Open Access

Analysis and visualisation of movement: an interdisciplinary review

  • Urška Demšar1Email author,
  • Kevin Buchin2,
  • Francesca Cagnacci3,
  • Kamran Safi4, 5,
  • Bettina Speckmann2,
  • Nico Van de Weghe6,
  • Daniel Weiskopf7 and
  • Robert Weibel8
Movement Ecology20153:5

DOI: 10.1186/s40462-015-0032-y

Received: 17 November 2014

Accepted: 2 February 2015

Published: 10 March 2015

Abstract

The processes that cause and influence movement are one of the main points of enquiry in movement ecology. However, ecology is not the only discipline interested in movement: a number of information sciences are specialising in analysis and visualisation of movement data. The recent explosion in availability and complexity of movement data has resulted in a call in ecology for new appropriate methods that would be able to take full advantage of the increasingly complex and growing data volume. One way in which this could be done is to form interdisciplinary collaborations between ecologists and experts from information sciences that analyse movement. In this paper we present an overview of new movement analysis and visualisation methodologies resulting from such an interdisciplinary research network: the European COST Action “MOVE - Knowledge Discovery from Moving Objects” (http://www.move-cost.info). This international network evolved over four years and brought together some 140 researchers from different disciplines: those that collect movement data (out of which the movement ecology was the largest represented group) and those that specialise in developing methods for analysis and visualisation of such data (represented in MOVE by computational geometry, geographic information science, visualisation and visual analytics). We present MOVE achievements and at the same time put them in ecological context by exploring relevant ecological themes to which MOVE studies do or potentially could contribute.

Keywords

Movement ecology Animal movement Trajectories Spatio-temporal analysis Spatio-temporal visualisation Geographic information science Computational geometry Visualisation Visual analytics Interdisciplinary developments

Introduction

Understanding the processes that cause and influence movement is one of the challenges in ecological enquiry with consequences for other disciplines, such as biodiversity [1-3]. Movement ecology investigates fundamental questions about organismal movement, which include why, how, when and where the organisms move and how this process is linked to external factors [1,4]. This knowledge leads to understanding not only movement but also how and why animals use specific resources, how they interact with each other, with other species and with their environment and how they compete and reproduce - the key elements of evolutionary processes that determine survival and fitness [5]. Understanding the processes at the basis of movement will provide the link to population distribution and dynamics [6], which is essential to forecast the impact of human-caused environmental change and outline conservation strategies.

With recent advances in positional technology, ubiquitous accessibility and widespread use of global positioning devices, researchers are now able to track movement at unprecedented levels of spatial and temporal detail. Tracking devices have and will become smaller, cheaper and more accessible, new satellite tracking technologies are introduced, data download methodologies become more efficient, battery life increases, numbers and variety of sensors on tracking tags increase, and all this leads to more data being collected at even higher spatial and temporal resolutions. Thus, movement ecology transformed itself from its data-poor beginnings into a data-rich discipline, allowing to find new answers to the burning research questions in animal ecology. Additionally, due to the miniaturisation of devices, more and more species can be tracked, such as birds, small mammals or even insects, opening new possibilities for quantitative ecological investigation of species hitherto considered too small [7-11].

In addition, the diversity and integration of different sensors allow the focus of data collection to move from the observer to the observed individual. That is, by collecting information from various sensors (e.g. body temperature, heart rate, acceleration) and incorporating environmental information into movement analysis, it is now becoming possible to reconstruct an animal’s perception of the world. We can find out where the animal was, its activities in various places and at various times, how these places looked and felt like and how they might have impacted the behaviour. Such observations could eventually lead towards the animal becoming the sensor informing us about its environment [9,12,13].

The basis of all such investigations is positional information through time, which is currently mostly collected using some type of animal-borne GPS tracking device. Sometimes, data collection is also complemented with the conventional very high frequency (VHF) or satellite systems data (Argos system and ICARUS system- International Cooperation for Animal Research Using Space), but GPS data are becoming increasingly prevalent as locational information in movement studies [7,8,14].

Trajectory data, defined here as a discrete time series of measured locations, are collected at detailed temporal resolutions and on particular temporal schedules. Depending on sampling frequency and schedules, such data volume can be very large (long, densely sampled trajectories). Additional complexity is introduced with simultaneous collection of related information either directly from other sensors or derived from environmental data [5].

There are many challenges with trajectories, the most basic and urgent being to visualise and explore such data. New analytical and visualisation methods are necessary for this purpose [1,2]. While there has been little cross-disciplinary exchange so far, we believe that there is a significant potential in interdisciplinary connections between movement ecology and information sciences that analyse movement data. Such connections would facilitate and enhance the necessary new methodological developments to mutual benefit. They would provide information scientists with an opportunity to explore real problems and get access to real data, while movement ecologists would get support for challenging data issues from researchers who specialise in spatio-temporal data analysis and visualisation. New methodologies from such collaborations would be based on both data expertise and ecological domain knowledge, thus likely outperforming mono-disciplinary methods.

Trajectory data are commonly collected in many other disciplines where movements of objects are being observed (e.g. vehicle, vessel or plane trajectories for transportation, human trajectories in time geography, pedestrian trajectories for urban planning). Further, a set of disciplines across information sciences (geographic information science (GIScience), computational geometry, visualisation, visual analytics) specialises in analysis and visualisation of spatio-temporal data on movement, including trajectories [15]. Each of these disciplines has their own approaches to trajectory analysis and visualisation, but the underlying concepts are the same.

As an example of what can be achieved in such interdisciplinary collaborations, this paper presents a review of movement studies from the European COST Action IC0903 “Knowledge Discovery from Moving Objects (MOVE)” (http://www.move-cost.info/). A COST action is an international research network bringing together researchers from across Europe to collaborate on a common topic. The main objective of the MOVE action was to facilitate collaborations between researchers in disparate disciplines interested in movement, thus establishing a network of ICT researchers and domain specialists to enable the development of novel methods for movement analysis and visualisation. Researchers from various subdomains in computer and geographic information sciences joined domain specialists from a broad range of disciplines that collect movement data. The network was active in the period 11/2009 to 10/2013 and consisted of close to 140 individual researchers in 24 European countries. The network generated a wide range of activities including 6 network conferences; 13 workshops, including one in the Lorentz workshop series [16] and two in the Dagstuhl seminar series in computer science [17,18]; 5 PhD training schools; 7 data challenges; and 53 Short-Term Scientific Missions (short visits). The main activity was the formation of collaborative teams between researchers in information and communication technologies (ICT) and domain scientists, out of which movement ecologists were the most prominent group. These collaborations were active both formally (through funded joint research projects or PhD student co-supervision) and informally (through joint experiments and paper authoring) and are continuing after the end of the action.

On the ICT side, MOVE has inspired many novel methodological developments through the exposure of ICT researchers to real data and domain knowledge; [19] provides an example documenting this process of interdisciplinary collaboration. The aim of this review paper is to serve as the knowledge transfer vehicle into the opposite direction. By providing an overview of MOVE achievements and their potential relevance to movement ecology, we hope to contribute to increased recognition in the movement ecology community of the potential that collaborations with ICT researchers could bring.

This paper presents an overview of methods for analysis and visualisation of trajectory data that were developed in MOVE and were either 1) specifically aimed for movement ecology or 2) were not specifically developed for movement ecology, but have a potential to be used in this context, as they address similar topics in other application areas. We further list some related work from information sciences but outside MOVE which may be of interest to ecologists. To facilitate the interdisciplinary knowledge transfer, we put these studies in the context of four ecological themes that we were able to identify in MOVE collaborations. These themes are:
  • Theme 1: Spatio-temporal dynamics of home ranges and utilisation distribution

  • Theme 2: Identification of spatio-temporal patterns in movement

  • Theme 3: Classification or identification of behaviour from movement data

  • Theme 4: Linking movement data with environmental context

In the next section we provide a short introduction to each of these themes and discuss data analysis challenges inherent to each theme. In the second part of this review we then look at methods developed in MOVE and how each of these addresses one or more of the four ecological themes.

ECOLOGICAL THEMES related to research in MOVE

T1: Spatio-temporal dynamics of home ranges and utilisation distribution

Utilisation distribution, the probability of encountering an animal in a given location given the available locational data, is a formal way to quantify or represent animal home ranges [20-24], with the idea to identify areas that provide vital resources (food, protection, nest or bedding sites, support for mating encounters or group living), thus allowing maximisation of individual fitness. In practice, home range is still often derived as a certain probability contour of the utilisation distribution that represents the proportion of time spent by animals within this contour [25]. With the increased availability of detailed, and thus highly spatially and temporally autocorrelated data, the methodological limitations of most widely used methods of utilisation distribution quantification are becoming more and more apparent. For example, many methods are sensitive to sampling frequency, where with very high sampling the contours of the utilisation distribution hug the data increasingly tighter and therefore home range shrinks to the area of the measurement error around the trajectory. This means that current utilisation distribution methods may not properly represent the home range concept, as data are not invariant to sampling method and frequency and therefore violate the requirement of the statistical independence of the observations.

In this section we focus on a selection of methods for estimation of the utilisation distribution directly relevant to MOVE. Other reviews include [24,26-29]. A common approach is to employ kernel density estimators, which place a decay probability function on each observed location and sum these up into a surface [30]. The choice of density parameters (kernel function types, bandwidth size) is widely debated [25,26,31-33]. As in many statistical methods, there is a trade-off between bias and variance that needs to be taken into consideration with KDEs. While a higher level of smoothing increases precision, it also increases the bias. However as we collect data at increasingly higher sampling frequencies, the variance decreases – this means that less smoothing is necessary and bias is reduced [33].

One of the problems with standard kernel density estimators is that they rarely consider the temporal dimension and sequentiality of points in a trajectory. Linking spatial variation to fluctuations in size and distribution over time and between populations is sometimes done by determining space use separately in each temporal period (in different seasons, months or years) [34-36]. Alternatively, time can be included in definition of kernels, either in the calculation of the kernel [37] or by extending the kernels to cover trajectory segments between two consecutive points rather than individual points [38]. Among the latter the Brownian bridge kernels incorporate uncertainty in movement between two consecutive locations in the definition of the kernel [39-42]. Other approaches incorporate movement behaviour (such as periodicity of visits), landscape properties or memory into home range estimators [43,44].

Most of the existing temporal approaches to home range estimation in ecology are based on 2-D statistical tradition. Developments in MOVE however have addressed the problem of the temporal dynamics of space use from the perspective of analysis and visualisation of multidimensional spatio-temporal data resulting in conceptually new approaches discussed in the review section.

T2: Identification of spatio-temporal patterns in movement

Movement of an individual organism is an interplay of four mechanistic components: its internal state, its motion capacity, its navigation capacity and the external factors [1]. The dynamic interaction of these components at various spatio-temporal scales is reflected in spatio-temporal patterns in movement data, which is why ecologists are particularly interested in identifying these patterns. We are searching for patterns from a variety of perspectives: within individuals or groups over time, between individuals, between groups, between populations, between species. In some cases, similarity in movement patterns is of interest, but so is identification of differences and relationships of these to both geographic space and time. In the following we present (non-exhaustive) lists of spatio-temporal pattern types of interest as well as of methods developed in movement ecology to identify these patterns in movement data.

One particular type of patterns is related to routines. These patterns are usually linked to temporal development of migration behaviour. Migration is a regular, seasonal pattern of movement that is strongly directional and seasonally reversible [45] and an obvious challenge is how to identify such routines or regular returns [46,47]. An unsolved problem in the study of migration is how learning affects migration journeys and migration ranges and how the range and/or route fidelity develop over time with an individual’s progressing age [48]. Route fidelity is a focus of many studies and can be investigated between individuals (birds flying in pairs), between the same individual at different times (consistency of the migration journey or route across seasons), or between individuals of co-existing species [49-52].

A related important question addresses navigation in migration [48]. How do orientation and long-distance navigation mechanisms influence the geometry of migration routes? What is the relationship between these mechanisms and migratory decisions (when to go, where to turn, what route to take)? Do the same navigational decisions occur at the same time and location in every migratory circle? What is the consistency of these decisions across individuals, groups and species? How do new migration routes evolve? Many studies are now exploring patterns in trajectory data in an attempt to answer these questions [48,53].

Another pattern type describes dynamic interaction, which is the inter-dependency of the movements of two or more individuals and is sometimes also called association, correlation or relative motion between two objects [54]. It can be investigated between individuals or groups of the same species, to see identify the frequency of individual encounters and patterns of avoidance, attraction, grouping or following [52,55]. Alternatively, patterns of interaction between co-occurring species can be of interest [51,56].

Identifying patterns from trajectory data requires a diverse set of methodologies. Temporal variability in movement can be explored through comparing long-term vs. short-term patterns, looking at seasonal patterns or presence of periodicity of varied lengths [47,51,55-59]. Route fidelity calls for geometric similarity analysis [50,52,53]. Migratory behaviour can be investigated through segmentation of trajectories at various spatio-temporal scales [46,47,53], where cross-scale analysis is of particular importance [60]. Interaction patterns can be identified using geometric approaches [61-63].

MOVE’s contribution to this theme is a series of alternative methods for spatio-temporal pattern identification. Computational geometry developed methods for median trajectories, segmentation, geometric similarity of trajectories and quantification of dynamic interaction. Spatio-temporal and attribute similarity of trajectories has been explored by GIScience through development of new data mining methods, such as geometric clustering, spatio-temporal clustering and clustering based on derived parameters of movement. Several contributions have also been made to cross-scale analysis.

T3: Classification or identification of behaviour from movement data

Animal movement is linked to behavioural responses [64,65], so that specific behaviours correspond to different movement types. For example, foraging, escaping predators, sitting in the nest, soaring in search of prey, all intuitively correspond to different movement patterns. Two recent technological developments support new ways of analysing behaviour beyond traditional methodologies (direct observation). First, the ever increasing availability of movement data provides the opportunity to infer behaviour from movement types [5]. Second, behaviour can be remotely monitored through a variety of sensors [66,67]. The advantages of these two approaches with respect to direct observations are twofold: they limit the interference of the observer, and exponentially increase the range of analysis.

