Department of Science and Technology Institutionen för teknik och naturvetenskap Linköping University Linköpings Universitet SE-601 74 Norrköping, Sweden 601 74 Norrköping

LiU-ITN-TEK-A--09/012--SE

Visualization of Weather andShip Data

Oskar EureniusTobias Heldring

2009-03-18

LiU-ITN-TEK-A--09/012--SE

Visualization of Weather andShip Data

Examensarbete utfört i vetenskaplig visualisering vid Tekniska Högskolan vid

Linköpings universitet

Oskar EureniusTobias Heldring

Handledare Patrik LundbladHandledare Mikael JernExaminator Mikael Jern

Norrköping 2009-03-18

Upphovsrätt

Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –under en längre tid från publiceringsdatum under förutsättning att inga extra-ordinära omständigheter uppstår.

Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat förickekommersiell forskning och för undervisning. Överföring av upphovsrättenvid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning avdokumentet kräver upphovsmannens medgivande. För att garantera äktheten,säkerheten och tillgängligheten finns det lösningar av teknisk och administrativart.

Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman iden omfattning som god sed kräver vid användning av dokumentet på ovanbeskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådanform eller i sådant sammanhang som är kränkande för upphovsmannens litteräraeller konstnärliga anseende eller egenart.

För ytterligare information om Linköping University Electronic Press seförlagets hemsida http://www.ep.liu.se/

Copyright

The publishers will keep this document online on the Internet - or its possiblereplacement - for a considerable time from the date of publication barringexceptional circumstances.

The online availability of the document implies a permanent permission foranyone to read, to download, to print out single copies for your own use and touse it unchanged for any non-commercial research and educational purpose.Subsequent transfers of copyright cannot revoke this permission. All other usesof the document are conditional on the consent of the copyright owner. Thepublisher has taken technical and administrative measures to assure authenticity,security and accessibility.

According to intellectual property law the author has the right to bementioned when his/her work is accessed as described above and to be protectedagainst infringement.

For additional information about the Linköping University Electronic Pressand its procedures for publication and for assurance of document integrity,please refer to its WWW home page: http://www.ep.liu.se/

© Oskar Eurenius, Tobias Heldring

Abstract

This paper focus on the development of a tool for Ship and Weather Information Monitor-

ing (SWIM) visualizing weather data combined with data from ship voyages. The project

was done in close collaboration with the Swedish Meteorological and Hydrological Institute

(SMHI) who also evaluated the result. The goal was to implement a tool which will help

shipping companies to monitor their fleet and the weather development along planned routes

and provide support for decisions regarding route choice and to evade hazard. A qualitative

usability study was performed to gather insight about usability issues and to aid future de-

velopment. Overall the result of the study was positive and the users felt that the tool would

aid them in the daily work.

ii

Acknowledgements

A special thanks to SMHI Global Product Manager Lennart Cederberg and NCVA [1] Professor

Mikael Jern.

iii

Contents

1 Introduction 11.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1.1 Swedish Meteorological and Hydrological Institute . . . . . . . . . 11.1.2 Weather Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 SWIM 52.1 Development Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62.2 GeoAnalytics Visualization (GAV) . . . . . . . . . . . . . . . . . . . . . . 62.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.3.1 Weather . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62.3.2 Voyage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.4 Visual Representation and Interaction . . . . . . . . . . . . . . . . . . . . 72.4.1 World Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.4.2 Parallel Coordinates Plot . . . . . . . . . . . . . . . . . . . . . . . 92.4.3 Voyage Time Graph . . . . . . . . . . . . . . . . . . . . . . . . . . 102.4.4 Linked Views . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.5 Task Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3 Results 143.1 Usability Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.1.1 Participants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143.1.2 Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143.1.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4 Discussion 174.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.2.1 Forecast Certainty . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

Bibliography 19

List of Figures 21

i

1 Introduction

In this report chapter 1 will present background information to the thesis subject areatogether with an introduction to SMHI and the weather parameters used. This is followedby a description of the purpose of the thesis work along with a review of related work.Chapter 2 presents an overview of the usage of SWIM, describes the tools used duringdevelopment, introduces the data sources and how the data is handled as well as adetailed description of the visual components in SWIM and how the interaction is linkedbetween them. In chapter 3 the results are presented in form of a usability study. Lastly,chapter 4 consists of conclusions drawn and presents ideas for further work.

1.1 Background

Visualizing multiple data sets from different sources represents a major challenge forthe future. Using data fusion [2], [3] to visualize integrated spatial and temporal datait is possible to gain insight and retrieve information that would otherwise have to beobtained using much more time consuming methods.

