schematize maps automatically for mobile devices.

Schematization refers to misrepresentations of shape and size of spatial objects and distortions of distance and direction relations among them. The research starts from findings by Latecki and Lakämper, who developed the discrete curve evolution, an algorithm to schematize topographic maps. Barkowski et al. applied this algorithm to map schematization.


Latecki, L.J. and R. Lakämper (2000). Shape similarity measure based on correspondence of visual parts. IEEE Trans. Pattern Analysis and Machine Intelligence (PAMI) 22(10): 1185-1190.

Barkowsky, T., L.J. Latecki, and K.-F. Richter (2000). Schematizing maps: Simplification of geographic shape by discrete curve evolution. Spatial cognition II: Integrating abstract theories, empirical studies, formal methods and practical applications. C. Freksa, et al., Springer: 41–53.

Contact: Angela Schwering

Author: Andreas Ohrem

Supervisor: Angela Schwering

SILVisualization of global landmarks on mobile devices.

In the WayTo research project, we developed a prototype to visualize global landmarks which are outside the map section displayed on a mobile device by visualizing them at the edge of screen to indicate directions.

An amendment to this design is to incorporate distance into the design of icons. The thesis could explore different techniques (e.g. halo (circles at the edge), wedge (triangle at the edge), or distance-encoding arrows). These methods distinguish only two or three levels (or four maximum) of icons size (that means we only have 4 different sizes of each global landmarks to indicate very close, close, far, and very far) on screen. Although there is hardly any empirical data to support that, drawing from cartographic theories regarding visual variables in animated map making, humans are not able to distinguish changes on map representation at fine levels.

Possible bachelor and master thesis could address:

1.    Implementing distance and direction indicating landmarks (DDL) display on mobile phones

2.    Investigating the effects of DDL on mobile screens on spatial orientation and spatial knowledge

3.    Comparing the variations of DDL design (gradient to true distance vs. categorical distance)

4.    Comparing the effects of DDL with Halo and Wedge design

Reference: Li, R.; Korda, A.; Radtke, M.; Schwering, A. (2014): Visualising distant off-screen landmarks on mobile devices to support spatial orientation.

Contact: Angela Schwering

Author: x

Supervisor: Angela Schwering

SITCOMPost-record Georeferencing of Videos

Videos are inherently spatial. However during recording most often the spatial component is not captured as metadata. Work on spatial videos has underpinned the potential and benefits of locating video scenes. 
This thesis will develop an approach to add the spatial component to the video stream in a post production step allowing for efficient storing and query the video spatially. The student will develop a concept to provide capturing locations, heading and further information as metadata with the video stream. Therefore a framework has to be developed to provide a middleware for different input and interaction modalities with the video data using map-based interfaces.
The framework will exemplify the approach for a disaster scenario like the recent heavy rain event in Münster using a number of available youtube videos. Additionally a series of user studies will be performed as evaluation of the interaction modalities.

Contact: Holger Fritze

Author: Samuel Navas Medrano

Supervisor: Prof. Chris Kray

Co-supervisor: Holger fritze

STMLArray databases for in situ sensor services

Most environmental in situ monitoring stations record observations at a fixed number of stations with constant frequency, for instance every minute or every hour. Arrays are a natural way to represent these, with space (station ID) and time as two of the dimensions. Current implementations of sensor observation services (SOS, [1]) however typically use a normalized relational database as the data backend. If the data size grows to billions of records, querying and updating the database becomes relatively slow, with indexes taking up a lot of resources. This thesis will evaluate the use of an array data base (SciDB) as a back-end for fixed sensor SOS for massive data sets, and compare it with relational databases. It will also compare the two approaches for complex aggregations, such as computing for each station the mean pollutant concentration profile by hour of the week averaged over several years, and queries that involve more than one parameter such as the number of days per year where more than one parameter exceeds a threshold. Following the air quality directives [1] the EEA currently collects all air quality observations reported by member states. This involves hourly observations on a few dozen parameters for thousands of stations.  This dataset is open [3], and will be the main use case to compare the two approaches.