The challenge is how to identify different types of behaviour from movement data. Behaviour types are often extracted from trajectories with various forms of statistical modelling, including state-space models, various types of random walk models and behavioural change point analysis [68-71]. Alternatively, data mining techniques, such as clustering are used for this purpose [72,73]. Or movement-derived parameters, such as speed are used to classify behaviour types [74].

Another promising perspective is the simultaneous recording of movement and information derived from other sensors, especially accelerometers. Accelerometers measure changes in velocity over time in three dimensions at very high temporal resolutions (10 Hz). These data can be used to identify two types of patterns: first, it is possible to identify changes in body posture and behaviour and second, the variation in measurements has been linked to speed and energy expenditure [67,75,76].

When accelerometers are used in combination with a GPS tracking device, acceleration data can be used to segment bird trajectories into behaviour classes including flying, foraging, body care, standing and sitting [76,77]. Accelerometer data can also be linked to GPS and magnetometer data [78] or alternatively in combination with a gyroscope, which measures the orientation and aids the accelerometer in high-frequency motion situations [79].

In many of these cases, video observations are collected simultaneously with locational and/or accelerometer data. Video footage serves as ground truthing for behaviour types, automatically derived through data mining [76,78,79]. Behaviour can also be identified directly from video-tracked data [80] or movement parameters can be derived from 3-D trajectories derived from video using computer vision: an example is data mining of movement parameters on 3-D trajectories of zebrafish movement [81].

MOVE studies utilise methods from computational geometry, spatial data mining and visualisation/visual analytics to support behaviour identification from movement data, as described in the second part of this review.

T4: Linking movement data with environmental context

The movement of an organism is affected by the internal state of the organism and by the external factors including environmental context of the individual’s location [1]. External factors affect the movement in many ways. They can trigger behavioural patterns or migratory decisions. Animals may decide to move at times with conditions supportive for a particular movement type while allowing them to optimise energy expenditure. Movement is therefore often linked to spatial and temporal variability in environmental conditions [82].

To investigate the influence of environment on movement, tracking data can be complemented by environmental data from many sources and of many types. Some studies incorporate remotely sensed satellite data with trajectories [82-85], others link trajectories to either meteorological information such as wind direction and speed [86] or to weather radar data [87,88]. For interested readers, there is an on-going COST Action ES1305 on this topic: “European Network for the Radar surveillance of Animal Movement (ENRAM)”, 2013–2017. For land animals, weather effects (snow) and topographic factors (slope) can be linked to movement [47]. Trajectories can be linked to field data on home range productivity and related indices derived from remotely sensed data, e.g. the normalised difference vegetation index (NDVI) [89]. For marine mammals (whales, dolphins, seals), passive and active acoustic monitoring is used in combination with trajectories [90].

An alternative to external environmental data is to include more than one sensor on an animal tag. This is particularly common for marine animals, where tags are traditionally referred to as bio-loggers and incorporate both locational and environmental sensors [5,8,66,91]. Frequently used are oceanographic sensors, such as the conductivity, temperature and depth (CTD) loggers and specific sensors for salinity, turbidity, fluorescence, level of chlorophyll, presence of cyanobacteria and other oceanographic parameters [92-94]. Locational sensors for marine species are often a combination of GPS tracking devices and various marine Satellite Relay Data Loggers (SRDL), which measure location, speed and depth of the diving animal [92,95]. Researchers have used data from combined oceanographic and locational sensors to model not only animal movement, but also the state of the oceans in remote areas inaccessible for human observers, but which animals (e.g. polar seals) visit regularly and periodically [93,94].

In many studies, movement data are linked to environmental or other sensor data through trajectory annotation. This is a process that semantically enriches trajectories with environmental and sensor information at each location and time [86,96]. Due to large data volumes of both tracking and environmental or sensor data, this has to be done automatically and systems are being developed to support this procedure. An example is the Environmental Data Automated Track Annotation System (EnvDATA) that allows annotation of trajectories from the animal movement online data repository Movebank (www.movebank.org, [10]) with satellite remotely sensed information [82]. Another example is the spatial database of environmental data linked to the species-distribution range in the Eurodeer project (www.eurodeer.org, [47]).

Is semantic trajectory annotation the best way to connect movement and environmental data? The problem is that these two types of data are collected at different spatial and temporal scales. Animal movement can be collected with 1 Hz resolution at times with accurate GPS locational measurements down to sub meter accuracy, while satellite data for a particular location may only be available from half-daily or daily satellite passes and collected at spatial scales of several tens or hundreds of meters [82]. Data pre-processing measures (e.g. spatio-temporal interpolation or aggregation) are therefore required prior to semantic trajectory annotation. This process may propagate the uncertainty related to coarser spatio-temporal resolution of environmental data into higher-resolution trajectory analysis. A question is therefore how to capture and describe or eliminate the uncertainty resulting from matching the spatially and/or temporally misaligned data. This is an area where MOVE contributed with cross-scale analysis for context-aware trajectory analysis.

Review

This section presents methodologies for movement analysis and visualisation developed in MOVE grouped in the following categories:
  1. 1.

    Geometrical analysis of trajectories

     
  2. 2.

    Similarity and clustering

     
  3. 3.

    Visualisation and visual analytics

     

For each of these categories we describe methods developed in MOVE and link them to the ecological themes from the first part of this review.

Category 1: Geometrical analysis of trajectories

Geometrical analysis of spatial data is developed in a number of information sciences in which methods are based on geometry and location of data entities in space. In terms of trajectories, geometrical methods are useful for a number of problems relating to the form and relative positions of trajectories in the 3-D (or 4-D) physical framework space of 2-D (or 3-D) position and time. In MOVE, geometrical analysis was represented by computational geometry [97] and GIScience [98] and the studies can be grouped into the following topics:
  • trajectory segmentation,

  • identifying a representative path from a set of trajectories,

  • scale-dependent geometric analysis and

  • identification of spatio-temporal patterns.

These methods may be useful for themes T2 and T3 (Table 1).
Table 1

MOVE studies in Geometrical analysis, categorised per method type vs. ecological themes (T) they address

 

Sources relevant for ecological themes

1. Geometrical analysis

T2

T3

Trajectory segmentation

104, 105

99-104

Identifying a representative path

106-110

 

Scale-dependent geometric analysis

113

114, 115

Identification of spatio-temporal patterns

116, 117

 

T2: Identification of spatio-temporal patterns in movement and T3: Classification or identification of behaviour from movement data.

Trajectory segmentation

Geometric trajectory segmentation refers to the problem of splitting a trajectory into pieces (referred to as segments) such that each piece fulfils a geometric criterion. If the geometric criteria characterise behaviours of a moving entity, this problem is closely linked to classification of behaviour (T3). Buchin et al. [99] developed a segmentation method for animal trajectories based on individual movement states of the moving object (animal). They look at different types of bird movement (flying, foraging and resting) and by linking these to different types of geometrical properties of trajectories (location, speed, angular range, heading, time, etc.) they developed a method that automatically segments bird trajectories into segments that correspond to these states (Figure 1). The results of such an algorithm can then be used for further exploratory ecological analysis of birds’ movement.
Figure 1

(after [ 103 ] ). A geometric segmentation of a trajectory. Red/pink segments are migration flight, yellow segments are stopovers (one stopover is shown in more detail in the lower right corner). Blue markers indicate end of a stopover. Stopovers are described by staying within a bounded activity radius for at least 2 days. This is a conjunction of a monotone-decreasing criterion (if the sequence (B, C,…, G) stays in a disk of small radius then so does every subsequence, e.g. (D, E, F)) and a monotone-increasing criterion (if the sequence (B,…, G) corresponds to a duration of at least two days then so does every sequence which includes it, e.g., (A, B,…,G, H)).

Geometric segmentations are often optimised based on a set of general spatio-temporal criteria. This means that the methods aim to minimise the number of segments while guaranteeing that each segment fulfils one of the criteria [100]. Various approaches are concerned with the scale at which the criteria are satisfied, for example, if they are satisfied on one particular segment and all its sub-segments or all larger segments that include this one particular segment. An efficient framework for geometric segmentation was proposed by [100] for criteria that are monotonically decreasing, i.e. if they are fulfilled on a segment, then they also are on its sub-segments. This framework was extended [101] to include combinations of monotonically decreasing and increasing criteria (as in Figure 1). If segments are not restricted to start at points, the general segmentation problem becomes computationally intractable [102], but can still be solved efficiently for monotonically decreasing criteria [100].

Geometric segmentation can also be linked to statistical analysis, which is an approach that may be more familiar to ecologists. Alewijnse et al. [103] propose a model-based approach to segmentation of movement data. In this approach, a segment is defined by a uniform model parameter and an information criterion is used to select the number of segments. This approach assumes little knowledge on geometric characteristics of the input trajectory data, yet it identifies the optimal segmentation by optimising the information criterion, linked to the complexity of the movement model. This particular approach uses Brownian bridges, but can be generalised to any parameterised movement model.

Segmentation methods were also developed in MOVE for other types of trajectories. Sester et al. [104] present a method to link segments to human movement behaviour. Their segmentation is based on identifying important places from trajectories (most frequently visited places for the longest time). Segments between these places are classified based on movement parameters linked to travel mode (walk, bus). This approach could serve to identify equivalent patterns in animal movement, for example, important stopover places of longest duration in data on annual migration (theme T2).

Panagiotakis et al. [105] present a method for segmenting trajectories into representative and non-representative segments based on other nearest trajectories. They use vehicle trajectories for their experiments. As their data vary according to spatial and temporal density in location sampling, this could be relevant to animal trajectories obtained with irregular sampling schemes.

Identifying a representative path from a set of trajectories

Another frequent geometric problem is how to identify a representative path for a set of given similar trajectories. Is there an optimal route that can be used to represent this set? How can this route or path be defined, while the path may or may not be one of the actual trajectories? This could be useful in ecological terms for theme T2: identification of routines and consistency in migration [49,50].

In MOVE, Buchin et al. [106] introduced a computational geometry approach to compose such a representative route from parts of the actual trajectories. Here the trajectories do not need to be temporally correlated, but just need to follow a similar spatial route. They call their representative route a median trajectory and build it from pieces of the trajectories in the data set (Figure 2).
Figure 2

(after [ 106 ]). a) three trajectories with a common start and end point and b) a median trajectory (bold) representing these three trajectories. The median trajectory is built of segments of the original trajectories and switches the original underlying trajectory at each intersection.

Identifying a representative route from a set of trajectories is a common topic in GIScience and there are many approaches outside MOVE. Brudson [107] uses principal curves to identify the most probable route from a set of GPS pedestrian trajectories. Similar methods are widely used in navigation and even developed for reconstruction of representative 3-D trajectories: [108] reconstruct 3-D bicycle tracks from GPS trajectories – a method that could be of interest for movement of animals freely moving in 3-D (birds, sea mammals). In MOVE, Etienne et al. [109] developed a method to identify the main naval route from a set of vessel trajectories, sampled at equal times. Pelekis et al. [110] take an alternative approach and consider the uncertainty in trajectory measurements by constructing a fuzzy vector representation of each trajectory. They use this representation to construct a so-called centroid trajectory as the representative path based on density of trajectory points at each moment in time.

Note that purely geometrical methods of identifying a representative trajectory have certain limitations when considering animal data. For example, a median trajectory provides a population-average summary of paths, however, its characteristics may not match up with the movement of any particular individual. The difference between the average model of movement vs. individual models is a well-known issue in ecology [111], which is analogous to the problem of global vs. local modelling in spatial statistics. We discuss this similarity in concepts in ecology and GIScience as part of one of the future challenges.

Scale-dependent geometric analysis

Movement characteristics are influenced by processes operating at different spatio-temporal scales [1]. There is therefore a need to support movement analysis across different scales and investigate how these multiple-scale processes act together. On this topic, movement ecology could benefit from the dependency of geographic phenomena on spatio-temporal scale, which is one of the most well-known and longest-standing topics in GIScience [112].

In MOVE, a number of GIScience studies explored the issue of scale. Laube and Purves [113] investigate how temporal scale affects the calculation of movement parameters (speed, sinuosity and turning angle) of animal trajectories. They demonstrate what they call the “granularity grief”: the fact that derivation of any kind of movement parameter from trajectory data is influenced by the temporal sampling rate and thus scale-dependent. They further demonstrate the relationship between uncertainty in individual GPS measurements at different scales and how these affect the fine-scale movement descriptions. This could be of interest to theme T2.

In a recent study on cross-scale movement analysis [114] show how the derivation of different movement parameters over a range of spatial and temporal scales significantly improves the subsequent classification of movement behaviour from a set of zebrafish trajectories, compared to a single-scale approach. This could be relevant to theme T3. Soleymani et al. [115] use cross-scale extraction of movement parameters and context information as input features to detect foraging behaviour in GPS trajectories of wading birds. Their results suggest that it is possible to classify, with high accuracy, fine-grained behaviours based on high-resolution GPS data, providing an opportunity to build a prediction model in cases where no additional sensor (e.g. accelerometer) or observational data is available.

Identification of spatio-temporal patterns

This sub-topic is linked to theme T2. Predating MOVE, a study [116] provides a general framework of spatio-temporal movement patterns that can be identified from trajectories. In MOVE, Orellana et al. [117] propose a method to identify suspension patterns in movement, which represent an attraction or an obstruction for the moving object.