One subject area of particular interest is the shipping industry where the need ofbeing able to draw accurate conclusions and make well planned decisions regarding routesand weather factors are important. There are a large number of variables affecting thesuccess of every sea voyage performed by vessels such as cargo ships. Important aspectsbeing considered are speed, safety and costs because of their direct impact on the voyageperformance. Weather factors are the main concern when considering these aspects andtherefore shipping companies consult meteorologists about weather information and routeoptimization.

1.1.1 Swedish Meteorological and Hydrological Institute

SMHI’s mission is to manage and develop information on weather, water and climatethat provides knowledge and advanced decision-making data for public services, the pri-vate sector and the general public. SMHI is a government agency under the Ministryof the Environmental and offers products and services that provides various kinds ofenterprises and organizations with an important foundation for decision-making wheregeneral forecasts and weather warnings, industry-specific services, simulations and ana-lyzes, statistics, climate studies and contracted research are some examples [4].

During our work we have been positioned at the Business and Media Services Depart-ment which markets and produces customized and industry-specific forecasting and dataservices aimed at customers in spheres such as media, industry and commerce, agricul-ture, the private market, shipping, land-based transport, energy and the building trade

1

1.1. BACKGROUND 2

and property management.The Shipping Department at the Business and Media Services Department uses a

weather routing system to monitor voyages according to weather criteria, voyage dataand ship performance to support captains decision-making regarding route optimizationand harsh weather. This weather routing system enables shipping companies with theirrespective captains to receive updated weather forecast information and guidance. In sit-uations when shipping companies prefer to monitor their voyages themselves a web-basedsystem called FleetWeb is used which provides similar, but more limited, functionalityas the weather routing system.

1.1.2 Weather Introduction

Weather parameters are not always easy interpreted but important aspects to understandand therefore follows a short description of the most common parameters used in SWIM.

Waves

The wave height parameter is separated into three different types: significant wave height,wind wave height and swell height. Significant wave height is the average wave height,trough to crest, of the one-third largest waves and was intended to mathematically ex-press the height estimated by a trained observer. It is today commonly used as a measureof the height of ocean waves. Wind wave height is the wave height produced by the localwind and swell height is the wave height generated by earlier formations of wind systemsin different geographical locations. The wave period parameter and wave direction pa-rameter are also categorized according to significant waves, wind waves and swell wherethe period is the unit time between two crests and the direction the path the wave travels.

Wave spectral kurtosis or freak wave index is a wave parameter describing the proba-bility of the generation of waves with extreme heights compared to the mean wave heightin that area. In probability theory and statistics, kurtosis is a measure of the ”peaked-ness” of the probability distribution of a real-valued random variable, in this case thesignificant wave height.

Wind

The wind parameter is commonly classified by its spatial scale, wind speed and thegeographic regions in which it occur. Wind direction is reported by the direction fromwhich it originates, e.g. a northerly wind blows from the north to the south. Windspeed is the movement of air in an atmosphere and is a scalar quantity, the magnitudeof the vector of motion (figure 1.1). Wind measured at level 89 of an isobaric scale,approximately 40 meters above sea level, is a suitable wind parameter when evaluatingshipping conditions.

Figure 1.1: Wind speed and direction is usually represented in visualizations bywind flags in unit knots.

1.2. PURPOSE 3

Air Pressure

The force per unit area exerted against a surface by the weight of air above that surfaceat any given point in the earth’s atmosphere is a common definition of atmosphericpressure. Pressure decreases with increasing elevation where low pressure areas have lessatmospheric mass above their location and pressure increases with decreasing elevationwhen high pressure areas have more atmospheric mass above their location.

Air pressure measured at sea level is referred to as the mean sea level pressure (MSLP)and is the pressure normally given in weather reports. Using MSLP, pressure consideredhigh pressure or low pressure do not depend on geographical location and the reductionto sea level normalizes fluctuations in pressure which makes isobars on weather mapsmeaningful and useful tools.

Temperature

The temperature parameter suitable when evaluating shipping conditions is the 2 metertemperature, measured two meters above sea level, which is given in Celsius.

Cloud Cover

Cloud cover refers to the fraction of the sky obscured by clouds when observed from aparticular location and its parameter is given in unit percent.

Precipitation

Precipitation is any product of the condensation of atmospheric water vapor that isdeposited on the earth’s surface. The precipitation parameter describes either rain orsnow in their corresponding height of water measured in meters.

