Name: Babs Hagendoorn

Master Thesis

Supervisors:
Dr. A.C. Seijmonsbergen and
Dr. J.G.B. Oostermeijer

Year: 2015

Relating Biodiversity to Geodiversity in a Mountain Ecosystem in Vorarlberg

The decline and extinction of species and the loss of habitat and landscape connectivity have resulted in an increased importance of biodiversity. A relatively new approach is to focus on the abiotic environment instead of on individual species. Diversity of geological, geomorphological and soil features is also referred to as "geodiversity". It is hypothesized that a high geodiversity results in a high biodiversity since a diverse geo-environment provides lots of space for species to inhabit. The aim of this study was to develop a GIS-based method to quantify and compare geodiversity and biodiversity. This was done in the alpine area Vorarlberg. The method consisted of 1) dividing the area into equal cells using a grid, 2) calculating the geodiversity index and 3) biodiversity index in each grid cell, 4) investigating the relationship between the geodiversity and biodiversity indices, 5) adjusting the geodiversity index and examine if the new geodiversity index could explain a greater part of the biodiversity variability and 6) adding ranking to investigate if this could improve the relation. The geodiversity and biodiversity index were successfully created in a GIS based manner and a positive relation between the two indices was observed. Adjusting the geodiversity index showed that elevation, slope and solar radiation diversity were the most important parameters for explaining biodiversity. Adding a ranking to the different geodiversity parameters did not improve the geodiversity index greatly. Altogether, this study shows a successful way of calculating a geodiversity index and correlating it to biodiversity. The results showed that a maximum of 12.5% of the variety in biodiversity could be explained by geodiversity, adjusting the parameters could improve this in further research.

Keywords: Geodiversity – Biodiversity – Conservation - Vorarlberg

The map (click here) shows the results of the regional susceptibility analysis

Back to top

Name: Erik Spronk

Master Thesis

Supervisors:
Dr. A.C. Seijmonsbergen and
Dr. L.H. Cammeraat

Year: 2014

Coping with Natural Hazards in the Alps. A multi-criteria GIS-based management-directed natural hazard assessment using multi-temporal and multi-scale approaches

Natural hazards receive frequent attention in the media. Due to the increased world population and the more intensive land use, these natural hazards form an increasing potential detrimental thread. Especially in countries with a higher welfare standard, there is a heed for these menaces, because the means to counter these are available. In recent years, on-going research has been carried out to improve and automate the way these hazards are identified. In this thesis, a new GIS-based and automated method will be explored for an area in Vorarlberg, Austria. First, a multi-criterion regional susceptibility analysis is made. These criterions are based on a literature study. Then, the results of the analysis are used to focus on a local scale. For these areas, a hazard assessment is made. Thirdly, this analysis will also be carried out based upon more recent data, by which changes will become clear. The results will be presented as maps, datasets, and models, as well as a workflow for policy purposes. In addition, two case studies have been developed.

Keywords: natural hazards – hazard assessment – GIS – susceptibility analysis - Vorarlberg

The map (click here) shows the results of the regional susceptibility analysis

Back to top

Name: Rachael Chambers

Master Thesis

Supervisors:
Dr. A.C. Seijmonsbergen and
Dr. K. F. Rijsdijk

Year: 2014

Development of a geodiversity index on a volcanic island: Tenerife, Canary Islands

The variability in abiotic factors within an environment provide the building blocks to variations in biotic factors. Therefore, there is a need to study the concept of geodiversity at the same level as biodiversity. The aim of this study is to develop an index which measures geodiversity on a volcanic island, through the use of geographic information systems (GIS), which can ultimately be used to relate to a (potential) biodiversity index. The island of Tenerife, in the Canary Islands is the chosen case study area. The method proposed here is based on that of Serrano et al. (2009) and Pereira et al. (2013). Two formulas are tested, one at the whole island scale and another at a case study scale. A grid is used to measure the geodiversity index (GDI) per 500 m2 quadrant across the whole island. The formula tested here, GDI1, is based on geology, pedology and slope angle. Due to the fact that there is no available geomorphological map for the whole island, ve case study areas have been selected to represent the various geomorphology of a volcanic island so that it can be included in the geodiversity index. The geomorphology is delineated at these sites and a formula is then applied to the case study areas based on geomorphology, geology and slope angle at two dierent scales: GDI2 at 500 m2 and GDI3 at 250 m2. This results in three dierent geodiversity indices, GDI1, GDI2, and GDI3 that delineate their respective areas into ve categories of geodiversity: Very Low; Low; Medium; High; and Very High. The accuracy assessment based on Congleton (1991) rates their overall accuracy at 57%, 60% and 54%, respectively. A potential biodiversity index (PBI) is derived from potential vegetation. Calculations are made to determine where there are low, medium and high areas of PBI and GDI, thus identifying hotspots of geodiversity and potential biodiversity. The methodological design has room for some improvements, but what is presented here provides a starting point for many areas of future research. Particularly, in analysing the geodiversity of volcanic islands and the development of a useful tool to assessing matching index values, which speeds up the recognition of hotspots for conservation.