Category 2: Similarity and clustering

A frequent task in trajectory analysis is to partition the data into groups of similar trajectories. In data mining, clustering takes a set of data objects and partitions these into groups (clusters) so that the objects in the same group are more similar to each other than to objects in other groups [118]. The procedure consists of two steps: first, a similarity measure has to be defined based on the data domain and second, a grouping procedure is used to partition data into clusters based on similarity between data objects. In this section we focus on similarity measures for trajectories – for reviews of clustering methods see [118,119].

We group trajectory similarity methods from MOVE based on what part of trajectory data space they consider. Trajectories are spatio-temporal data and their data space can therefore be partitioned into three separate sub-spaces: spatial part (location), temporal part (time) and attribute part (derived movement parameters or other). The following trajectory similarity measures were developed in MOVE:
  • geometrical similarity, based on location and time only,

  • similarity based on physical attributes of movement (speed, acceleration, direction, etc.) and

  • context-aware similarity based on a combination of attributes.

For ecological themes, similarity and clustering of trajectories can support identification of particular spatio-temporal patterns in movement, that may also be related to behaviour, thus supporting themes T2 and T3 (Table 2).
Table 2

MOVE studies in Similarity and clustering, categorised per method type vs. ecological themes (T) they address

 

Sources relevant for ecological themes

2. Similarity and clustering

T2

T3

T4

Geometrical grouping and similarity

121-128

  

Similarity based on physical attributes of movement

123, 129–132, 135, 136

114, 130, 133, 134

 

Context-aware similarity

  

137, 138

T2: Identification of spatio-temporal patterns in movement, T3: Classification or identification of behaviour from movement data and T4: Linking movement data with environmental context.

Geometrical grouping and similarity

Geometrical similarity methods rely on the notion of distance between trajectory points, which is usually a spatial distance. Originally such measures were developed to compare shapes of polygonal lines and only considered locations of points. Traditional geometrical similarity measures include Euclidean distance between each pair of points, Hausdorff distance that identifies the largest distance from a point on one trajectory to the closest point on the other trajectory, and Fréchet distance that takes into account the location and ordering of trajectory points and is sometimes also called the dog-walking distance (it represents the minimum length of a leash between two objects, i.e. a person and its dog, that move along respective trajectories without backtracking). More recently the temporal aspects of movement are also considered [120].

In MOVE, Buchin et al. [121] compare data points at equal times, whereas [122] compare data points with a bounded local time shift. Pelekis et al. [123] develop a similarity measure based on the area between two trajectories. Merki and Laube [124] define a set of grouping patterns (pursuit, escape, avoidance and confrontation) and present algorithms for their detection. Outside of MOVE, Rinzivillo et al. [125] use several geometric similarities in their clustering. First they consider only the distance between the start and end points of each trajectory; then they use distances between a selected number of sampled points on each trajectory and finally the smallest distance between two trajectories at a certain time.

Several of these methods have been used for clustering of sub-trajectories to identify entities moving in a group over some period of time [120,126]. Groups occur when a large set of moving entities moves sufficiently close for a sufficiently long time and may split or merge with other groups. This may be useful for identification of dynamic interaction patterns (T2). In MOVE, Buchin et al. [127] propose a representation of how such groups evolve over time (Figure 3). Outside of and pre-dating MOVE, methods that identify similar sub-trajectories are often based on either dynamic time warping or identification of the longest common subsequence [128].
Figure 3

(after [ 127 ]). Progression of a set of trajectories through time (represented as horizontal axis and progression is from left to right). Colours indicate groupings of various sizes based on location and proximity of moving objects at that moment. A beige group has three objects, a yellow one four, an orange one five, a grey one six. At each moment in time the grouping is maximal.

Similarity based on physical attributes of movement

A number of recent studies in GIScience approaches trajectory similarity by looking at physical properties of movement, which include physical descriptors of movement (speed, direction, acceleration, turning angle, angular speed) and path shape properties (curvature, sinuosity, tortuosity). These quantities are either measured by the tracking device or derived from trajectories and are referred to as movement parameters [116,129]. In terms of movement ecology, clustering based on these parameters can be used for inference about movement behaviour [130] and thus contribute to theme T3. In the following we describe MOVE studies that use movement parameters to define trajectory similarity.

Pelekis et al. [123] introduce four types of trajectory similarity, out of which two are geometrical (spatio-temporal similarity and spatial-only similarity) and two based on movement parameters (speed/acceleration-based similarity and directional similarity).

Dodge et al. [129] define the term “movement parameter” and use various individual parameters (velocity, acceleration, turning angle, displacement, straightness index) to build temporal movement parameter profiles. These profiles are used to decompose trajectories into segments of homogeneous movement. Dodge et al. [131] use the conceptual space of movement parameters (MP space) to compare two or more trajectories and define their similarity based on the temporal progression of the respective trajectories in the MP space. They use speed, azimuth, turning angle and acceleration to identify groups of concurrent and coincident trajectories. Concurrence is defined as similar progression through MP space and coincidence as similar progression through 3-D space, a space-time cube (STC, see visualisation section). The study is performed on a well-known trajectory data set of hurricanes in the Atlantic. Dodge et al. [132] extend their 2009 segmentation method with an alternative similarity measure based on producing a string of symbols for each temporal profile in the MP space and using a modified string edit distance metric for trajectory clustering. Soleymani et al. [114] use the MP space parameters for cross-scale spatio-temporal identification of different animal behaviours.

McArdle et al. [133,134] combine physical properties of movement in the space-time cube with time series clustering methods to classify a set of pedestrian tracking trajectories into several behaviour types. They decompose the 3-D space-time cube into two 2-D projections of time vs. one of the two geographical coordinate axes and then compare similarities of trajectories in each (or both) of these projected spaces based on their shape as mathematical curves.

Çöltekin et al. [135] use a set of eye tracking specific movement parameters for eye- and mouse-tracking trajectories obtained during a task of visual search on a computer display to investigate hand-eye interaction. In Human-Computer Interaction (HCI), eye tracking is a way to evaluate the usability of visual interfaces, where trajectories of gaze on the screen are collected using an eye tracking device. Trajectories of mouse movement on the screen are also collected for a similar purpose and this study investigates if there is a connection between the two trajectory types using trajectory analysis. Movement parameters used to evaluate similarity of the eye and mouse movement are distances from gaze or mouse to target and distance between gaze and mouse trajectory (Figure 4). Such studies, while using trajectory data from an unrelated domain (HCI), could be relevant for analysis of dynamic interaction patterns in theme T2, as data type (trajectories) and conceptual formulation of the problem (interaction of two moving objects, in this case gaze and mouse pointer on the screen) are the same as in the dynamic interaction problem.
Figure 4

(after [ 135 ]). Eye and mouse trajectories in a visual search task: the participant was asked to identify the target (green square) on a map (not shown) and click on it. a) Eye and mouse trajectories generated in this task. b) Time series plot of distances from eye & mouse to target vs. time. In both charts, eye is in red, mouse is in blue.

Outside of MOVE, Ranacher and Tzavella [136] provide a broader review of physical movement trajectory similarity measures in GIScience.

Context-aware similarity based on a combination of attributes

As discussed in theme T4, animal movement is inherently embedded in the environmental context. In MOVE, several developments integrated contextual information into similarity analysis contributing to theme T4.

Buchin et al. [137] present a method for integrating land cover information into similarity analysis. They extend geometric similarity (equal time distance, Hausdorff distance and Fréchet distance) with context distance, which for their hurricane trajectories consists of external and internal factors that influence hurricane movement. External factors include atmospheric conditions (temperature, air pressure), land use (land, sea) and topography of the region, while internal factors relate to properties of hurricanes themselves (intensification, wind speed, move speed, diameter). Their method is able to distinguish between hurricanes that have a similar spatio-temporal track, but different context.

In a more sophisticated attempt at incorporating context into similarity analysis [138] consider the temporal variation in the sequential use of environmental features (relevant to T2 and T4). Their objective is to explore spatio-temporal patterns in the sequential habitat use by animals and they propose a tree-based approach using sequence alignment method (SAM) [139]. Sequences are constructed from roe deer trajectories by linking location to four habitat use classes, defined from two geographical parameters: habitat type and elevation. SAM is used to cluster the sequences into dendrograms (Figure 5), where clusters of similarly-moving animals can be identified at different levels of detail. By linking additional covariates to their results, they explore the relationship between identified clusters and animal characteristics.
Figure 5

(after [ 138 ]). Tree classifying individuals based on spatio-temporal sequential habitat use during May-June. An extract of the habitat use sequences for the first five days (01/05 – 05/05) is shown. Covariates can be associated to each individual and help to identify relations between identified clusters of similar sequential habitat use and animal characteristics.

Category 3: Visualisation and visual analytics

Two final MOVE disciplines interested in movement are visualisation and visual analytics [140]. Vision is the most important sense in communication between humans and computers and visualisation plays an important part in cognitive processing. It supports data analysis in several ways: it provides an ability to portray and understand large amounts of data; it allows identification of patterns in the data that were not previously evident and thus supports hypothesis generation; the patterns are identifiable at large and small scales; and problems with the data can become quickly apparent [140]. Different visualisation communities include scientific visualisation, information visualisation and visual analytics. The first two portray spatial and non-spatial data respectively [140,141], while visual analytics combines human reasoning and pattern recognition ability with the computational capabilities of a computer to support a more efficient data analysis [142-144].

The increased recent availability of all types of movement data has kick-started visualisation and visual analytics developments for movement, resulting in a wide variety of methods and tools [145-147]. Many of these applications, while firmly anchored in visualisation or visual analytics, use animal movement data as inspiration [148-150] as did many of the participants in MOVE. We categorise MOVE contributions into the following three categories:
  • Spatio-temporal visualisations: space-time cube and other approaches

  • Visual aggregations: geometric aggregations and kernel densities

  • Visual analytics of movement

Many of these studies (Table 3) contribute to themes T2 and T4 by providing the ability to visually identify various previously unknown spatio-temporal patterns. Aggregations support theme T1 by visualising temporal dynamics of space use.
Table 3

MOVE studies in Visualisation and visual analytics, categorised per method type vs. ecological themes (T) they address

 

Sources relevant for ecological themes

3. Visualisation and visual analytics

T1

T2

T4

Spatio-temporal visualisations

 

133-135, 157, 158, 160, 161, 164-167

 

Visual aggregations

15, 62, 175-190

145, 168–173, 182, 188

 

Visual analytics, attribute visualisations, linked views

 

135, 146, 147, 158, 165, 166, 170, 173, 183, 193-198

193, 194

T1: Spatio-temporal dynamics of home ranges and utilisation distribution, T2: Identification of spatio-temporal patterns in movement and T4: Linking movement data with environmental context.

Spatio-temporal visualisations: Space-time cube and other approaches

The problem of representing time along with two spatial dimensions has a long tradition in GIScience [151,152]. In 1970s, a branch of geography called time geography established one of the most frequently used visual representations of interlinked geography and time: the space-time cube (STC) [153]. In an STC, the spatio-temporal data are shown in a 3-D space, where the bottom 2-D plane represents the 2-D geographic space and the third axis represents time. The main assumption of time geography is that geographic space and time are inseparable and the STC was developed to portray this assumption in a visual manner. Since then, the STC has become popular in GIScience for visualisation of human activity patterns [154-156]. In the visualisation community (and outside of GIScience), the STC popularity to show the temporal component of any type of spatio-temporal data (not just trajectories) has also recently increased [157].

In MOVE, the basic form of STC for trajectories (where trajectories are shown as polylines in the STC space) was used in several studies. McArdle et al. [133,134] superimposes the STC on a virtual globe (Google Earth), so that the third dimension consists of a sum of elevation and time, thus making it appropriate for locations with flat terrain. Of note is the linkage to Google Street View [134], which allows for visual ground-truthing of automatically-derived stopping points (e.g. one frequent stopping point turned out to be a shop, another a dentist’s office). This may be of interest to ecologists who try to understand stops in migration (relevant to T2). At present the availability of Street View limits this kind of exploration to specific countries, however, the coverage is likely to be extended in the future.

Çöltekin et al. [135,158] apply the STC concept to visualise interaction between gaze and mouse trajectories. Here the base 2-D plane represents the display of the stimulus on the computer screen and the third dimension represents time. Their approaches might be of interest to ecologists who are exploring the dynamic interaction between animals (relevant to T2), to which the interaction between the eye and mouse movement is analogous.

The inherent inseparability of space and time in the 3-D STC is difficult to achieve if only spatial or only temporal visualisations are used. However, as a 3-D display, it is complex to use and its usability needs to be empirically examined [159]. MOVE contributed to this through usability experiments in which the authors deconstructed the STC from the traditional cartographic point of view and made recommendations about the strengths and weaknesses of this popular 3-D display [160,161].

STC is only one of many temporal visualisations (see [162,163] for reviews). In MOVE, several collaborations between practitioners and visual developers resulted in alternative temporal displays, some specifically aimed at animal ecology. One study [164] developed bespoke spatio-temporal displays of bird migration patterns (relevant to T2). They focus on a specific set of migration-related questions, such as timing of annual migration, route fidelity and identification of stops. They also explore how these events relate to the time of the year (onset of spring) and how the spatio-temporal patterns vary between individuals and years. Another collaboration between visualisation experts and ecologists developed space-time visualisations to explore changes in biodiversity [165], using timeline and species density displays, relevant to visualising temporal dynamics of population distribution (T2).

In MOVE, Zhang et al. [166] present a timeline display developed for a set of identification, localisation and movement comparison tasks to study urban movement trajectories. Outside MOVE, Wang and Yuan [167] use a similar set of temporal visualisations, including a timeline, a straightness plot and others to investigate spatio-temporal patterns in urban movement.

Visual aggregations: geometric aggregations and kernel densities

When movement data sets are large, visual displays that show all trajectories become unsatisfactory, as the overprinting and clutter increase to the point that no patterns can be reliability identified anymore. In MOVE, Netzel et al. [168] have investigated how different line rendering styles can help improve the perception of dense visualisations of trajectories and do so using oystercatcher trajectories as a case study.