Contact: Edzer Pebesma

Author: Jigeeshu Joshi

Supervisor: Edzer Pebesma

SILEnhancing Dynamic Geometry with Artificial Intelligence: Applications in Medicine and GIS

Powerful methods in an area called "dynamic geometry" provide a rapid way of solving a wide range of spatial reasoning problems. For example, as a user manipulates some shapes (circles, lines, points) in an "intelligent" sketch, the sketch automatically updates itself to maintain certain spatial constraints (e.g. lines being tangent to circles, lines being parallel to each other, and so on).

In medicine these shapes could represent different types of cells that have been automatically recognised from images of a tissue section, and the task could be to determine whether the cells are cancerous (histopathology). In geographic information systems these shapes could represent streets, buildings, and landmarks recognised from satellite images or directly from a hand-drawn sketch.

In this thesis project you will be:

  • extending the spatial language of dynamic geometry to work with common-sense qualitative spatial relations (near, left of, inside, etc.)
  • integrating dynamic geometry into a knowledge representation (artificial intelligence) framework

You will then use this enhanced technology for a range of exciting tasks, such as:

  • improving computer-based image recognition by automatically correcting errors based on knowledge about objects in the domain
  • providing new ways of interacting with complex data using "intelligent" sketches

Through this research project you will develop skills and experience in the application of methods in artificial intelligence (AI). You will be introduced to the necessary tools and existing projects to build on. No prior experience with methods in AI is necessary (you will be given considerable support in this area).

Contact: Carl Schultz

Author: Nikolai Gorte

Supervisor: Angela Schwering

Co-supervisor: Carl Schultz

SILEmbodied 3d Isovist as a Predictor of Spatial Experience

Isovist analysis is used to predict how people behave in and 'feel about' the geometry of space. One of the best-documented examples of such relation is provided by Wiener et al. (2007). In this work, the authors controlled the shape of a Virtual Reality environment and measured how people experience individual spaces, depending on its underlying isovist properties.

Recently, 3d isovist analysis became increasingly popular in contexts where its traditional 2d counterpart suffers major limitations. However, it is not clear if the influence of isovists on human experience can be directly extrapolated from the 2d to the 3d analysis. In this thesis, the student will replicate the VR experiment of Wiener and colleagues, taking into account the vertical component of human experience and using a 3-dimensional isovist in the analysis of final results.


Wiener, J. M., Franz, G., Rossmanith, N., Reichelt, A., Mallot, H. A., & Bülthoff, H. H. (2007). Isovist analysis captures properties of space relevant for locomotion and experience. Perception, 36(7), 1066 – 1083.


Krukar, J., Schultz, C., Bhatt, M., (forthcoming). Towards Embodied 3d Isovists: Incorporating cognitively-motivated semantics of `space’ and the architectural environment in 3D visibility analysis

Contact: Jakub Krukar

Author: Charu Manivannan

Supervisor: Jakub Krukar

SILAn Artificial Intelligence System that Learns by Spatial Induction

"Learning by induction" is the ability to take a number of observations or examples and discover rules that account for the observations - it's the ability to generalise from examples.

Being able to generalise spatial patterns from observations is essential for artificial intelligence systems to perform a variety of tasks in a flexible and robust way. We don't want to tell the computer system exactly how to solve every specific problem it will encounter. Instead, we want to give it some examples and have the system use common sense and background knowledge to figure out ways of solving new problems that have a similar structure.

In this thesis project you will be:

(a) integrating a new powerful "common sense" spatial reasoning method called Enhanced Geometric Constraint Solving within a more general artificial intelligence framework for inductive spatial learning;

(b) evaluating the system in a range of exciting applications in geographic information science, architectural design, cognitive psychology (analogical reasoning), and medicine (histopathology).

Through this project you will develop skills and experience in the application of methods in artificial intelligence (AI). You will be introduced to the necessary tools and existing projects to build on. No prior experience with methods in AI is necessary (you will be given considerable support in this area).

Contact: Carl Peter Leslie Schultz

Author: Lars Syfuß

Supervisor: Edzer Pebesma

Co-supervisor: Carl Schultz

STMLDas könnte Sie auch interessieren

Im Arbeitsfeld der reproduzierbaren Forschung werden wissenschaftliche Artikel gemeinsam mit Daten und Programmcode in Form von Kompendien organisiert. Ziel solcher Kompendien ist es, die Daten oder die Analysen austauschbar zu gestalten, sowie den Zugang zu Daten und Software langfristig zu sicherzustellen. Für ein gutes Benutzererlebnis sollte der Austausch von Daten und Analysen zwischen Kompendien einfach und stabil sein, die einen Daten also kompatibel mit dem anderen Code.