The figure (click here) shows the final geodiversity index map based on geology, slope angle and pedology. Black lines indicate manual delineation of areas with similar GDI values, the black numbers indicate the majority GDI value within that area. GDI values are based on five categories: Very Low (7-10), Low (11-14), Medium (15-18), High (19-22) and Very High (>23).

Back to top

Name: Janina Bösken

Master Thesis

Supervisors:
Dr. L.H. Cammeraat and
Dr. A.C. Seijmonsbergen

Year: 2013

Geomorphology and landscape evolution of the Parroquia-Zarzilla region, SE Spain

The study area is located in the catchment of the Guadalentín River in the eastern Betics, a key region for desertification studies. Desertification is one of the biggest problems facing European Mediterranean countries (Geeson et al., 2002). Understanding how landscapes are affected by degradation will broaden our knowledge in order to reduce human impact on land. The geology of the eastern Betic Cordillera offers an interesting research setting in the convergence zone of the African and Eurasian plates. The research area has not yet been subject to other studies and forms a key area in understanding the landscape evolution of the region. The objective of this thesis is to reconstruct the landscape evolution of the Parroquia-Zarzilla region since the late Pliocene. Investigated research questions are: 1. What can we learn from the present geomorphologic features about the evolution of the landscape? 2. Do the lithological and sedimentological findings match up with findings of other research in the region and which new ideas can be deduced? 3. How do climate fluctuations influence the evolution of the landscape? 4. Considering the evolution of the landscape in a broader picture, what can be said about geological and neo-tectonical drivers of the evolution? In order to answer these questions the area was investigated through the generation of a geomorphological map and geomorphological cross sections on-site. Prior to the field work, a literature research was conducted and a GIS database with satellite images, topographical, geological and soil maps was created. After fieldwork drawn field maps were digitized to visualize the findings in a geomorphological map on a scale of 1:5,000. The results were compared to other studies conducted in the region. Finally, the map was transferred into a 3D-model in ArcScene to draw regional conclusions about the landscape evolution. The result of this master thesis is a geomorphological map that contains a fluvio-lacustrine landscape in the northwest, an alluvial fan in the southwest and a bedrock-pediment dominated landscape in the north and south, as well as a lacustrine landscape in the southeast. The bedrock-pediment dominated landscape has two valleys – a steeper v-valley in the north and a less steep valley with highly eroded terraces in the south. The generation of pediments is controlled by phases of neo-tectonic uplift and neo-tectonic stability. Strong earthquakes possibly triggered mass movement processes that blocked the Luchena River and created a lacustrine environment in the northwest. When the blockage was overcome through erosion, the paleo-lake emptied suddenly, and a fluvial dominated landscape developed. This blocking/deblocking process happened several times, as three limnic terrace levels (10 m, 30 m, 50 m height above floodplain) are observed nowadays. A river used to flow through the southern valley, but was cut off from its source by neo-tectonic uplift of the bedrock-pediment dominated part of the research area. The cut-off river changed its direction to connect with the downstream located Guadalentín Basin via the Luchena. Comparison to other research shows similar internal dynamics, where the Guadalentín River was blocked creating a paleo-lake to the east, but does not allow a correlation to climatic conditions as driver for terrace accumulation and river erosion. This suggests that internal dynamics are of higher significance than external forces. Finally a rough sequence of events is proposed: 1. formation of eastern Betic bedrock (Cretaceous-Neogene); 2. development of alluvial fan in subsidized valley, drainage in southern valley (Pliocene-Pleistocene); 3. uplift of bedrock hills, pediment formation, calcrete development, cut-off southern river (Pliocene-Pleistocene); 4. Luchena river blockage, development of 50 m and 30 m terraces (Lower Pleistocene); 5. last 4.5 m of eastern basin fill (after 19,500 BP); 6. two phases of uplift (19,500-2,500 BP); 7. development of paleo-lake east (17,000-13,800 BP); 8. Lorca fan aggradation (7,900-4,200 BP); 9. deposition of limnic L2 terraces east (2,000-1,870 BP); 10. development of fluvial and limnic 10 m, 5 m terraces & current floodplains (1,870-1,300 BP and/or 485-50 BP). Due to difficulties in classifying the temporal events of the landscape evolution precisely, it is strongly advised to conduct further research in the area with regard to dating of terraces and pediment surfaces, measurement of sedimentation rates and effects of human impact.