Even with improved trajectory rendering methods, the overplotting problem eventually becomes too severe for complex and large data sets. In such cases a frequently used concept is visual data aggregation. Aggregation refers to combining several data elements into a single unit that is then shown in some other way than the original data would be. This operation reduces the size of the data to be displayed, while at the same time there is inevitable information loss as patterns are generalised. Andrienko et al. [145] present an overview of aggregation methods for movement data and [157] discuss this concept in the context of an STC. Studies in MOVE developed two types of aggregations: geometric aggregations (relevant for T2) and aggregations using kernel density estimation (relevant for T1 and T2).

Geometric aggregations

A common approach to aggregate movement data for spatial visualisation is edge bundling, which merges nearby sub-paths into one. It is often used for origin–destination data (movement data where only start and end points are known). Methods for edge bundling are popular and include force-directed approaches [169,170], combinatorial techniques [171] and image-based methods [172]. In MOVE, Hurter et al. [172] demonstrate the usefulness of edge bundling for simplifying and aggregating various types of movement trajectories, in particular eye tracking data. Höferlin et al. [170] enrich the aggregation of groups of trajectories by applying abstracted schematic rendering that reflects the aggregation process. Another MOVE study [173] incorporates the edge-bundling principle into a time lens display of car movement.

Aggregations with kernel densities

As mentioned, the concepts of utilisation distribution and home range are often shown using kernel density estimators of trajectory points. These methods traditionally assume that the points form an independent sample taken from a static 2-D probability distribution of the individual’s locations and ignore the sequentiality of points. This is often addressed through sequential kernels where kernels are placed not over trajectory points but over trajectory segments and are added into a two-dimensional probability surface, as is normally done with point kernels. Note that for visualisation purposes we call a segment a line between two consecutive trajectory points. This is different from the segments in trajectory segmentation in geometric analysis, where a segment refers to a sub-trajectory, which may consist of any number of consecutive points. Figure 6 presents an overview of segment kernel approaches and lists relevant GIScience and ecological references, some from MOVE and some preceding MOVE.
Figure 6

(after [ 190 ]). Two-dimensional kernels for trajectories that produce two-dimensional density surfaces. The point-based kernels in panel a) do not consider the temporal dimension of trajectory points, but treat them as independent observations in a point data set. Panels b), c) and d) show line-segment kernels, where sequentiality of two consecutive trajectory points is taken into account in kernel construction.

In GIScience, segment kernels are often defined based on principles of time geography [153] and space-time probability prisms [174]. Considering movement in terms of space-time prisms results in elliptical kernels that define the area covered by all possible movement paths between two trajectory points. This is a popular GIScience approach to model animal movement [15,62,175,176]. In MOVE, researchers modelled vessel line density based on similar principles and convolution of density fields around two consecutive points [177,178]. Vessel line densities were linked into interactive systems with other geovisualisations [179,180]. Other researchers in MOVE took into account acceleration and velocity of movement through directional segment kernels [181].

A MOVE collaboration addressed the problem of low sampling rate between consecutive trajectory points using Brownian bridges [182]. If the temporal sampling rate is too low for the linear movement between two observed locations to make sense, a segment kernel that takes into account movement uncertainty is more appropriate to use. Buchin et al. [182] develop their approach to visualise uncertainty in movement as well as identify interaction patterns such as encounter, avoidance/attraction, regular visits, and following (Figure 7).
Figure 7

(after [ 182 ]). Two trajectories (blue/purple) with a potential encounter (red/green) computed based on the Brownian bridge movement model.

All of the kernel density models of utilisation distribution described so far, whether they incorporate time or sequentiality into the algorithm or not, are represented as surfaces in two geographical dimensions. MOVE researchers have combined the STC principle with a generalisation of the 2-D kernel density into three dimensions (Figure 8). A MOVE study used density of gaze points from eye tracking to enrich the STC visualisation of gaze points from a large number of eye tracking experiments by colour coding [183]. Outside of MOVE, a 3-D point kernel density in STC has been used in crime visualisation [184] and in spatial epidemiology [185]. Another GIScience study combined the 3-D point kernel density with the principle of space-time prisms in time geography to generate probabilistic space-time prisms for understanding the movements and activities of animals at fine temporal and spatial scales [186]. Also outside of MOVE, a recent ecological study [187] develops 3-D densities in real physical 3-D space (i.e. not the STC) taking into consideration elevation as well as the two geographical coordinates of location.
Figure 8

(after [ 190 ]). Three-dimensional kernels for trajectories that produce volumes in geo-time space. Panel a shows a) point space-time density with cylindrical kernels that do not take into account the temporal sequence of points in a trajectory. Panels b), c) and d) show polyline kernels, where there is one kernel for the entire trajectory (and not a separate kernel for each line segment). Distance from each voxel to trajectory in panel b) (shown in kernel with a dashed grey line) is calculated in 3-D, perpendicularly to the trajectory. Distance from each voxel to trajectory in panels c)-d) is measured at a constant moment in time (i.e. horizontally) and is calculated as 2-D distance. Panel d) shows the Brownian bridges version of the stacked 3-D kernel, where the width of the kernel at each moment in time depends on the position on the trajectory between each two consecutive points.

Generalising the concept of point density into a model that considers sequentiality of movement, one MOVE study developed the space-time density of polylines [188]. Here, the kernels are not calculated for each point nor each segment, but span the entire trajectory, which is represented as a polyline in the STC. The kernel around each trajectory is built within a volumetric union of rounded-cylinders, one cylinder for each trajectory segment. A further MOVE collaboration produced a faster optimised version of the 3-D polyline density, the stacked space-time density, and linked it to the concept of home range dynamics. This study also provides several alternative kernels, including a three-dimensional Brownian bridge kernel [189,190]. Figure 9 shows an example of the stacked space-time density for a month of daily trajectories of one lesser black-backed gull [190]. As volumes, such 3-D densities can be displayed in various ways, either using direct volume rendering or isosurfaces. They allow for identification of spatio-temporal patterns that are otherwise undistinguishable from spatial-only patterns in a standard 2-D density surface.
Figure 9

(after [ 190 ]. Stacked space-time density of animal trajectories. a) Space-time cube representation of one month of trajectories of one individual bird. The x-y plane represents geographic space and the z-axis is time (0-24 hrs). b) Brownian stacked space-time density of the trajectories from the space-time cube. c) Gaussian stacked space-time density of the same data and d) isosurface of the highest values in the Gaussian density, indicating a temporal column and a space-time hotspot.

An HCI study [158] investigates the level of interaction between the eye and mouse trajectories using the 3-D stacked space-time density [190] combined with 3-D volumetric change detection methods to quantify the level of interaction between eye and mouse trajectories. In terms of movement ecology, this may be relevant to dynamic interaction (T2).

Visual analytics of movement

A common design approach for visual analytics systems for highly-dimensional complex data is to use linked views, that is a set of interactively connected visualisations, each of which provides a different perspective on the data [191]. This methodology takes a set of data displays (each showing a selection of the given dimensions in some particular way) and then allows the interaction in one view (e.g. selection, zoom in, zoom out, etc.) to simultaneously modify displays in all views [192]. This is relevant to visual exploration of spatio-temporal data, since it enables generating a unique spatial, temporal or spatio-temporal perspective on a pattern appearing in a linked attribute-only view [144].

Many MOVE studies mentioned above employ the linked views paradigm (e.g. [165,166,183]). Others include not only spatio-temporal displays, but also attribute visualisations. Tominski et al. [173] introduce a trajectory wall. Here, car trajectories are represented as ribbons in a 3-D space, where the bottom 2-D plane represents the geographic space and the third axis the vehicle count. Ribbons are stacked over their geographic path and their segments coloured according to an attribute (speed). This view is interactively linked to a time lens, where clusters of temporally similar trajectories are shown using edge-bundling. The system allows identification of temporal patterns in car movement and in particular anomalies in regular flow, such as traffic jams.

Andrienko et al. [193,194] present linked views for a comprehensive visual exploration of any type of movement trajectories. These systems include STCs, a number of attribute visualisations, density maps, temporal visualisations and a number of other displays that allow incorporation of contextual information.

Another trajectory type widely represented in MOVE are eye movement trajectories, generated in HCI studies of visual displays. A number of MOVE studies used visual analytics for exploration of eye trajectories from such experiments. Andrienko et al. [195] introduce a comprehensive visual analytics methodology for exploring eye movement – their system is based on their previous work and visualisations [144,145,193]. Ooms et al. [196] use a combination of selection, simplification and aggregation operations to visualise and analyse patterns in eye tracking data. Kurzhals and Weiskopf [183] and [197] use a system with multiple linked views including a density-based STC representation for a set of gaze trajectories collected in an experiment with dynamic visual stimuli (videos, Figure 10). Finally, as described above, [158] and [135] use the STC and stacked space-time densities for concurrent visualisation of eye and mouse trajectories in an attempt to quantify the interaction between the eye and mouse.
Figure 10

Eye tracking data shown in the visual analytics tool ISeeCube [ 183 , 197 ] with multiple linked views. The data shows gaze information from multiple participants watching the same video [189]. The visual workspace is separated into several regions that can be freely adjusted by the user. Region a) displays the scanpaths of selected participants in an STC (coloured lines), along with clustered gaze point data (coloured regions on the two grey walls). The STC also contains a snapshot of the video at a time frame that can be adjusted by the user. At the top-left of region a), the same video replay is shown with two areas of interest marked by blue boxes (person and kite). Region b) provides a hierarchical clustering of the trajectories according to the similarity of their distribution of attention to areas of interest. Region c) shows the detailed information of one of the areas of interest (kite), including overall distribution of attention, as well as size and position of the area over time.

The visual interfaces of visual analytics systems often include the aforementioned multiple linked views to provide a comprehensive visual representation of complex data. Another component that makes visual analytics particularly interesting for complex data is its incorporation of (semi)-automatic analysis methods. Here we can make direct use of the new MOVE trajectory analysis methods reviewed earlier in this paper. In particular similarity measures and clustering methods are useful because they allow us to group, aggregate and simplify large data sets. One example of MOVE research on this topic is a study that employs trajectory clustering to group the trajectories and then allow for interactive selection of subgroups and re-clustering [170]. The process allows for top-down exploration of the data set, repeatedly selecting one or a few clusters and re-clustering remaining trajectories. Clustering supports faceted exploration that allows to cluster trajectories according to a variety of similarity measures between trajectories (e.g. coverage, distance between means, distance between standard deviations) and facets (related to geometric information from the trajectories, e.g. position, velocity, direction of motion, time, and object class). Another example from MOVE is the analysis of gaze trajectories with the system by [183] and [197,198]. They support spatio-temporal clustering of gaze points on trajectories, as well as hierarchical clustering of the sequences of gaze trajectories based on some distance metric. Many of these metrics are based on string metrics such as the Levenshtein distance [199], which measures the difference between two words (sequences of characters) based on how many single-character edits (insertions, deletions, substitutions) are needed to convert one sequence into another. In [197] the Levenshtein distance is used to compare and cluster gaze trajectories represented as a string of subsequently viewed areas of interest.

A comprehensive overview of spatio-temporal visual analytics for movement is given by [146,147].

Conclusions

Interdisciplinary collaborations such as the ones fostered in MOVE and described in this paper are reducing boundaries between disciplines that are interested in movement. To conclude, we propose as set of challenges that will be important to address in continued interdisciplinary collaborations between animal ecologists and ICT researchers. We identified five challenges, three based on specific themes in movement research and two more general ones, linked to characteristics of disparate scientific communities interested in movement.

Challenge 1: Navigation

An important topic in animal movement analysis is the question of navigation, specifically long-distance animal navigation [48]. How do animals navigate in their migration? Some species (terrestrial birds) may exhibit genetically or culturally inherited patterns, others (pelagic birds) do not, and other as yet unexplained mechanisms seem to account for their migratory navigation control [50]. Yet other species may rely on map related cues and relationships between celestial and magnetic compass for their orientation [48]. Long-term tracking can assist investigations into navigation and migration mechanisms, but this has to be combined with behavioural experiments and exploration of internal mechanisms, such as sensory perception, neurobiological state and genetic characteristics of migrant animals. Identifying spatial principles of long-distance animal navigation from such a complex set of sources will require interdisciplinary collaborations of the type that MOVE has shown, but on a broader scale with inclusion of biology, genetics and neuroscience. It will also require simulation methods to efficiently generate null model trajectories for long-haul displacements [200].

Challenge 2: Spatio-temporal dependency and heterogeneity

Movement trajectories are a special type of spatio-temporal data, that is, data with specific geographic and temporal location. Since recorded positions in a trajectory are not random, but are generated by continuous movement, the points in a trajectory are highly correlated in both space and time [5,201]. Indeed, the higher the temporal frequency of collection, the higher the correlation. This property is called the spatio-temporal dependency or the spatio-temporal autocorrelation and is a well-known issue in GIScience and spatial statistics [98,202-204]. A recent movement ecology study [205] incorporates geostatistical semivariogram modelling into movement analysis and another recent GIScience study incorporated spatial statistics measures (Getis-Ord Gi* statistic) with kernel density estimation and spatial data mining to identify periodicity patterns in migration trajectories of Arctic Barnacle Goose [206]. However, these are just first attempts and movement ecology could benefit from other developments in spatial and spatio-temporal statistics.

The second well-known property of spatio-temporal data is spatio-temporal heterogeneity or the property of geographical processes to vary over space and time [98]. Global statistical models used on spatial data often average characteristics from each location into descriptors that are valid over the entire area, but in fact are not valid anywhere. To address this, spatial statistics uses various local models, such as geographically and temporally weighted methods [203,204] to disaggregate descriptions of processes to individual locations in space and/or time. In ecology, an analogous problem is the distinction between modelling of movement at the population and individual levels [111] and exploration of local modelling could perhaps help address this problem in alternative ways.