Suchdatenbanken, wie zum Beispiel Elasticsearch, spielen beim Auffinden von Dokumenten im Web eine zentrale Rolle. Eine typische Funktion einer Suche ist das Vorschlagen ähnlicher Dokumente auf Basis hochperformanter invertierter Indizes.

Einen ersten Schritt hin zur Kompabilitätanalyse stellen direkte und mittelbare Metadaten dar die in Suchdatenbanken gesammelt werden. Diese Metadaten werden heute meist vom Autor erstellt (abstract, keywords) und nicht umfassend. Auf der Basis von Kompendien können diese und weitere Informationen aus den Sekundärdateien (Daten, Quellcode) abgeleitet werden. Zum Beispiel könnten ähnliche genutzte Softwarekomponenten oder Datenausschnitte einen Hinweis darauf geben, dass zwei gegebene Kompendien so weit kompatibel sind, dass die Daten des einen mit der Analyse des anderen kombiniert werden können.

Ziel dieser Arbeit ist es, die möglichen Quellen von Metadaten wissenschaftlicher Publikationen zu sichten und mit den Anforderungen des Anwendungsfalls zusammen zu führen. Es sollen neue Wege zur Erweiterung, Integration und Vergleich der Metadatensätze entworfen und mittels einer prototypischen Implementierung evaluiert werden.

Die Arbeit kann auf Deutsch oder Englisch verfasst werden.


Contact: Daniel Nüst

Author: Lukas Lohoff

Supervisor: Edzer Pebesma

Co-supervisor: Daniel Nüst

SITCOMLocation-Based Notifications to Encourage VGI Participation and Contribution


Contact: Jakub Krukar

Author: Joanna Kwong

Supervisor: Chris Kray

Co-supervisor: Jakub Krukar

SILValidity of sketch maps under varied tasks

Sketch maps are drawings which represent human spatial memory of an area of interest. Sketch maps, however, are commonly distorted - even when the knowledge of a certain area is well established. There are two contrary models that describe the relation between human spatial memory and the drawn sketch map. Model (1) assumes that the relation between human memory and the map drawn is relatively stable - i.e., that the quality of the sketch map is directly linked to the quality of spatial knowledge. Model (2) assumes that the quality of a drawn map can differ depending on the task at hand. This would imply that the number and type of errors is different on two maps drawn subsequently if the task changes (even if the map represents a well-known area).
This thesis will involve designing an experiment in which participant draw two maps of the same area for two different tasks / reasons / motivations. You will then be required to analyse the type of errors (and the type of correct drawn information) on these sketch maps in order to verify whether model (1) or (2) better explains your data.

Blades, M., 1990. The reliability of data collected from sketch maps. Journal of Environmental Psychology, 10(4), pp.327–339.

Tversky, B. (1992). Distortions in cognitive maps. Geoforum, 23(2), 131–138.
Wang, J., & Schwering, A. (2015). Invariant spatial information in sketch maps — a study of survey sketch maps of urban areas. Journal of Spatial Information Science, 11(11), 31–52.
Tversky, B. (2009). Spatial cognition: Embodied and situated. In M. Aydede & P. Robbins (Eds.), The Cambridge Handbook of Situated Cognition (pp. 201–216). New York: Cambridge University Press.      (section “Space of Navigation”)

Contact: Jakub Krukar

Author: Antonia van Eek

Supervisor: Angela Schwering

Co-supervisor: Jakub Krukar

SITCOMSpatial Query Interface for an Open City API

Various open data portals are currently emerging as catalogs for data being made open by public institutions. Enabling an efficient access to, and searching of these open datasets is still not fully understood. [1] proposed semantic APIs as a way of improving access to open data. The purpose of this thesis is to design and implement functionalities to enable spatial search through this API.

The thesis will involve two basic tasks:
  - addition of curated spatial datasets to the Open City Toolkit
  - development of an intuitive interface for the spatial query of these spatial datasets
The intuitiveness of the Spatial Query Interface could be tested, for example, through (sample) usability tests.