Back to top

Name: Bep Schrammeijer

Master Thesis

Supervisors:
Dr. L.H. Cammeraat and
Dr. A.C. Seijmonsbergen

Year: 2012

A modern Tragedy of the Commons: An ecosystem service analysis of development along the Noordzeekanaal

In order to achieve sustainable development the physical environment needs to be organised and used in a way that minimises ecosystem degradation and enhances the benefits that society obtains from them. Ecosystems provide crucial services vital to human well-being and development is directly and indirectly dependent on ecosystem services. The Millenium Assessment conducted in 2005 found that 60% of ecosystem services assessed face degradation or unsustainable use. Scientists at the Stockholm Resilience Centre pointed out in 2009 that more planetary boundaries are being approached than just global climate change, for example land use change and chemical pollution. Currently spatial planning is guided by Environmental Impact Assessment to identify potential impacts on the environment. However, ecological aspects that are difficult to quantify are easily neglected, negated or insufficiently understood. Where they are identified, or potential cumulative effects are identified, the process is not comprehensive enough to ensure that the identified effects are addressed. This creates a social dilemma as longer-term social and ecological costs tend to be shoved aside in the interest of short-term economic gain. Valuing ecosystem services can help to enhance decision making, highlight equity issues and facilitate sustainability. By valuing ecosystem services decision makers have a more comprehensive view of costs and benefits to society posed by potential developments. It can highlight the necessity of ecosystems, help avoid irreversible degradation and highlight possible negative side effects. The main challenge is the development of a transparent method of ecosystem service analysis that can be used by relevant stakeholders and be integrated into the decision making process. This research assesses the impact of current uses of, and future plans for, the Noordzeekanaal and surrounding area on ecosystem services that regulate air and water quality. Using a framework recommended by KNAW for earth science research on the behalf of spatial planning past, present and potential future land cover maps are developed and inform the calculation of the use and value of several regulatory ecosystem services. In order to do this a tool has been developed that utilises existing spatial data to calculate and visualise supply, demand and values of selected ecosystem services in a GIS (Geographic Information System) program. Results indicate that the current landscape of the Noordzeekanaal area cannot regulate all the emissions to the air that are produced in the area. This indicates that a cost is being carried by society and the quality of common pool resources is being affected while it is not certain who benefits from these activities.

Back to top

Name: Manon de Vries

Master Thesis

Supervisors:
Dr. A.C. Seijmonsbergen and
Dr. L.H. Cammeraat

Year: 2012

A rapid GIS-based hazard zonation method for use in managing ski areas with a case study in Vorarlberg, Austria

Natural hazards in mountainous terrain include the occurrence of karst, rockfall and landslide processes, and poses threats to the natural and human-built environment. Mapping the locations sensitive to these types of mass movements and their activity is fundamental for management purposes to prevent future damage of e.g. skiing infrastructures. Therefore, Geographic Information System (GIS) models are designed to rapidly outline the spatial distribution of the existing mass movements in an alpine region. The safety factor analysis is used to show a first overview of stable and unstable areas, with the help of soil thickness and groundwater level height calculations. It serves as input for the landslides GIS model. Input parameters for the karst and rockfall models include lithology and slope gradient characteristics, based on a geology map and a high resolution Digital Elevation Model (DEM) derived from laser altimetry data, both covering the area of interest: the Lech municipality in the most western federal state of Austria, Vorarlberg. The three GIS models are set up as executable ArcInfo toolboxes, by which the dynamic model input data can be edited if updated data becomes available. On community scale, the hazard zonation maps of karst, rockfall and landslides give a sufficient outline of hotspots where danger is expected, so policymakers can make appropriate decisions regarding future infrastructural planning, with the help of layered PDFs and geographic visualization files for use in virtual globe software, in which the output maps are provided.

The figure (click here) shows the outcome of the safety factor calculation, with red colours being unstable slopes, and green colours stable slopes. The SF value layer is visualized on top of the outcome of the soil thickness model (h in the safety factor formula [m]), and the groundwater level height (mz in the formula [m]). Vertical exaggeration: 1,5 times the original elevation.

Back to top

Name: Cindy Teeven

Master Thesis

Supervisors:
Dr. A.C. Seijmonsbergen and
Dr. L.H. Cammeraat

Year: 2012

A multi-criteria geomorphological analysis method for geodiversity assessment in Vorarlberg, Austria