Challenge 3: Human movement behaviour vs. animal movement behaviour

Recently, a number of studies in computer science focused on identification of human movement behaviour patterns from GPS trajectories. These studies look at identification of significant places (i.e. locations which play an important role in the activities of a user) [207-209], classification of human behaviour in these places [210,211] and analysis of spatial interactions between significant places identified from trajectories [212]. In addition to this, new tracking technologies for observation of human movement have been deployed as alternatives to GPS technology: an example are short range wireless technologies such as Bluetooth [213-216] and wireless networks [217].

The deployment of wireless sensors is a relatively new avenue in biologging sciences, with an ever increasing number of wildlife studies using them as ‘proximity sensors’ [218-221]. An interesting problem is therefore the question if and how much are human-centric movement methods and technologies transferrable into the animal-tracking context. For example, Bluetooth experiments generate a data set of flows between sensor locations; the animal-tracking analogue could be flows between most-visited places in a home range, which can be detected by deployment of wireless sensors on animals and in the environment [219]. Alternatively, methods for human significant places could be used for identification of animal significant places and the movement between these further investigated with human-behaviour related methods, such as spatial interaction.

Challenge 4: Statistics vs. exploratory data analysis

Ecology as a field is geared towards a hypothesis-driven approach that seeks confirmation through statistical testing. Indeed, most methodological questions on the recent list of one hundred fundamental ecological questions [2] are of a statistical nature. The information sciences are at the other side of the spectrum, geared towards exploratory data analysis, data-driven methods, data mining, visualisation and visual analytics. These perspectives are complementary and can enhance each other, but the challenge is how to best communicate and inform and learn from each other in achieving the goal: developing new methodologies for the analysis of large and complex new ecological movement data. One of the areas that could contribute to this challenge is visual analytics, where the links between exploration/visualisation and statistical methods are already being developed [222,223].

Challenge 5: Publication and dissemination

Interdisciplinary collaborations face the challenge of disciplinary differences in how results are published and disseminated in each discipline, as is well-illustrated by the vast variety of the publication venues in our literature list. In MOVE, animal ecologists paired up with researchers from GIScience and computer science, such as specialists in computational geometry, data mining, spatio-temporal databases, or visualisation. These disciplines and subdisciplines all have their specific cultures in how results are published and disseminated and target specific venues.

A challenge for interdisciplinary teams is therefore how to decide upon the best target for their work: should their work find place in a technical journal or a domain specific journal in order to reach the broadest possible audience? With increased availability of electronic sources, search engines are likely to return results from all sources, and hence findability is not an issue anymore. Perhaps more challenging is the culture of scientific recognition of publications, which differs widely between different disciplines. We support a holistic approach by disseminating interdisciplinary results as widely as possible and in all relevant communities. For this, information scientists could be encouraged to explore new possibilities and publish their novel methods in outlets of the target domain science, in this case movement ecology. Many of these methods have been developed for a purpose, and the ICT researchers have been able to benefit in the interdisciplinary collaborations from the data and expert knowledge contributed by the domain specialists. On the other hand, ecologists should be encouraged to collaboratively publish in ICT venues and follow these in order to benefit from the state-of-the-art in movement analysis in technical disciplines. This paper offers a useful list of ICT sources that are less familiar to ecologists, but worth exploring.

The above are some of the challenges that interdisciplinary collaborations in movement ecology may be facing. Of course, there are more challenges, particularly regarding technical issues or priorities of the research agenda. Some of these have already been addressed in the review of this paper. Challenges for the research agenda in movement research have also been sketched out in other initiatives that took place within the MOVE network, such as [224], who defined a research agenda focusing on the implications of working with real movement data, or [225] who defined the grand challenges of computational movement analysis. With new developments in tracking and other sensor technology, movement research is entering a golden age, with many more opportunities than challenges lying ahead. We believe that working across disciplines will allow researchers to address more and more ambitious questions about movement and we trust that MOVE has demonstrated this in a first attempt to raise interdisciplinary awareness. Further, we believe that interdisciplinarity can spark and foster unusual, innovative and exciting new ideas for movement research that no single discipline can produce on its own. We have experienced this within the MOVE network, as we hope this review demonstrates. We therefore also hope that the paper will achieve its goal of serving as a catalyst for further interdisciplinary collaborations in movement research.

Declarations

Acknowledgements

This paper presents the interdisciplinary methodological developments resulting from the COST Action IC0903, “MOVE: Knowledge Discovery from Moving Objects” (http://www.move-cost.info/). Funding from the COST Program (European Cooperation in Science and Technology) is gratefully acknowledged, as are the many contributions made by the MOVE participants to the results of this COST action. Figure 5 was provided by Johannes De Groeve and Figure 10 by Kuno Kurzhals. We also thank the three anonymous reviewers and Dr Jed Long from the University of St Andrews, Scotland, UK, for close reading of the manuscript and constructive suggestions for improvement.

Authors’ Affiliations

(1)
School of Geography & Geosciences, University of St Andrews
(2)
Department of Mathematics and Computer Science, Technical University Eindhoven
(3)
Biodiversity and Molecular Ecology Department, IASMA Research and Innovation Centre, Fondazione Edmund Mach
(4)
Department of Migration and Immuno-ecology, Max Planck Institute for Ornithology
(5)
Department of Biology, University of Konstanz
(6)
Department of Geography, Ghent University
(7)
Visualization Research Center, University of Stuttgart
(8)
Department of Geography, University of Zurich