[1] Degbelo, A., Trilles, S., Kray, C., Bhattacharya, D., Schiestel, N., Wissing, J. and Granell, C. (2016) ‘Designing semantic APIs for open government data’, JeDEM - eJournal of eDemocracy and Open Government, 8(2), pp. 21–58.


Contact: Auriol Degbelo

Author: Ang Sherpa

Supervisor: Auriol Degbelo

SITCOMOpen City Toolkit Recommender System

The Open City Toolkit has been suggested in [1,2] as a way of improving data re-use in the city context. This thesis will develop and test a module which suggests
relevant datasets for a new Open City Toolkit app. The development may follow user-centered design principles.

The thesis will involve two basic tasks
 - modelling of (spatial) datasets and apps in the context of the Open City Toolkit
 - specification of an annotation strategy of both apps and datasets, based on open data formats (e.g., RDF, JSON, JSON-LD)
 - design and implementation of recommendation functionalities (i.e., which datasets could be re-used in which apps)  
The feasibility of the ideas suggested is expected to be demonstrated through a prototypical implementation.

[1] Degbelo, A., Granell, C., Trilles, S., Bhattacharya, D., Casteleyn, S. and Kray, C. (2016) ‘Opening up smart cities: citizen-centric challenges and opportunities from GIScience’, ISPRS International Journal of Geo-Information, 5(2), p. 16. doi: 10.3390/ijgi5020016.

[2] Degbelo, A., Trilles, S., Kray, C., Bhattacharya, D., Schiestel, N., Wissing, J. and Granell, C. (2016) ‘Designing semantic APIs for open government data’, JeDEM - eJournal of eDemocracy and Open Government, 8(2), pp. 21–58.


Contact: Auriol Degbelo

Author: Brhane Bahrishum

Supervisor: Auriol Degbelo

GeoSimSpatial-Temporal Analysis of Type 2 Diabetes in Westfalen-Lippe (Germany)

The main aim of this thesis is to analyze the spatial distribution of diabetes type 2 (T2DM) in Westfalen-Lippe (Germany) based on socio-geographic and economic data.

Contact: Judith Verstegen

Author: Mónica Magán da Fonseca

Supervisor: Judith Verstegen

ERSLAre wind farms responsible for frost damage in grain fields along a ridge in Australia?

Grain farmers around a newly developed wind farm on a ridge in Australia observed a significant increase in frost damage since the establishment of a wind farm. The hypothesis from climatologists is that under certain synoptic weather conditions, the wind farm can impact local wind patterns that may lead to anomalous micro-climates around the farmlands in the vicinity of the existing wind farm. The aim of this project is to support the farmers by a data driven approach to study changes in surface temperature patterns before and after the development of the wind farm and answer the question if increasing frost occurrence can be attributed to the new wind farm. In this context, this thesis would approach the problem from a satellite perspective. Lacking a sufficient number of climate stations, time series of multi-sensor remote sensing data (e.g. MODIS, Landsat) will be processed and analyzed this project.


The thesis will be conducted in collaboration with the Center for Atmospheric Research in Christchurch, New Zealand. We are looking for a motivated candidate, ideally with previous experience in remote sensing analyses. For more information, please contact Hanna Meyer ( (IfGI)

Contact: Hanna Meyer

Author: Boris Stöcker

Supervisor: Hanna Meyer

Co-supervisor: Edzer Pebesma

Externalerror prevention in WebGIS

Contact: Christian Kray

Author: Sadegh Karampanah

Supervisor: Christian Kray

Co-supervisor: Morin Ostkamp

STMLValidation of Earth observation cloud-processing services

openEO develops an open API to connect R, Python and Javascript clients to big Earth observation cloud back-ends in a simple and unified way. Back-ends process user-defined algorithms on remote sensing data sets - usually image-based - within their cloud infrastructure. An important aspect is to facilitate users to switch between back-ends easily while still getting consistent and comparable processing results. Back-ends use different IT infrastructure and software to process data although they share the same specification for processes and for communication between clients and back-ends: the openEO API. It is still necessary to ensure that processes comply to the specification. As a consequence, the results from back-ends are often not comparable by default and need to be checked for compliance with the specification. One way to ensure compliance is by processing a certain standardized, reference data sets and validating the results. The openEO project still has to select such data sets. Additionally, the differences in infrastructure and software may eventually lead to at least small differences in the processing results, either due to rounding in floating point arithmetic or implementation details. Therefore there needs to be a certain threshold that the results are allowed to differ. This thesis aims to solve the issues raised by

  • defining which aspects an image-based data set need to fulfil for our validation purposes,
  • selecting suitable image data sets for validation purposes,
  • defining the concrete rules and a workflow for validation,
  • and implementing a prototype for the specified workflow.