The conservation of geodiversity is important since landscapes and ecosystems depend on their underlying soil, geology and geomorphology. This study contributes to and refines the method for mapping and assessing potential geodiversity by newly including the influence of the socio-economic factors. To meet this objective, four research questions are formulated: 1. Can the segmentation and the classification steps which are necessary for geomorphological mapping be improved within the area of interest? 2. How are geoconservation features influenced by socio-economic factors? 3. How can socio-economic factors be included in the ranking and weighting system of geoconservation? 4. How do socio- economic factors influence geodiversity over time? The mapping of geodiversity is based on four criteria: Scientific relevance, Frequency of occurrence, Vulnerability and Disturbance. In the new method, 15 socio-economic factors (e.g. cable railways and ski runs) are used to determine the Disturbance criteria for each landform in an alpine area in Vorarlberg, Austria. The socio-economic factors affect Disturbance within a specified zone of influence, the buffer. A first model calculates the disturbance of a single buffer, a second tool is created for two buffers, and the third one combines all the socio-economic factors into one feature class which will leads to the final potential geodiversity map. The ranking values of the disturbance and the size of the buffer range are not fixed and can therefore be easily altered in an Access database which is directly linked to ArcGIS. Four datasets with mapped and classified geomorphological landforms resulted in different spatial distribution of the potential geodiversity in the same key area. This is due to the difference in geomorphologic landform classification. Overall, 80-90% was classified as low potential geodiversity, 3-7% as moderate and 2-12% as high. In the key areas Kriegerhorn and Zürsersee, the area mapped as high potential geodiversity decreases by 70.000 m2 and 32.000 m2 respectively after implementing the new classification method which includes the socio-economic factors. The new method can also be used for future management in the ski region of Lech. A future scenario was implemented where two ski areas were connected by two new cable railways. Applying the new methodology on this scenario in the key area Auenfeld, potential geodiversity was shown to decrease form high to medium in an area of 10.000 m2. Main conclusion of the study is that socio-economic factors do influence the potential geoconservation value in the municipality of Lech since they a) fragment the landscape and b) contribute to a lowering of the disturbance factor, which is used in the existing weighting and ranking scheme.

The figure (click here) shows potential geodiversity with the influence of the SE-factors with the CAM dataset (zoom on key area Kriegernhorn). The potential geodiversity is downgraded where the ski run and local road intersect the karst landform. Base map: orthophoto (true colour).

Back to top

Name: Marijn van Veelen

Master Thesis

Supervisors:
Dr. A.C. Seijmonsbergen and
Prof. Dr. Ir. W. Bouten

Year: 2012

Development and application of tools for the use of multi-temporal LiDAR data in analysis of topographic change following the 2005 flood - A case study for the Meng valley in Vorarlberg, Austria

In the last decades, airborne laser altimetry (also LiDAR) rapidly become the most used topographic mapping technique in several landscapes. The accuracy of approximately 15 cm in the vertical make the technique appropriate for the detection of small scale morphological changes by comparing DTMs obtained at different times. A methodology is developed for the processing of LiDAR point data to construct DTMs. Furthermore, accuracy assessment techniques are developed, making use of the data abundance of LiDAR surveys. The methodology is applied for an alpine catchment in Vorarlberg, Austria. The area is chosen because of the high morphological dynamics in this area, which make it possible to use laser scanning data of only two years apart. It is shown that although vertical morphological changes of as little as 0.3 meter can be identified from multi-temporal LiDAR derived DTMs, this is only so for areas which are not vegetated or are vegetated, but have a low relief. The presence of vegetation clearly raises the accuracy level of the LiDAR data, and while the original data accuracy is not affected by the terrain relief, the accuracy of the LiDAR derived DTM is sharply decreased. The use of terrestrial laser scanning might provide a way to more detailed topographic mapping in such areas, and in specific the combined use of the techniques is deemed worthwhile. Although the presence of erosion or deposition is obscured by the presence of vegetation, with expert knowledge also the data in more complex areas can be used for change detection assessments, although a lot more postprocessing might be needed.

The figure (click here) shows a map of the elevation differences (m) between the unfiltered DEMs from 2004 and 2006 LiDAR data. While patterns of geomorphological activity are clearly visible in the stream channel, they are obscured by high rates of apparent change due to differences in forest stand in the vegetated parts of the area.

Back to top

Name: Siska Leek, Niels Raes and Harry Seijmonsbergen

Course: Advanced GIS and Remote Sensing Mini-Project

Year: 2011

A method for the detection of limestone assisted by GIS and remote sensing in the rainforest of Borneo

Multi-source data is used to prepare a potential limestone distribution map of part of Borneo’s rainforest. Supervised classification, DEM-analysis and expert interpretation is used in ERDAS Imagine 10, ArcGIS, ArcScene and combined into a potential limestone exposure map including geological structure. Firstly a supervised classification of potential limestone exposures in the Sankulirang peninsula in Borneo using ERDAS2010 was prepared showing surface limestone exposures. Secondly, a landscape depression distribution map (inArcGIS10) was prepared using GDEM data and thirdly a geological interpretation map (made with ArcScene) based on Landsat image interpretation and manual digitalization was prepared. The final potential limestone map is assumed to better reflect limestone occurrences than existing geological information alone.

The figure (click here) a part of the supervised land cover classification map (left) and manually digitized geological interpretation map overlain with landscape depressions (right) for an area in Borneo.