References

  1. Nathan R, Getz WM, Revilla E, Holyoak M, Kadmon R, Saltz D, et al. A movement ecology paradigm for unifying organismal movement research. Proc Natl Acad Sci. 2008;105(49):19052–9.Google Scholar
  2. Sutherland WJ, Freckleton RP, Godfray HCJ, Beissinger SR, Benton T, Cameron DD, et al. Identification of 100 fundamental ecological questions. J Ecol. 2013;2013(101):58–67.Google Scholar
  3. Jeltsch F, Bonte D, Pe'er G, Reineking B, Leimgruber P, Balkenhol N, et al. Integrating movement ecology with biodiversity research - exploring new avenues to address spatiotemporal biodiversity dynamics. Movement Ecology. 2013;2013:1–6.Google Scholar
  4. Holden C. Inching Toward Movement Ecology. Science. 2006;313:779–82.Google Scholar
  5. Cagnacci F, Boitani L, Powell RA, Boyce MS. Animal ecology meets GPS-based radiotelemetry: a perfect storm of opportunities and challenges. Phil Trans Biol Sci. 2010;365:2157–62.Google Scholar
  6. Mueller T, Fagan WF. Search and navigation in dynamic environments from individual behaviors to population distributions. Oikos. 2008;117:654–64.Google Scholar
  7. Wikelski M, Kays RW, Kasdin NJ, Thorup K, Smith JA, Swenson Jr GW. Going wild: what a global small-animal tracking system could do for experimental biologists. J Exp Biol. 2007;210:181–6.Google Scholar
  8. Tomkiewicz SM, Fuller MR, Kie JG, Bates KK. Global positioning system and associated technologies in animal behaviour and ecological research. Phil Trans Biol Sci. 2010;365:2163–76.Google Scholar
  9. Bridge ES, Thorup K, Bowlin MS, Chilson PB, Diehl RH, Flacron RW, et al. Technology on the Move: Recent and Forthcoming Innovations for Tracking Migratory Birds. Bioscience. 2011;61:689–98.Google Scholar
  10. Kranstauber B, Cameron A, Weinzerl R, Fountain T, Tilak S, Wikelski M, et al. The Movebank data model for animal tracking. Environ Model Software. 2011;26:834–5.Google Scholar
  11. Rodriguez A, Negro JJ, Mulero M, Rodriguez C, Hernandez-Pliego J, Bustamante J. The Eye in the Sky: Combined Use of Unmanned Aerial Systems and GPS Data Loggers for Ecological Research and Conservation of Small Birds. PLoS One. 2012;7(12):e50336.Google Scholar
  12. Fedak MA. Marine animals as platforms for oceanographic sampling: a “win/win” situation for biology and operational oceanography. Memoir Natl Inst Polar Res. 2004;58:133–47.Google Scholar
  13. Calenge C, Draya S, Royer-Carenzia M. The concept of animals’ trajectories from a data analysis perspective. Ecol Informat. 2009;4:34–41.Google Scholar
  14. Wikelski M, Rienks F. Global Satellite Tracking of (Small) Animals Will Revolutionize Insight Into Migration, Human Health, and Conservation. ICARUS white paper. 2008. http://icarusinitiative.org/sites/default/files/ICARUS White Paper Oct 2008.pdf.Google Scholar
  15. Long JA, Nelson TA. A review of quantitative methods for movement data. Int J Geogr Inform Sci. 2012;27(2):292–318.Google Scholar
  16. Shamoun-Baranes J, van Loon EE, Purves RS, Speckmann B, Weiskopf D, Camphuysen CJ. Analysis and visualization of animal movement. Biol Lett. 2011;8(1):6–9.Google Scholar
  17. Sack J, Speckmann B, Van Loon EE and Weibel R (Eds.) 2010. Representation, Analysis and Visualization of Moving Objects, Dagstuhl Seminar Proceedings 10491, 5–10 Dec 2010, http://drops.dagstuhl.de/portals/index.php?semnr=10491.
  18. Gudmundsson J, Laube P, Van Loon, EE, 2012. Representation, Analysis and Visualization of Moving Objects, Dagstuhl Seminar Proceedings 12512, 16–21 Dec 2012. http://drops.dagstuhl.de/opus/volltexte/2013/3996/.
  19. Slingsby A, Dykes J. Experiences in involving analysts in visualisation design. Proceedings of BELIV ‘12: Beyond Time and Errors - Novel Evaluation Methods for Visualization, 14–15 Oct 2012. Seattle: USA; 2012.Google Scholar
  20. Seton ET. Life-histories of northern animals: an account of the mammals of Manitoba. New York, NY: Charles Scribner; 1909.Google Scholar
  21. Burt WH. Territoriality and Home Range Concepts as Applied to Mammals. J Mammal. 1943;24:346–52.Google Scholar
  22. Powell RA, Mitchell MS. What is a home range? J Mammal. 2012;93:948–58.Google Scholar
  23. Worton BJ. Kernel Methods for Estimating the Utilization Distribution in Home-Range Studies. Ecology. 1989;70(1):164–8.Google Scholar
  24. Kie JG, Matthiopoulos J, Fieberg J, Powell RA, Cagnacci F, Mitchell MS, et al. The home-range concept: are traditional estimators still relevant with modern telemetry technology? Phil Trans Biol Sci. 2010;365:2221–31.Google Scholar
  25. Getz WM, Fortmann-Roe S, Cross PC, Lyons AJ, Ryan SJ, Wilmers CC. LoCoH: Nonparameteric Kernel Methods for Constructing Home Ranges and Utilization Distributions. PLoS One. 2007;2(2):e207.Google Scholar
  26. Worton BJ. A review of models of home range for animal movement. Ecol Model. 1987;38:277–98.Google Scholar
  27. Laver PN, Kelly MJ. A Critical Review of Home Range Studies. J Wildl Manage. 2008;72(1):290–8.Google Scholar
  28. Börger L, Dalziel BD, Fryxell JM. Are there general mechanisms of animal home range behaviour? A review and prospects for future research. Ecol Lett. 2008;11(6):637–50.Google Scholar
  29. Moorcroft PR. Mechanistic approaches to understanding and predicting mammalian space use: recent advances, future directions. J Mammal. 2012;93(4):903–16.Google Scholar
  30. Silverman BW. Density estimation for statistics and data analysis. New York: Chapman and Hall; 1986.Google Scholar
  31. Seaman DE, Powell RA. An Evaluation of the Accuracy of Kernel Density Estimators for Home Range Analysis. Ecology. 1996;77(7):2075–85.Google Scholar
  32. Calenge C. The package “adehabitat” for the R software: A tool for the analysis of space and habitat use by animals. Ecol Model. 2006;198:516–9.Google Scholar
  33. Fieberg J. Kernel density estimators of home range: smoothing and the autocorrelation red herring. Ecology. 2007;88:1059–66.Google Scholar
  34. Brito JC. Seasonal Variation in Movements, Home Range, and Habitat Use by Male Vipera latastei in Northern Portugal. J Herpetol. 2003;37(1):155–60.Google Scholar
  35. Fieberg J, Börger L. Could you please phrase “home range” as a question? J Mammal. 2012;93:890–902.Google Scholar
  36. Walter DW, Beringer J, Hansen LP, Fischer JW, Millspaugh JJ, Vercauteren KC. Factors affecting space use overlap by white-tailed deer in an urban landscape. Int J Geogr Inform Sci. 2011;25(3):379–92.Google Scholar
  37. Lyons AJ, Turner WC, Getz WM. Home range plus: a space-time characterization of movement over real landscapes. Movement Ecology. 2013;2013:1–2.Google Scholar
  38. Long JA, Nelson TA. Time Geography and Wildlife Home Range Delineation. J Wildl Manage. 2011;76(2):407–13.Google Scholar
  39. Bullard F. Estimating the Home Range of an Animal: a Brownian Bridge Approach. University of North Carolina at Chapel Hill: MSc thesis; 1999.Google Scholar
  40. Horne JS, Garton EO, Krone SM, Lewis JS. Analyzing Animal Movements Using Brownian Bridges. Ecology. 2007;88(9):2354–63.Google Scholar
  41. Benhamou S. Dynamic Approach to Space and Habitat Use Based on Biased Random Bridges. PLoS One. 2011;6(1):e14592.Google Scholar
  42. Kranstauber B, Kays R, LaPoint SD, Wikelski M, Safi K. A dynamic Brownian bridge movement model to estimate utilization distributions for heterogeneous animal movement. J Anim Ecol. 2012;81:738–46.Google Scholar
  43. Benhamou S, Cornélis D. Incorporating Movement Behavior and Barriers to Improve Kernel Home Range Space Use Estimates. J Wildl Manage. 2010;74(6):1353–60.Google Scholar
  44. Benhamou S, Riotte-Lambert L. Beyond the Utilization Distribution: Identifying home range areas that are intensively exploited or repeatedly visited. Ecol Model. 2012;227(2012):112–6.Google Scholar
  45. Sinclair ARE. The function of distance movements in vertebrates. In: Swingland IR, Greenwood PJ, editors. The ecology of animal movement. Oxford, UK: Claredon Press; 1983. p. 248–58.Google Scholar
  46. Bunnefeld N, Börger L, Van Moorter B, Rolandsen CM, Dettki H, Solberg EJ. A model‐driven approach to quantify migration patterns: individual, regional and yearly differences. J Anim Ecol. 2011;80:466–76.Google Scholar
  47. Cagnacci F, Focardi S, Heurich M, Stache A, Hewison AJM, Morellet N, et al. Partial migration in roe deer: migratory and resident tactics are end points of a behavioural gradient determined by ecological factors. Oikos. 2011;120:1790–802.Google Scholar
  48. Alerstam T. Conflicting Evidence About Long-Distance Animal Navigation. Science. 2006;313:791–4.Google Scholar
  49. Freeman R, Mann R, Guilford T, Biro D. Group decisions and individual differences: route fidelity predicts flight leadership in homing pigeons (Columba livia). Biol Lett. 2010;7:63–6.Google Scholar
  50. Guildford R, Freeman R, Boyle D, Dean B, Kirk H, Phillips R, et al. A Dispersive Migration in the Atlantic Puffin and Its Implications for Migratory Navigation. PLoS One. 2011;6(7):e21336.Google Scholar
  51. Hansen IJ, Johnson CH, Cluff HD. Synchronicity of movement paths of barren-ground caribou and tundra wolves. Polar Biology. 2013;36:1363–71.Google Scholar
  52. Pettit B, Perna A, Biro D, Sumpter DJT. Interaction rules underlying group decisions in homing pigeons. J R Soc Interface. 2013;10:20130529.Google Scholar
  53. Dias MP, Granadeiro JP, Catry P. Individual variability in the migratory path and stopovers of a long-distance pelagic migrant. Anim Behav. 2013;86:359–64.Google Scholar
  54. Long JA, Nelson TA. Measuring Dynamic Interaction in Movement Data. Trans GIS. 2013;17(1):62–77.Google Scholar
  55. Williams DM, Dechen Quinn AC, Porter WF. Informing Disease Models with Temporal and Spatial Contact Structure among GPS-Collared Individuals in Wild Populations. PLoS One. 2014;9(1):e84368.Google Scholar
  56. Raisman EA, Barner Rasmussen H, King LE, Ihwagi FW, Douglas-Hamilton I. Feasibility study on the spatial and temporal movement of Samburu’s cattle and wildlife in Kenya using GPS radio-tracking, remote sensing and GIS. Prev Vet Med. 2013;111:76–80.Google Scholar
  57. Harris KJ, Blackwell PG. Flexible continuous-time modelling for heterogeneous animal movement. Ecol Model. 2013;255:29–37.Google Scholar
  58. Reimers E, Tsegaye D, Colman JE, Efetstøl S. Activity patterns in reindeer with domestic vs. wild ancestry. Appl Anim Behav Sci. 2014;150:74–84.Google Scholar
  59. Riotte-Lambert L, Benhamou S, Chamaillé-Jammes S. Periodicity analysis of movement recursions. Journal of Theoretical Biology. 2013;317:238–43.Google Scholar
  60. Fryxell JM, Hazell M, Börger L, Dalziel BD, Haydon DT, Morales JM. Multiple movement modes by large herbivores at multiple spatiotemporal scales. Proc Natl Acad Sci. 2008;105(49):19114–9.Google Scholar
  61. Benhamou S. How to reliably estimate the tortuosity of an animal’s path: straightness, sinuosity, or fractal dimension? J Theor Biol. 2004;229:209–20.Google Scholar
  62. Long JA, Nelson TA, Nathoob FS. Toward a kinetic-based probabilistic time geography. Int J Geogr Inform Sci. 2014;28(5):855–74.Google Scholar
  63. Miller JA. Using Spatially Explicit Simulated Data to Analyze Animal Interactions: A Case Study with Brown Hyenas in Northern Botswana. Trans GIS. 2012;16(3):271–91.Google Scholar
  64. Levitis DA, Lidicker Jr WZ, Freund G. Behavioural biologists do not agree on what constitutes behaviour. Anim Behav. 2009;78(1):103–10.Google Scholar
  65. Dingle H, Drake VA. What is migration? Bioscience. 2007;57:113–21.Google Scholar
  66. Ropert-Coudert Y, Wilson RP. Trends and perspectives in animal-attached remote sensing. Front Ecol Environ. 2005;3(8):437–44.Google Scholar
  67. Brown DD, Kays R, Wikelski M, Wilson R, Klimley AP. Observing the unwatchable through acceleration logging of animal behavior. Animal Biotelemetry. 2013;2013:1–20.Google Scholar
  68. Beyer HL, Morales JM, Murray DL, Fortin M-J. Estimating behavioural states from movement paths using Bayesian state-space models: a proof of concept. Methods in Ecology and Evolution. 2013;4:433–41.Google Scholar
  69. Gurarie E, Andrews RD, Laidre KL. A novel method for identifying behavioural changes in animal movement data. Ecol Lett. 2009;12:395–408.Google Scholar
  70. Gautestad AO, Mysterud A. The Lévy flight foraging hypothesis: forgetting about memory may lead to false verification of Brownian motion. Movement Ecology. 2013;2013:1–9.Google Scholar
  71. Morales JM, Haydon DT, Frair J, Holsinger KE, Fryxell JM. Extracting More from Relocation Data: Building Movement Models as Mixtures of Random Walks. Ecology. 2004;85(9):2436–45.Google Scholar
  72. Schwager M, Anderson DM, Butler Z, Rus D. Robust classification of animal tracking data. Comput Electron Agr. 2007;56:46–59.Google Scholar
  73. Wilmers CC, Wang Y, Nickel B, Houghtaling P, Shakeri Y, Allen ML, et al. Scale Dependent Behavioral Responses to Human Development by a Large Predator, the Puma. PLoS One. 2013;8(4):e60590.Google Scholar
  74. Shamoun-Baranes J, Bouten W, Camphuysen CJ, Baaij E. Riding the tide: intriguing observations of gulls resting at sea during breeding. Ibis. 2011;153(2):411–5.Google Scholar
  75. Nathan R, Spiegel O, Fortmann-Roe S, Harel R, Wikelski M, Getz WM. Using tri-axial acceleration data to identify behavioral modes of free-ranging animals: general concepts and tools illustrated for griffon vultures. J Exp Biol. 2012;215:986–96.Google Scholar
  76. Shamoun-Baranes J, Bom R, van Loon EE, Ens BJ, Oosterbeek K, Bouten W. From Sensor Data to Animal Behaviour: An Oystercatcher Example. PLoS One. 2012;7(5):e37997.Google Scholar
  77. Kays R, Jansen PA, Knecht EMH, Vohwinkel R, Wikelski M. The effect of feeding time on dispersal of Virola seeds by toucans determined from GPS tracking and accelerometers. Acta Oecol. 2011;37:625–31.Google Scholar
  78. Guo Y, Poulton G, Corke P, Bishop-Hurley GJ, Wark T, Swain DL. Using accelerometer, high sample rate GPS and magnetometer data to develop a cattle movement and behaviour model. Ecol Model. 2009;220:2068–75.Google Scholar
  79. Gerencser L, Vasarhely G, Nagy M, Viscek T, Miklosi A. Identification of Behaviour in Freely Moving Dogs (Canis familiaris) Using Inertial Sensors. PLoS One. 2013;8(10):e77814.Google Scholar
  80. Rutz C, Bluff LA, Weir AAS, Kacelnik A. Video cameras on wild birds. Science. 2007;318(5851):765–765.Google Scholar
  81. Cachat J, Stewart A, Utterback E, Hart P, Gaikwad S, Wong K, et al. Three-Dimensional Neurophenotyping of Adult Zebrafish Behavior. PLoS One. 2011;6(3):e17597.Google Scholar