The scope of the thesis can be adapted to to fit the requirements of either a bachelor thesis or a master thesis. Some more information can be found in the corresponding openEO API GitHub issue.


Contact: Matthias Mohr

Author: Simon Schulte

Supervisor: Edzer Pebesma

Co-supervisor: Matthias Mohr

GeoSimParameterbasierte Identifikation von Realdaten-Streckenabschnitten im OpenDRIVE-Format

Parameterbasierte Identifikation von Realdaten-Streckenabschnitten im OpenDRIVE-Format

Contact: Judith Verstegen

Author: Peter Konopatzky

Supervisor: Judith Verstegen

ERSLComparison of spatial validation strategies in machine learning for environmental science

A major task in environmental science is to obtain spatially comprehensive data from limited field samples (e.g. climate stations, soil profiles, vegetation records,...). This is often done using machine learning algorithms that learn the relationships between field data and remotely sensed predictor variables (e.g. from satellites). The developed model is then used to make spatial predictions for the entire area of interest (i.e. create a "map" of the variable of interest).
Such a map is only valuable when the error of the model is known. The error assessment however causes major difficulties for models with spatial dependencies and cause standard validation methods to fail. There is increasing consent in literature that spatial validation is necessary and several strategies have been proposed (e.g. Roberts 2018, Meyer 2018, Valavi 2018, Brenning 2012). However, little research is done on how different strategies compare, which however is important to know for model comparisons and for finding the "right" validation strategy for a dataset.
This project aims at comparing and (as far as possible) evaluating different validation strategies for machine learning based spatial mapping of environmental variables. Usually in most projects the "true" performance is hard to assess due to the fact that the reference data are limited. Therefore we want to adress the problem using reference data that are available in a spatially continuous way hence providing a continuous reference. Such data are rare but a possible research task would be available in the field of rainfall monitoring: The RADOLAN dataset contains high quality rainfall data from a radar network in 1km spatial resolution and serves as an excellent reference set. We then want to model rainfall for Germany using data from the the geostationary satellite sensor MSG SEVIRI and auxiliary predictor variables such as elevation (see Kühnlein 2014 for the idea of mdoelling rainfall based on MSG SEVIRI).
Using the RADOLAN data we simulate different sampling designs (hence we simulate raingauges in different constellations: random, clustered,...) and train machine learning models to predict rainfall from MSG SEVIRI in a spatial way. Different strategies for spatial model evaluation are then compared.


R programming is an advantage, interest in working with machine learning algorithms


Brenning, A. (2012): Spatial cross-validation and bootstrap for the assessment of prediction rules in remote sensing: The R package sperrorest. IEEE International Geoscience and Remote Sensing Symposium.

Kühnlein, M., T. Appelhans, B. Thies, and T. Nauss, 2014: Precipitation Estimates from MSG SEVIRI Daytime, Nighttime, and Twilight Data with Random Forests. Journal of Applied Meteorology and Climatology, 53 (11), 2457–2480.

Meyer, H., Reudenbach, C., Hengl, T., Katurji, M., Nauss, T. (2018): Improving performance of spatio-temporal machine learning models using forward feature selection and target-oriented validation. Environmental Modelling & Software 101: 1-9.

Roberts, D. R., V. Bahn, S. Ciuti, M. S. Boyce, J. Elith, G. Guillera-Arroita, S. Hauenstein, J. J. Lahoz-Monfort, B. Schröder, W. Thuiller, D. I. Warton, B. A. Wintle, F. Hartig, and C. F. Dormann, 2017: Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure. Ecography, doi:10.1111/ecog.02881.

Valavi R, Elith J, Lahoz‐Monfort JJ, Guillera‐Arroita G. blockCV: An r package for generating spatially or environmentally separated folds for k‐fold cross‐validation of species distribution models. Methods Ecol Evol. 2018;00:1–8.

Contact: Hanna Meyer

Author: Marc Dragunski

Supervisor: Hanna Meyer

Co-supervisor: Edzer Pebesma