Back to top

Name: Friso Dekker

Course: Advanced GIS and Remote Sensing Mini-Project

Year: 2011

Using object-based classification to map land cover on the island of Tenerife

Research has indicated the added value of object-based classification over pixel-based classification, with classification accuracies being up to 36.77% higher. Both methods were used to perform a supervised classification of the land covers on Tenerife; the pixel-based method in ERDAS Imagine® 9 and the object-based method in eCognition® Developer 8. In ERDAS, 11 categories of land cover were generated, in eCognition some land covers were subdivided to create a total of 15 land cover categories. An Accuracy Assessment was prepared in ERDAS, resulting in 64.8%. In eCognition, a Best Classification Result was generated, resulting in an average mean of 0.759 for all classes. All objects thus have a 75.9% similarity with the classes they are assigned to. The Object-based classification has an accuracy of 11.1% more if you consider the 75.9% of this classification to be an accuracy assessment. Furthermore, the object-based classification method has a better ability to classify urban areas, different lava flows and different types of agriculture.

The figure (click here) shows the classification of land cover on Tenerife, as output from eCognition.

Back to top

Name: Jesus Aguirre Gutierrez, Harry Seijmonsbergen, Jan Sevink

Course: Internship

Year: 2011

Water pollution hazards from geological formations in the Rio Santa region, Peru

De Callejon de Huaraz, a tectonic basin between the Cordillera Blanca en Cordillera Negra in Northern Peru, forms the catchment of the Rio Santa, one of the major rivers in Peru. The catchment comprises the high, glaciated summits of the Cordillera Blanca (the largest glaciated region of the Tropics and the subtropical lower parts of the Rio Santa catchment in the north of the province Ancash. This area is characterized by a variation in geology, such as mineralized formations (polysulfide ores). Exceptions are the batholitic (granodiorite / tonalite) sections of the Cordillera Blanca. The quality of the water in the many rivers and lakes differs as a result of the often strong availability of sulphide ores in the parent rocks. Through oxidation heavy metals can be mobilized. As a consequence, the drainage derived from such rocks almost always is characterized by its low pH values, and sometimes high concentrations of heavy metals. In case of batholithic or limestone derived parent bedrock formations (Mesozoic in age), neutral pH values are common, and low to zero contamination with heave metals occur.
During this project the following GIS data have been prepared:

• DEM for the study area obtained and pre-processing in order to eliminate errors
• Construction of hydrological network of the study area Rio Santa
• Delineation of catchments present in the study area
• Digitization of the geological units
• The pollution values per geological formation were obtained
• Digitization of the mines present in the study area
• Preparation of the shape file showing the final pollution hazard value per catchment
• By overlapping the information of the geological maps, the catchments layer and the rivers map a water contamination map were constructed

Back to top

Name: Cindy Teeven

Course: Advanced GIS and Remote Sensing Mini-Project

Year: 2010

Semi-automated reconstruction of glacier extent in the Cordillera Blanca, Peru (1989 - 2007) using ArcGIS

Glaciers are known to be very sensitive when climate changes. The variability in temperature and precipitation has a great effect on the position of the glaciers in the past. With the data of images of 1989, 1991, 1999 and 2007 a classification as well a reconstruction of the past position is made. A methodology, based on a semi-automated classification, has been made within ArcMap. A three dimensional view of the final product is created in ArcScene. With various tools in ArcMap, a reconstruction has made of the former position of the glaciers in four different years using a SRTM DEM as background. The final product (figure right) was the result of both manual and automated procedures.

The figure shows the method used in ArcGIS (click here) and a reconstruction of the glaciated area during 4 different years visualized on a digital elevation model (click here)

Back to top

Name: Yiting Wang

Course: Advanced GIS and Remote Sensing Mini-Project

Year: 2009

Study on Land Use Change and Urban Sprawl in Nanjing, China with GIS and Remote Sensing

Nanjing, the capital city of Jiangsu province, locates in the middle-lower Yangtze Plains, within latitudes of 31.14^o to 32.37^oN and longitudes of 118.22^oE to119.14^oE, which is also the junction of Yangtze River and eastern costal development zones. Because of its central geographic location, Nanjing is among the most important cities in Yangtze River Delta and has experienced rapid development and huge reformations since the Chinese Economic Reform in 1979. Stimulated by economic development and population growth, the urban area in Nanjing has been sprawling, and land use and land cover have changed significantly. This study focuses on the land use change and urban sprawl in Nanjing city between 1979 and 2001. Three remotely sensed data in 1979, 1988, and 2001 are used to extract land use information; spatial analysis methods are utilized to analyze land use change; the mechanism of such change is also discussed to find out the main driving forces.

Please click here and here for full-sized figures.

Back to top

Name: C. Tovar & A.C. Seijmonsbergen

Course: Advanced Remote Sensing Mini-Project

Year: 2008

Land cover classification of a Landsat ETM+ satellite image using pixel and object based techniques

A Landsat ETM+ image of northern Peru in the vicinity of the city of Cajamarca was classified using ERDAS Imagine and Definiens Developer. First an unsupervised classification using 20 classes was made which included preliminary knowledge of the local ecosystems and presence of the altitude of natural occurring habitats. This preliminary classification knowledge was used in a next classification with 17 final classes in ERDAS imagine. The same training samples were used in the object-based segmentation and classification using the Definiens Developer software. Accuracy assessments were made for all tests. Comparison of all tests showed that the 17 class in a final classification using ERDAS Imagine gave best statistical performance. The map shows the classified Landsat ETM+ image. It is concluded that a preliminary classification with trained unsupervised knowledge can serve as an efficient first recognition for geo-ecological mapping in relatively unknown areas.