  82. Dodge S, Bohrer G, Weinzierl R, Davidson SC, Kays R, Douglas D, et al. The environmental-data automated track annotation (Env-DATA) system: linking animal tracks with environmental data. Movement Ecology. 2013;2013:1–3.Google Scholar
  83. Coyne MS, Godley BJ. Satellite Tracking and Analysis Tool (STAT): an integrated system for archiving, analyzing and mapping animal tracking data. Mar Ecol Prog Ser. 2005;301:1–7.Google Scholar
  84. Dodge S, Bohrer G, Bildstein K, Davidson SC, Weinzierl R, Bechard MJ, et al. Environmental drivers of variability in the movement ecology of turkey vultures (Cathartes aura) in North and South America. Phil Trans Biol Sci. 2014;369:20130195.Google Scholar
  85. Morellet N, Bonenfant C, Börger L, Ossi F, Cagnacci F, Heurich M, et al. Seasonality, weather and climate affect home range size in roe deer across a wide latitudinal gradient within Europe. J Anim Ecol. 2013;82(6):1326–39.Google Scholar
  86. Safi K, Kranstauber B, Weinzierl R, Griffin L, Rees EC, Cabot D, et al. Flying with the wind: scale dependency of speed and direction measurements in modelling wind support in avian flight. Movement Ecology. 2013;2013:1–4.Google Scholar
  87. Shamoun-Baranes J, van Loon EE, van Gasteren H, van Belle J, Bouten W, Buurma L. A comparative analysis of the influence of weather on the flight altitudes of birds. Bull Am Meteorol Soc. 2006;87:47–61.Google Scholar
  88. Robinson DW, Bowlin MS, Bisson I, Shamoun-Baranes J, Thorup K, Diehl RH, et al. Intergating concepts and technologies to advance the study of bird migration. Front Ecol Environ. 2010;8(7):354–61.Google Scholar
  89. Willems EP, Barton RA, Hill RA. Remotely sensed productivity, regional home range selection, and local range use by an omnivorous primate. Behav Ecol. 2009. doi:10.1093/beheco/arp087.Google Scholar
  90. Hastie GD, Gillespie DM, Gordon JCD, MacAulay JDJ, McConnell BJ, Sparling CE. Tracking technologies for quantifying marine mammal interactions with tidal turbines: pitfalls and possibilities. In: Shields M, Payne A, editors. Marine Renewable Energy Technology and Environmental Interactions. Humanity and the Sea. Dordrecht: Springer Science and Business Media; 2014.Google Scholar
  91. Cooke SJ, Hinch SG, Wikelski M, Andrews RD, Kuchel LJ, Wolcott TG, et al. Biotelemetry: a mechanistic approach to ecology. Trends Ecol Evol. 2004;19(6):334–43.Google Scholar
  92. Cook TR, Hamann M, Pichegru L, Bonadonna F, Gremillet D, Ryan PG. GPS and time-depth loggers reveal underwater foraging plasticity in a flying diver, the Cape Cormorant. Mar Biol. 2012;159:373–87.Google Scholar
  93. Fedak MA. The impact of animal platforms on polar ocean observation. Deep Sea Research Part II: Topical Studies in Oceanography. 2013;88–89(1):7–13.Google Scholar
  94. Roquet F, Wunsch C, Forget G, Heimbach P, Guinet C, Reverdin G, et al. Estimates of the Southern Ocean general circulation improved by animal-borne instruments. Geophys Res Lett. 2013;40:1–5.Google Scholar
  95. Chevaillier D, Karpytchev M, McConnell BJ, Moss S, Vincent C. Can gray seals maintain heading within areas of high tidal current? Preliminary results from numerical modeling and GPS observations. Mar Mamm Sci. 2014;30(1):374–80.Google Scholar
  96. Urbano F, Cagnacci F, Calenge C, Dettki H, Cameron A, Neteler M. Wildlife tracking data management: a new vision. Phil Trans Biol Sci. 2010;365(1550):2177–85.Google Scholar
  97. De Berg M, Cheong O, van Kreveld M, Overmars M. Computational Geometry. 3rd ed. Berlin-Heidelberg: Springer Verlag; 2008.Google Scholar
  98. Wilson JP, Fotheringham AS. The Handbook of Geographic Information Science. Oxford: Blackwell Publishing; 2008.Google Scholar
  99. Buchin M, Kruckenberg H, Kölzsch A. Segmenting Trajectories by Movement States. In: Timpf S, Laube P, editors. Advances in Spatial Data Handling, Advances in Geographic Information Science. Berlin Heidelberg: Springer-Verlag; 2013. p. 15–25.Google Scholar
  100. Buchin M, Driemel A, van Kreveld M, Sacristan V. Segmenting Trajectories: A framework and algorithms using spatiotemporal criteria. Journal of Spatial Information Science. 2011;3:33–63.Google Scholar
  101. Alewijnse SPA. A framework for trajectory segmentation by stable criteria and Brownian bridge movement model. Master’s thesis: Eindhoven University of Technology; 2013.Google Scholar
  102. Aronov B, Driemel A, van Kreveld M, Löffler M, Staals F. Segmentation of Trajectories on Non-Monotone Criteria. New Orleans, Louisiana, USA: Proceedings of the 24th Annual ACM-SIAM Symposium on Discrete Algorithms, 6-8 January 2013; 2013. p. 1897–911.Google Scholar
  103. Alewijnse SPA, Buchin K, Buchin M, Sijben S, Westenberg MA. Model-based segmentation and classification of trajectories. Dead Sea, Israel: Proceedings of the 30th European Workshop on Computational Geometry March 3-5, 2014; 2014.Google Scholar
  104. Sester M, Feuerhake U, Kuntzsch C, Zhang L. Revealing Underlying Structure and Behaviour from Movement Data. KI - Künstliche Intelligenz. 2012;2012:1–9.Google Scholar
  105. Panagiotakis C, Pelekis N, Kopanakis I, Ramasso E, Theodoridis Y. Segmentation and Sampling of Moving Object Trajectories based on Representativeness. IEEE Transactions on Knowledge and Data Engineering. 2011;2011:1–17.Google Scholar
  106. Buchin K, Buchin M, van Kreveld M, Löffler M, Silveira RI, Wenk C, et al. Median Trajectories. Algorithmica. 2013;66(3):595–614.Google Scholar
  107. Brundson C. Path Estimation from GPS Tracks. Maynooth: Ireland: Proceedings of Geocomputation 2007; 2007.Google Scholar
  108. Goren-Bar T, Greenfeld J. A Method for Constructing 3D Traveling Routes from GPS Navigation Data. Redondo Beach, California, USA: Proceedings of the Third ACM SIGSPATIAL International Workshop on GeoStreaming, November 2012; 2012.Google Scholar
  109. Etienne L, Devogele T, Bouju A. Spatio-temporal trajectory analysis of mobile objects following the same itinerary. Advances in Geo-Spatial Information Science. 2012;10:47–57.Google Scholar
  110. Pelekis N, Kopanakis I, Kotsifakos EE, Frentzos E, Theodoridis Y. Clustering Uncertain Trajectories. Knowledge and Information Systems. 2011;28(1):117–47.Google Scholar
  111. Fieberg J, Rieger RH, Zicus MC, Schildcrout JS. Regression modelling of correlated data in ecology: subject specific and population averaged response patterns. J Appl Ecol. 2009;46:1018–25.Google Scholar
  112. Goodchild MF. Scale in GIS: An overview. Geomorphology. 2011;130:5–9.Google Scholar
  113. Laube P, Purves RP. How fast is a cow? Cross-Scale Analysis of Movement Data. Trans GIS. 2011;15(3):401–18.Google Scholar
  114. Soleymani A, Cachat J, Robinson K, Dodge S, Kalueff AV, Weibel R. Integrating cross-scale analysis in the spatial and temporal domains for classification of behavioral movement. Journal of Spatial Information Science. 2014;8:1–25.Google Scholar
  115. Soleymani A, van Loon E, Weibel R. Capability of movement features extracted from GPS trajectories for the classification of fine-grained behaviors. Castellón, Spain: Proceedings of the AGILE 2014 International Conference on Geographic Information Science, 3–6 June 2014; 2014. http://www.agile-online.org/Conference_Paper/cds/agile_2014/agile2014_150.pdf.Google Scholar
  116. Dodge S, Weibel R, Lautenschütz K. Towards a taxonomy of movement patterns. Information Visualization. 2008;2008:1–13.Google Scholar
  117. Orellana D, Wachowicz M, De Knegt HJ, Ligtenberg A, Bregt AK. Uncovering Patterns of Suspension in Movement. Zurich, Switzerland: Proceedings of GIScience, 4-7 September 2010; 2010.Google Scholar
  118. Jain AK, Murty MN, Flynn PJ. Data Clustering: A Review. ACM Computing Surveys. 1999;31(3):264–323.Google Scholar
  119. Xu R, Wunsch D. Survey of Clustering Algorithms. IEEE Trans Neural Netw. 2005;16(3):645–78.Google Scholar
  120. Buchin K, Buchin M, Gudmundsson J, Löffler M, Luo J. Detecting Commuting Patterns by Clustering Subtrajectories. International Journal of Computational Geometry and Applications. 2011;21(3):253–82.Google Scholar
  121. Buchin K, Buchin M, van Kreveld MJ, Luo J. Finding long and similar parts of trajectories. Computational Geometry. 2011;44(9):465–76.Google Scholar
  122. Buchin K, Buchin M, Gudmundsson J. Constrained free space diagrams: A tool for trajectory analysis. Int J Geogr Inform Sci. 2010;24(7):1101–25.Google Scholar
  123. Pelekis N, Andrienko G, Andrienko N, Kopanakis I, Marketos G, Theodoridis Y. Visually Exploring Movement Data via Similarity-based Analysis. Journal of Intelligent Information Systems. 2012;38(2):343–91.Google Scholar
  124. Merki M, Laube P. Detecting reaction movement patterns in trajectory data. Avignon, France: Proceedings of the AGILE 2012 International Conference on Geographic Information Science, 25-27 April 2012; 2012.Google Scholar
  125. Rinzivillo S, Pedreschi D, Nanni M, Giannotti F, Andrienko N, Andrienko G. Visually Driven Analysis of Movement Data by Progressive Clustering. Information Visualization. 2008;7:225–39.Google Scholar
  126. Nanni M, Pedreschi D. Time-focused clustering of trajectories of moving objects. Journal of Intelligent Information Systems. 2006;27:267–89.Google Scholar
  127. Buchin K, Buchin M, van Kreveld M, Speckmann B, Staals F. Trajectory Grouping Structure. London, Ontario, Canada: Proceedings of 2013 Algorithms and Data Structures Symposium (WADS), August 12-14, 2013; 2013.Google Scholar
  128. Yanagisawa Y, Satoh T. Clustering Multidimensional Trajectories based on Shape and Velocity. Atlanta, Georgia, USA: Proceedings of the 22nd IEEE International Conference on Data Engineering Workshops (ICDEW'06), 3-6 April 2006; 2006.Google Scholar
  129. Dodge S, Weibel R, Forootan E. Revealing the physics of movement: Comparing the similarity of movement characteristics of different types of moving objects. Comput Environ Urban Syst. 2009;33:419–34.Google Scholar
  130. Hanks EM, Hooten MB, Johnson DS, Sterling JT. Velocity-Based Movement Modeling for Individual and Population Level Inference. PLoS One. 2011;6(8):e22795.Google Scholar
  131. Dodge S, Weibel R, Laube P. Trajectory Similarity Analysis in Movement Parameter Space. Plymouth, UK: Proceedings of GISRUK, 27-29 April 2011; 2011.Google Scholar
  132. Dodge S, Laube P, Weibel R. Movement similarity assessment using symbolic representation of trajectories. Int J Geogr Inform Sci. 2012;26(9):1563–88.Google Scholar
  133. McArdle G, Demšar U, van der Spek S, McLoone S. Interpreting Pedestrian Behaviour by Visualising and Clustering Movement Data. In: Liang S, Wang X, Claramunt C, editors. 12th International Symposium on Web and Wireless Geographical Information Systems W2GIS 2013. 4–5 April 2013, vol. 7820. Banff, Alberta, Canada: Lecture Notes in Computer Science, Springer; 2013. p. 64–81.Google Scholar
  134. McArdle G, Demšar U, van der Spek S, McLoone S. Classifying Pedestrian Movement Behaviour From GPS Trajectories using Visualistion and Clustering. Annals of GIS. 2014;20(2):85–98.Google Scholar
  135. Çöltekin A, Demšar U, Brychtova A, Vandrol J. Eye-hand coordination during visual search on geographic displays. Vienna, Austria: Spatial Eye-tracking Workshop at GIScience 2014. 23-26 Sept 2014; 2014.Google Scholar
  136. Ranacher P, Tzavella K. How to compare movement? A review of physical movement similarity measures in geographic information science and beyond. Cartography and Geographic Information Science. 2014;41(3):286–307.Google Scholar
  137. Buchin M, Dodge S, Speckmann B. Context-Aware Similarity of Trajectories. Proc. 7th International Conference on Geographic Information Science (GIScience). Lecture Notes in Computer Science. 2012;7478:43–56.Google Scholar
  138. De Groeve J, Van de Weghe N, Neutens T, Cagnacci F. Roe deer (Capreolus capreolus) spatio-temporal sequential habitat use. An application of Sequence Alignment Methods. Vienna, Austria: MOVE Final conference, 30 Sept – 1 Oct 2013; 2013.
  139. Sankoff D, Kruskal J. Time Warps, String Edits, and Macromolecules: The Theory and Practice of Sequence Comparison. Reading: Addison Wesley; 1983.Google Scholar
  140. Ware C. Information Visualization: Perception for Design. 3rd ed. Elsevier/Morgan: Kaufmann; 2013.Google Scholar
  141. Tory M, Möller T. Rethinking visualization: A high-level taxonomy, vol. 2004. Austin, Texas, USA: Proceedings of the IEEE Symposium on Information Visualization 10-12 October 2004; 2004. p. 151–8.Google Scholar
  142. Thomas JJ, Cook KA. Illuminating the path: The research and development agenda for visual analytics. Los Alamitos, CA: IEEE Computer Society; 2005.Google Scholar
  143. Keim D, Andrienko G, Fekete J-D, Görg C, Kohlhammer J, Melançon G. Visual analytics: definition, process, and challenges. In: Kerren A, Stasko JT, Fekete J-D, North C, Kerren A, Stasko JT, Fekete J-D, North C, editors. Information visualization – human-centered issues and perspectives. LNCS state-of the-art survey, vol. 4950. Berlin-Heidelberg: Springer; 2008. p. 154–75.Google Scholar
  144. Andrienko G, Andrienko N, Demšar U, Dransch D, Dykes J, Fabrikant SI, et al. Space, time and visual analytics. Int J Geogr Inform Sci. 2010;24(10):1577–600.Google Scholar
  145. Andrienko G, Andrienko N. A general framework for using aggregation in visual exploration of movement data. The Cartographic Journal. 2010;47(1):22–40.Google Scholar
  146. Andrienko N, Andrienko G. Visual Analytics of Movement: an Overview of Methods, Tools, and Procedures. Information Visualization. 2013;12(1):3–24.Google Scholar
  147. Andrienko G, Andrienko N, Bak P, Keim DA, Wrobel S. Visual Analytics of Movement. Berlin-Heidelberg: Springer Verlag; 2013.Google Scholar
  148. Grundy E, Jones MW, Laramee RS, Wilson RP, Shephard ELC. Visualisation of Sensor Data from Animal Movement. Eurographics. 2009;28(3):815–22.Google Scholar
  149. Spretke D, Janetzko H, Mansmann F, Bak P, Kranstauber B, Davidson S. Exploration through Enrichment: A Visual Analytics Approach for Animal Movement. Chicago, IL, USA: Proceedings of ACM SIGSPATIAL GIS 2011. 1-4 Nov 2011; 2011.Google Scholar
  150. Janetzko H, Jäckle D, Deussen O, Keim DA. Visual Abstraction of Complex Motion Patterns. In: Wong PC, Kao DL, Hao MC, Chen C, editors. Visualization and Data Analysis 2014, vol. 9017. San Francisco, California, USA: Proceedings of SPIE-IS&T Electronic Imaging, 2 - 6 Feb 2014; 2014.Google Scholar
  151. Pequet D. Representations of Space and Time. New York, USA: The Guilford Press; 2002.Google Scholar
  152. Kwan M-P, Neutens T. Space-time research in GIScience. Int J Geogr Inform Sci. 2014;28(5):851–4.Google Scholar
  153. Hägerstrand T. What about people in regional science? Papers of the Regional Science Association. 1970;24:7–21.Google Scholar