Please click here for the full-sized map result.

Back to top

Name: Timo Kamminga

MSc Thesis

Year: 2008

Vulnerability assessment of potential soil erosion; a case study on Mauritius

A vulnerability assessment is a useful tool to research the decrease and fragmentation of natural habitats for sustainable landscape management. Soil erosion is a important abiotic factor that can cause landuse changes. To express and evaluate the vulnerability to soil erosion on Mauritius two vulnerability models were used, the Revised Universal Soil Loss Equation (RUSLE) model, and an adapted version of the South Pacific Applied Geoscience Commission (SOPAC) model. The aim was to express vulnerability to soil erosion in GIS by means of the RUSLE, to evaluate the abiotic components of the SOPAC model and discuss how accurately the results represent the spatial vulnerability in terms of soil erosion compared to the results of the RUSLE model. The RUSLE model showed high vulnerability in the southeast of the island, and in the cyclone season months February and March. Where rainfall exceeded 150-200 mm higher vulnerability was predicted. The adapted SOPAC model proved insufficient for an extensive comparison, due to incompatibility of factors. General trends were visible, some of which corresponded with the trends observed in the RUSLE model. Further research could incorporate elements of the adapted SOPAC model to improve the vulnerability assessment based on the RUSLE model.

Please click here for a map result (jpg).

Back to top

Name: Rui Cao

MSc Thesis

Year: 2007

Extraction of Forest Stand Information using LiDAR in Vorarlberg, Austria

Forest stand information is valuable and important for forest management and policy making. In this study, attempts are made to extract two kinds of forest stand information: forest structure and aboveground biomass. Forest structure plays a vital role in mountain forests' protective capacity against natural hazards and aboveground biomass represents potential carbon which can be added to atmosphere as greenhouse gas. The estimation of aboveground biomass helps countries calculate their carbon stock in the forests and meet the emission reduction targets specified by international treaties. Light Detection and Ranging (LiDAR) has the ability to obtain 3- dimensional information of earth and canopy surface. This study aims at finding an automated method to assess forest structure using landscape metrics, estimate aboveground biomass at stand level and analyze the interrelations between forest stand characteristics and their surroundings. Based on multi-resolution segmentation, forest stand objects and tree crown objects are obtained, tree crown objects were classified into four height classes and overlaid with forest stand objects. Two landscape metrics are calculated for each stand in the study area: Shannon Evenness Index (SHEI) and a modified Division Index (mDIVI). Canopy density values of each height class were used to distinguish uniform forest stands from multilayered stands and SHEI was used as assistance. Through mDIVI and canopy density value, forests in the study area were classified into seven forest structure types. Local Maxima technique was used to detect single tree tops and to calculate the canopy height of each tree, then tree height value was input into allometric equations, Diameter at Breast Height (DBH) and aboveground biomass were obtained. The total aboveground biomass of a stand was calculated by adding the biomass amounts of all the trees growing within this stand. Through the approach in this study, forest managers and policy makers can easily evaluate the forests' growing stages and fuel load level, carbon stock at country scale can also be calculated in an efficient and cost-effective fashion if LiDAR data are available.

Back to top

Name: Stefan Fritz

M.Sc Thesis

Year: 2003

Reconstruction of Late Glacial cirque and valley glaciers and modelling changes of the Equilibrium Line Altitude with GIS software

This study tests a method to reconstruct former cirque and valley glaciers in high-mountain environments. Special attention is paid to the snowline, or Equilibrium Line Altitude (ELA), and the influence of local circumstances on the snowline depression. The execution of this study is carried out by implementing field and expert knowledge into a Geographical Information System (GIS). The main objectives of this study are:
(1) To categorize Late Glacial ice-marginal landforms according to both geomorphological and glaciological features in the Rätikon mountains, Vorarlberg, Austria;
(2) To reconstruct glaciers making use of these glacial remnants, with taking notice of disturbances of the original situation by younger processes, and other limiting conditions;
(3) To distinguish individual Late Glacial recessional stadials; and
(4) To relate the recessional stadials to each other, locally, and to existing literature data on glacial chronostratigraphy, regionally. The figure shows an example of a GIS map with reconstructed glacier positions during the Bühl deglaciation phases.

Please click here for the full-sized map result.