  154. Kwan M-P. Interactive geovisualization of activity-travel patterns using three-dimensional geographical information systems: a methodological exploration with a large data set. Transportation Research Part C. 2000;8:185–203.Google Scholar
  155. Kapler T, Wright W. GeoTime information visualization. Information Visualization. 2005;4:136–46.Google Scholar
  156. Kraak M-J. Geovisualization and time – new opportunities for space-time cube. In: Dodge M, McDerby M, Turner M, editors. Geographic visualization: concepts, tools and applications. Chichester, UK: John Wiley & sons; 2008. p. 293–306.Google Scholar
  157. Bach B, Dragicevic P, Archambault D, Hurger C, Carpendale S. A Review of Temporal Data Visualizations based on Space-Time Cube Operations. STAR – State of the Art Report. Swansea, UK: Eurographics Conference on Visualization (EuroVis) 2014, 9-13 June 2014; 2014.Google Scholar
  158. Demšar U, Çöltekin A. Quantifying the interactions between eye and mouse movements on spatial visual interfaces through trajectory visualisations. Vienna, Austria: Workshop on Analysis of Movement Data at GIScience 2014. 23-26 Sept 2014; 2014.Google Scholar
  159. Kristensson PO, Dahlbäck N, Anundi D, Björnsad M, Gillberg H, Haraldsson J, et al. An Evaluation of Space Time Cube Representation of Spatio Temporal Patterns. IEEE Trans Vis Comput Graph. 2009;15(4):696–702.Google Scholar
  160. Kveladze I, Kraak M-J. What do we know about the space-time cube from cartographic and usability perspective? Columbus, Ohio, USA: Proceedings of Autocarto 2012. 16–18 Sept 2012; 2012.Google Scholar
  161. Kveladze I, Kraak M-J, van Elzakker CPJM. A Methodological Framework for Researching the Usability of the Space-Time Cube. The Cartographic Journal. 2013;50(3):201–10.Google Scholar
  162. Aigner W, Miksch S, Müller W, Schumann H, Tominski C. Visualizing time-oriented data – a systematic view. Computers & Graphics. 2007;31:401–9.Google Scholar
  163. Aigner W, Miksch S, Müller W, Schumann H, Tominski C. Visual Methods for Analyzing Time-Oriented Data. IEEE Trans Vis Comput Graph. 2008;14(1):47–60.Google Scholar
  164. Kölzsch A, Slingsby A, Wood J, Nolet BA and Dykes J. Visualisation design for representing bird migration tracks in time and space. Workshop on Visualisation in Environmental Sciences (EnvirVis), Jul 2013, Leipzig, Germany: 2013. http://openaccess.city.ac.uk/2384/
  165. Slingsby A and van Loon EE. Visual Analytics for Exploring Changes in Biodiversity. Workshop on Visualisation in Environmental Sciences (EnvirVis), Jul 2013. Leipzig, Germany: 2013. http://openaccess.city.ac.uk/2385/.
  166. Zhang QC, Slingsby A, Dykes J, Wood J, Kraak MJ, Blok CA, et al. Visual analysis design to support research into movement and use of space in Tallinn: A case study. Information Visualization. 2014;13(3):213–31.Google Scholar
  167. Wang Z, Yuan X. Urban Trajectory Timeline Visualization. Bangkok, Thailand: Proceedings of IEEE BigComp 2014; 2014.Google Scholar
  168. Netzel R, Burch M, Weiskopf D. Comparative eye tracking study on node-link visualizations of trajectories. IEEE Trans Vis Comput Graph. 2014;12(20):2221–30.Google Scholar
  169. Holten D, van Wijk JJ. Force‐Directed Edge Bundling for Graph Visualization. Computer Graphics Forum. 2009;28(3):983–90.Google Scholar
  170. Höferlin M, Höferlin B, Heidemann G, Weiskopf D. Interactive schematic summaries for faceted exploration. IEEE Transactions on Multimedia. 2013;15(4):908–20.Google Scholar
  171. Verbeek K, Buchin K, Speckmann B. Flow map layout via spiral trees. IEEE Trans Vis Comput Graph. 2011;17(12):2536–44.Google Scholar
  172. Hurter C, Ersoy O, Fabrikant SI, Telea A. Bundled Visualization of Dynamic Graph and Trail Data. IEEE Trans Vis Comput Graph. 2014;20(8):1141–57.Google Scholar
  173. Tominski C, Schumann H, Andrienko G, Andrienko N. Stacking-Based Visualization of Trajectory Attribute Data. IEEE Trans Vis Comput Graph. 2012;18(12):2565–74.Google Scholar
  174. Miller HJ. Modelling accessibility using space-time prism concepts within geographical information systems. Int J Geogr Inform Sci. 1991;5(3):287–301.Google Scholar
  175. Downs JA, Horner MW, Tucker AD. Time-geographic density estimation for home range analysis. Annals of GIS. 2011;17(3):163–71.Google Scholar
  176. Downs JA, Horner MW. Analysing infrequently sampled animal tracking data by incorporating generalized movement trajectories with kernel density estimation. Comput Environ Urban Syst. 2012;36:302–10.Google Scholar
  177. Willems N, van de Wetering H, van Wijk JJ. Visualization of vessel movements. Eurographics/IEEE-VGTC Symposium on Visualization. 2009;2009:28(3).Google Scholar
  178. Willems N. Visualization of Vessel Traffic. PhD thesis. The Netherlands: Eindhoven University of Technology; 2011.
  179. Scheepens R, Willems N, van de Wetering H, Andrienko G, Andrienko N, van Wijk JJ. Composite Density Maps for Multivariate Trajectories. IEEE Trans Vis Comput Graph. 2011;17(12):2518–27.Google Scholar
  180. Scheepens R, van de Wetering H, van Wijk JJ. Contour based visualization of vessel movement predictions. Int J Geogr Inform Sci. 2014;28(5):891–909.Google Scholar
  181. Krisp JM, Peters S, Burkert F. Visualizing Crowd Movement Patterns Using a Directed Kernel Density Estimation. Earth Observation of Global Changes (EOGC). Lecture Notes in Geoinformation and Cartography. 2013;2013:255–68.Google Scholar
  182. Buchin K, Sijben S, Arseneau TJ, Willems EP. Detecting Movement Patterns using Brownian Bridges. In: Proceedings of the 20th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (ACM GIS). 6-9 Nov 2012, Redondo Beach, California, USA. 2012.Google Scholar
  183. Kurzhals K, Weiskopf D. Space-Time Visual Analytics of Eye-Tracking Data for Dynamic Stimuli. IEEE Trans Vis Comput Graph. 2013;19(12):2129–38.Google Scholar
  184. Nakaya T, Yano K. Visualising crime clusters in a space-time cube: and exploratory data-analysis approach using space-time kernel density estimation and scan statistics. Trans GIS. 2010;14(3):223–39.Google Scholar
  185. Delmelle E, Dony C, Casas I, Jia M, Tang W. Visualizing the impact of space-time uncertainties on dengue fever patterns. Int J Geogr Inform Sci. 2014;28(5):1107–27.Google Scholar
  186. Downs JA, Horner MW, Hyzera G, Lamba D, Loraamm R. Voxel-based probabilistic space-time prisms for analysing animal movements and habitat use. Int J Geogr Inform Sci. 2014;28(5):875–90.Google Scholar
  187. Tracey JA, Sheppard J, Zhu J, Wei F, Swaisgood RS, Fisher RN. Movement-Based Estimation and Visualization of Space Use in 3D for Wildlife Ecology and Conservation. PLoS One. 2014;9(7):e101205.Google Scholar
  188. Demšar U, Virrantaus K. Space-time density of trajectories: exploring spatiotemporal patterns in movement data. Int J Geogr Inform Sci. 2010;24(10):1527–42.Google Scholar
  189. Demšar U, van Loon EE. Visualising movement: The seagull. Significance. 2013;10(5):40–2.Google Scholar
  190. Demšar U, Buchin K, van Loon EE, Shamoun-Baranes J. Stacked space-time densities: a geovisualisation approach to explore dynamics of space use over time. GeoInformatica. 2015;19(1):85–115.Google Scholar
  191. Roberts J. State of the art: Coordinated & multiple views in exploratory visualization. In: Proceedings of the International Conference on Coordinated and Multiple Views in Exploratory Visualization (CMV ‘07). 2 July 2007, Zurich, Switzerland. 2007. 2007. p. 61–71.Google Scholar
  192. Wills G. Linked Data Views. In: Chen C-H, Härdle W, Unwin A, editors. Handbook of Data Visualization. Berlin-Heidelberg: Springer Verlag; 2008.Google Scholar
  193. Andrienko G, Andrienko N, Heurich M. An eventbased conceptual model for context-aware movement analysis. Int J Geogr Inform Sci. 2011;25(9):1347–70.Google Scholar
  194. Andrienko N, Andrienko G, Barrett L, Dostie M, Henzi SP. Space Transformation for Understanding Group Movement. IEEE Trans Vis Comput Graph. 2013;19(12):2169–78.Google Scholar
  195. Andrienko G, Andrienko N, Burch M, Weiskopf D. Visual Analytics Methodology for Eye Movement Studies. IEEE Trans Vis Comput Graph. 2012;18(12):2889–98.Google Scholar
  196. Ooms K, Andrienko G, Andrienko N, De Maeyer P, Fack V. Analysing the spatial dimension of eye movement data using a visual analytic approach. Expert Systems with Applications. 2012;39(1):1324–32.Google Scholar
  197. Kurzhals K, Heimerl F, Weiskopf D. ISeeCube: Visual analysis of gaze data for video. Safety Harbor, Florida, USA: Proceedings of the ACM Conference on Eye Tracking Research and Applications (ETRA 2014), 26 – 28 March, 2014; 2014. p. 43–50.Google Scholar
  198. Kurzhals K, Bopp CF, Bässler J, Ebinger F, Weiskopf D. Benchmark data for evaluating visualization and analysis techniques for eye tracking for video stimuli. Paris, France: Proceedings of the 2014 BELIV Workshop: Beyond Time and Errors - Novel Evaluation Methods for Visualization. 10 Nov 2014; 2014.Google Scholar
  199. Levenshtein V. Binary codes capable of correcting deletions, insertions, and reversals. Soviet Physics-Doklady. 1966;10(8):707–10.Google Scholar
  200. Technitis G, Weibel R. An Algorithm for Random Trajectory Generation Between Two Endpoints, Honoring Time and Speed Constraints. Vienna, Austria: Proceedings of GIScience. 23-26 Sept 2014; 2014.Google Scholar
  201. Fieberg J, Matthiopoulos J, Hebblewhite M, Boyce MS, Frair JL. Correlation and studies of habitat selection: problem, red herring or opportunity? Phil Trans Biol Sci. 2010;365(1550):2233–44.Google Scholar
  202. Tobler WA. A Computer Movie Simulating Urban Growth in the Detroit Region. Economic Geography. 1970;46(2):234–40.Google Scholar
  203. Fotheringham AS, Brunsdon C, Charlton M. Geographically Weighted Regression: the Analysis of Spatially Varying Relationships. Chichester: Wiley; 2002.Google Scholar
  204. Demšar U, Harris P, Brunsdon C, Fotheringham AS, McLoone S. Principal Component Analysis on Spatial Data: a Review. Annals of Association of American Geographers. 2013;103(1):106–28.Google Scholar
  205. Fleming CH, Calabrese JM, Mueller T, Olson KA, Leimgruber P, Fagan WF. From Fine-Scale Foraging to Home Ranges: A Semivariance Approach to Identifying Movement Modes across Spatiotemporal Scales. Am Nat. 2014;183:E154–67.Google Scholar
  206. Sarkar D, Chapman CA, Griffin L, Sengupta R. Analyzing Animal Movement Characteristics From Location Data. Trans GIS: Advanced Online Publication. 2014.Google Scholar
  207. Ye Y, Zheng Y, Chen Y. Mining individual life pattern based on location history. In: Mobile data management: systems, services and middleware. Taipei, Taiwan: IEEE MDM’09, 18–20 May 2009; 2009. p. 1–10.
  208. Rodrigues A, Damásio C, Cunha JE. Using GPS Logs to Identify Agronomical Activities, Connecting a Digital Europe Through Location and Place. Lecture Notes in Geoinformation and Cartography. 2014;2014:105–21.Google Scholar
  209. Umair M, Kim WS, Choi BC, Jung SY. Discovering personal places from location traces. Bongpyong-myun, Pyeongchang, South Korea: Proceedings of the 16th International Conference on Advanced Communication Technology (ICACT 2014), 16–19 Feb. 2014; 2014. p. 709–13.Google Scholar
  210. Zhou C, Frankowski D, Ludford PJ, Shekhar S, Terveen LG. Discovering personally meaningful places: An interactive clustering approach. ACM Transactions on Information Systems. 2007;25:3.Google Scholar
  211. Huang W, Li M, Hu W, Song G, Xing X, Xie K. Cost sensitive GPS-based activity recognition, vol. 2013. Shenyang, China: Proceedings of the 10th International Conference on Fuzzy Systems and Knowledge Discovery (FSKD) 23-25 July 2013; 2013. p. 962–6.Google Scholar
  212. Siła-Nowicka K, Oshan T, Vandrol J, Fotheringham S. The Use of Volunteered Geographic Information for Spatial Interaction Modelling. Vienna, Austria: Proceedings of GIScience. 23-26 Sept 2014; 2014.Google Scholar
  213. Delafontaine M, Versichele M, Neutens T, Van de Weghe N. Analysing spatiotemporal sequences in Bluetooth tracking data. Applied Geography. 2012;34:659–68.Google Scholar
  214. Versichele M, Neutens T, Delafontaine M, Van de Weghe N. The use of Bluetooth for analysing spatiotemporal dynamics of human movement at mass events: A case study of the Ghent Festivities. Applied Geography. 2012;32(2):208–20.Google Scholar
  215. Versichele M, Neutens T, Goudeseune S, Van Bossche F, Van de Weghe N. Mobile Mapping of Sporting Event Spectators Using Bluetooth Sensors. Sensors. 2012;12(10):14196–213.Google Scholar
  216. Versichele M, De Groote L, Claeys Bouuaert M, Neutens T, Moerman I, Van de Weghe N. Pattern mining in tourist attraction visits through association rule learning on Bluetooth tracking data: A case study of Ghent, Belgium. Tour Manage. 2014;44:67–81.Google Scholar
  217. Bonné B, Barzan A, Quax P, Lamotte W. WiFiPi: Involuntary tracking of visitors at mass events. Madrid, Spain: 14th International Symposium on “A World of Wireless, Mobile and Multimedia Networks” (WoWMoM), 4-7 June 2013; 2013.Google Scholar
  218. Ji W, White PCL, Clout MN. Contact rates between possums revealed by proximity data loggers. J Appl Ecol. 2005;42:595–604.Google Scholar
  219. Prange S, Jordan T, Hunter C, Gehrt SD. New radiocollars for the detection of proximity among individuals. Wildl Soc Bull. 2006;34(5):1333–44.Google Scholar
  220. Rutz C, Burns ZT, James R, Ismar SM, Burt J, Otis B, et al. Automated mapping of social networks in wild birds. Curr Biol. 2012;22(17):R669–71.Google Scholar
  221. Krause J, Krause S, Arlinghaus R, Psorakis I, Roberts S, Rutz C. Reality mining of animal social systems. Trends Ecol Evol. 2013;28(9):541–51.Google Scholar
  222. Demšar U, Fotheringham AS, Charlton M. Exploring the spatio-temporal dynamics of geographical processes with Geographically Weighted Regression and Geovisual Analytics. Information Visualization. 2008;7:181–97.Google Scholar
  223. Foley P, Demšar U. Using geovisual analytics to compare the performance of Geographically Weighted Discriminant Analysis versus its global counterpart, Linear Discriminant Analysis. Int J Geogr Inform Sci. 2013;27(4):633–61.Google Scholar
  224. Purves R, Laube P, Buchin M, Speckmann B. Moving beyond the point: An agenda for research in movement analysis with real data. Comput Environ Urban Syst. 2014;47:1–4.Google Scholar
  225. Laube P. Computational Movement Analysis. Springer Briefs in Computer Science. Berlin-Heidelberg: Springer-Verlag; 2014.Google Scholar

Copyright

© Demšar et al.; licensee BioMed Central. 2015

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Advertisement