Back to top

Name: Jesus Aguirre, Harry Seijmonsbergen and Joost Duivenvoorden

M.Sc Thesis

Year: 2010

Optimizing land cover change detection using combined pixel-based and object-based image classification in a mountainous area in Mexico

A combination of object-based and pixel-based image classification methods and post-classification object-based change detection is applied to Enhanced Landsat Thematic Mapper images for optimizing land cover classification and change detection in a mountainous region in northern Mexico. The land cover categories with the highest individual classification accuracies for coniferous forest, shrubs and deciduous forest, bare soil, urban areas and water bodies in the object-based and pixel-based classifications were extracted and combined into single combined classification layers. Comparison of the overall classification accuracies of the object-based, pixel-based and the combined maps shows that the combination method produces higher overall accuracies. The combined classification maps are used as input for post-classification object-based change detection analysis. The combination of pixel-based and object-based change detection method for Landsat imagery leads to improved classification and change detection classification and has potential to similar mountain areas.

Figure shows examples of classifications of a Landsat satellite image of 1999 and 2006 and the changes detected using a post-classification technique.

Please click here for a full-sized figure.

Back to top

Name: Guusje Hoek

M.Sc Thesis

Year: 2008

Classification of a Dutch dune reserve with ERDAS imagine and Definiens Developer

The objective of this remote sensing study was to compare accuracies of pixel and object-based classification RGB and NIR images of the Kraansvlak coastal dune system in the Netherlands. Therefore, an evaluation of the influence of color balancing was necessary. With Definiens Developer rule sets for different shrub vegetations were tested for the entire Kraansvlak area. Field samples of the various classes were collected in the Kraansvlak. With these training sites a classification of a part of the Kraansvlak was done with ERDAS Imagine 9.1(pixel-based) and Definiens Developer 7.0 (object-based) and the accuracy results were compared. With Definiens Developer color balanced and original images were compared and the entire Kraansvlak was classified with the color images. For the RGB image an overall accuracy of 59.52% is obtained with ERDAS Imagine 9.1 and an overall accuracy of 67.6% is obtained with Definiens Developer. For the NIR image an overall accuracy of 66.67% is obtained with ERDAS Imagine 9.1 and an overall accuracy of 76.5% is achieved with Definiens Developer. For the color balanced images an overall accuracy of 64.8% is acquired. For the entire Kraansvlak an overall accuracy of 73.6% was obtained and rule sets are found for the shrubs which resulted in an overall accuracy of 71.6%. It can be concluded that the highest accuracies are obtained with NIR images in both approaches. However, a higher accuracy was obtained with Definiens Developer, the object based approach. In general, color balancing probably has a negative influence on the classification. The figure shows 4 examples of classifications using RGB and NIR images and corresponding overall accuracy of the Kraansvlak study area.

Please click here for the full-sized map result.

Back to top

Name: Jelle Harms

MSc Thesis

Year: 2007

Distribution and potential habitat mapping of herpetofaunal species at different scales

Herpetofaunal species decline is a well known issue nowadays. Habitat loss and degradation, introduced invasive species, deceases, pollution, and global climate change are all recognised to be of significant contribution to herpetofaunal diversity decline. To counteract this phenomenon, niche modeling is a powerful way of evaluating a species'potential habitat and assessing most relevant environmental variables for herpetofaunal species. Based on this knowledge, proper land management can be improved which will lead to better decision making on conservation. Species presence observation points in combination with environmental variables were used to assess potential distribution, diversity and variable importance of various herpetofaunal species at three different scales. This was done, using a niche modeling technique known as Maximum Entropy Species Distribution Modeling (Maxent). At a local scale, Sand lizard (L. agilis) habitat was modeled using ASTER and Digital Elevation Model (DEM) (derived) variables of two Northern Dutch islands Vlieland and Terschelling. Most relevant environmental variable at this scale was found to be the classified ASTER. At a regional scale, potential distribution and diversity was modeled for six herpetofaunal species on Crete using 9 different environmental variables. Important factors on this scale turned out to be NDVI and other land cover related variables like Ecoclimap and CORINE. Also the used soil variable had great impact on both variable importance and spatial distribution of the modeled species. At continental scale, six herpetofaunal species were modeled using maximum 18 environmental variables. Categorical data like Ecoclimap land cover and FAO soil have shown to be of significant value when mapping species potential distribution. Aside from these variables, also bioclimatic factors like cloud cover and annual mean temperature and seasonality appeared to be of considerable relevance to explain herpetofaunal species distribution. Furthermore, the potential distribution was compared with IUCN known species range. In general, potential habitat corresponded to the expected species range but suitable habitat was also broadly found outside the IUCN range. In the figure the potential distribution maps of models without categorical data representations of the Maxent model for the Cretan species are shown. Warmer colors show areas with better predicted conditions. From upper to lower the following species distribution map are presented: Bufo viridis, Chalcides ocellatus and Hemidactylus turcicus.

Please click here for a map result (jpg).

Back to top

Name: Leonarde Reyes Acosta

MSc Thesis

Year: 2007

Spatial distribution of the Lycaena dispar batava, from 1990 - 2005
A multivariate assessment

The Dutch Large Copper Lycaena dispar batava is an important endemic species of the Netherlands that depends directly on the unique preserved habitat in the national natural parks De Weerribben and De Wieden (Duffey 1968). The uniqueness of the butterfly habitat consists on a specific sphagnum-reed vegetation combination, which happens in the terrestrialisation succession process of old dug-peat areas (Duffey 1968). These areas have demonstrated to be highly biological diverse, hence different efforts are being carried out to preserve this unique vegetation community along with the animal populations that depend on it. In that context, the following research looks for describing, comparing and assessing the relation of the Large Copper's distributional patterns during time, with its host plant distribution, the Water Dock (Rumex hydrolapathum Huds.), in the national parks "De Wieden" and "De Weerribben". For assessing this relation two methods are going to be applied: GIS based spatial distribution analyses: "Hot spots" and "Spatial Autocorrelation" (Getis and Ord 1992, Griffith 2003) and multivariate variance analysis statistics: Spatial and Hierarchical Variation Partitioning analyses (Chevan and Sutherland 1991, Borcard et al. 1992). The GIS spatial analyses are focused on evaluating the spatial distribution patterns of the Large Copper during time. And the variance multivariate analyses assess the relation of the vegetation associations in specific periods of time (Rumex hydrolapathum varying covertures), physical variables (e.g. temperature and rain-fall) and biological records of the species (e.g spatial distribution, gender ratio, ovoposition ratio, etc.), with the abundance of the Lycaena dispar records. The results show that the Large Copper distributional patterns during time follows a clustered trend, following specific Water Dock vegetation associations (Vegetations of the False Oat grass order 'transitional grasslands' and Forest and shrubs, not specified), as the former depends on healthy Water Dock plant communities (Duffey 1977, Bink 1986). However, as the Water Dock has a wider distribution and more flexible ecological needs than the Large Copper, the multivariate test unveils a clearer effect of the "Water level" during the last years over the butterfly abudance.Moreover, the spatial analyses make evident a strong trend of the Lycaena data set related to the year of sampling, especially in the Weerribben national park. It is concluded, that even though the Lycaena dispar batava is strongly related to the Rumex hydrolapathum covertures and vegetation associations, the Water level regimes would importantly influence the abundance and distribution of the species. The final result comprises a set of maps for the Weerribben-Wieden area describing the distribution of the Large Copper and the Water Dock during specific periods of time, along with the variance partitioning analysis assessment. These findings increases and supports the knowledge about the specific distribution of this particular butterfly, and provides a set of knowledge that could provide important management insights. The map shows historical records of Lycaena dispar. In the Netherlands.

Please click here for a map result (pdf).

Back to top

Name: Marieke Kamp

M.Sc Thesis

Year: 2008

The use of GIS in Forensic Sciences

Geographical Information Systems (GIS) are extensively used in various disciplines to visualise and analyse spatial data. This research is focused on the use of GIS in Forensic Sciences and is divided into two main parts.

The first part is concerned with the exploration of the current applications of GIS in forensic research and how this role of GIS can be expanded for future application based on a literature study and personal interviews with field experts. Although GIS is relatively well-known as a potential tool for forensic data analyses, only few people are actually using it. GIS is likely an effective tool for e.g. analyses of individual tracks, relations between tracks within a single crime case, a combination of track and crime analyses and to analyse the spatial distribution and trends of forensic evidence. Beforehand GIS can be widely used for forensic purposes, (1) scientists should become more familiar with the possibilities of GIS and (2) data should become more widely available in easy accessible databases. In addition, combining current different databases or data sources might have additional value to analyse potential unknown patterns in crime analyses.

The second part of the research is focused to develop two GIS-modules for M.Sc. students Forensic Sciences.

Please click here for an example GIS-module.

Back to top

Name: Stogiannidis G. Efstathios

M.Sc Thesis

Year: 2008

MSc Chemistry: Spatio - temporal mapping of pollutants in the North-Sea

The spatial and temporal distribution of several pollutants in the North Sea bed have been studied and mapped by means of GIS and mathematical- geostatistical software. The data have been collected during monitoring campaigns in the years 1981-2006 by the Dutch National Institute for Coastal and Marine Management (RIKZ). During these sampling campaigns, the particles larger than 63 µm dry weight sediment fraction (clay and silt) were sampled and analyzed for heavy metals, polychlorinated biphenyls (PCBs) and polycyclic aromatic hydrocarbons (PAHs). In this project spatial (regression kriging) and temporal (moving average) interpolation together with validation and cross validation have been performed for 31 pollutants from the RIKZ data with a varying degree of success, depending on the substance and the sampling year. Maps of the area, showing pollutants' distribution and their annual trends have been produced. The concentrations of pollutants in the Dutch Coastal zone show a decrease over time. The PAH ratios' maps suggest point and non point sources of pollution, an increase of combustion sources over the last years and an increase of petroleum sources in specific locations around the middle of the last decade. Maps of uncertainty show that the amount of our model's uncertainty is significant and thus an alternative data treatment is suggested.

The animation (click here) illustrates the temporal change of PCB 138 concentration in the North Sea between 1988 and 2006.

Back to top