1.2 Purpose

FleetWeb suffers from design flaws such as low interactivity, cluttered displays, unstableinitialization and lack of filtering possibilities. SWIM was created as a prototype for afuture version of FleetWeb to aid shipping companies monitoring vessels according toforecasted weather development along planned routes (figure 2.1). The tool providessupport for decisions regarding route choice with respect to future weather developmentand warnings of extreme weather factors. With weather forecasts and voyage informa-tion provided by the European Centre for Medium-Range Weather Forecasts (ECMWF)[5] and SMHI [4] shipping companies can monitor fleets using highly interactive visualrepresentations visualizing weather parameters and detailed voyage information. Usinga geographic map visualization together with a parallel coordinates plot (PCP) bringsgreat possibilities when exploring the data.

SWIM was designed with the following tasks in mind:

• Detailed monitoring of voyages based on planned and reported waypoints.

• Easy exploration of weather forecast according to both geographic positions andalong routes.

1.3. RELATED WORK 4

• Using a PCP together with a geographic map visualization to find interesting voy-ages according to weather parameters.

• Using highly interactive linked visual representations to facilitate exploration ofdata.

1.3 Related Work

Visualization of spatio-temporal data has been the subject of several recent researchpapers, see [6] for an overview. GeoVista Studio [7] and CommonGIS [8] are general sys-tems which support exploratory data analysis with decision-making and Andrienko andAndrienko has illustrated other motivating approaches in earlier papers [9], [10]. Accord-ing to Jern and Franzn most of these systems lack support for analyzing simultaneouslymultiple attributes data and spatio-temporal behavior [11] and as a response a generic”GeoAnalytics” visualization (GAV) [12] toolkit is introduced. Jern et al. emphasize theadvantages and need of multiple linked views (MLVs) for dynamically exploring time-varying, geographically referenced and multivariate attribute data. Parallel coordinates[13], [14], [15], [16] with embedded visual inquiry methods that serve as a visual con-trol panel for dynamically linked and coordinated views is a well-known visualizationtechnique for multivariate data [17].

In VISPER [18] multi-dimensional, multi source, time-varying and geospatial infor-mation from voyage analysis is represented to facilitate decision-making. Therefore Lund-blad et al. propose the use of common InfoViz visual representations for multivariatedata together with geographic mapping to aid users in their work examining voyageperformance.

Roberts [19] provides a review of multiple linked-view tools, methodologies and mod-els, discusses related challenges and ideas, and provides some rudiments for coordinationwithin a geovisualization context. In accordance with Roberts linking and relating infor-mation in one view to that of other views will assist the user in the exploration processand may provide additional insight into the underlying information. Operations such asfiltering, dynamic queries [20] and selections applied simultaneously in different viewswhen affecting not only the same information but more effectively collections of differentdata is pointed out by Roberts as more interesting and an important aspect to consider.

Spirkovska and Lodha presented AWE - Aviation Weather Data Visualization En-vironment [21] which was designed to meet an urgent need of critical weather elementvisualizations. AWE focuses on interactive graphical displays of these weather elementssuch as, meteorological observations and terminal area forecasts and maps them onto acartographic grid specific to the pilot’s interest.

Song et al have designed an integrated atmospheric visual analysis and explorationsystem for interactive analysis of weather data sets [22]. These tools provide meteorolo-gists with new abilities to analyze their data and answer questions on regions of interests,ranging from physics-based atmospheric rendering to illustrative rendering containingparticles and glyphs.

In accordance with Dorneich et al. challenges lie in effectively integrating voyage andweather information in the same tool to facilitate decision-making [23].

2 SWIM

SWIM is developed using the GeoAnalytics Visualization (GAV) Framework [24] and isa tool shipping companies can use to monitor voyages and receive up to date weatherforecasts corresponding to their routes.

SWIM will on startup automatically find the latest weather forecast delivered bythe ECMWF. In SWIM the user is presented with three different modes accessible usingtabs. In the first mode a selection of static weather parameters are presented in a weatherparameter menu sorted according to priority level, e.g. basic weather, advanced weatherand extra weather (figure 2.1). The second mode presents a searchable list menu withall active voyages retrieved from the SMHI database. In the third mode focus lies on oneor more selected voyages with detailed information presented.

Figure 2.1: The SWIM overview. In the world map ships are visualized usingglyphs and their routes are plotted as lines. Significant wave height is displayedusing an iso-surface where warmer color indicates higher waves. The current timestep can be changed using the time slider positioned underneath the world map.Using the weather parameter menu to the left the user can select which parameterto visualize and which representation to use. At the bottom weather parametersare visualized in the parallel coordinates plot.

5

2.1. DEVELOPMENT PROCESS 6

2.1 Development Process

The first part of the thesis work was devoted to creating a reader for weather data.Developing a flexible reader for the binary weather data (described in section 2.3.1) wastime consuming yet rewarding later on. With the ability to read and store weatherparameter data focus shifted to develop map visualizations such as iso-surfaces, iso-lines,wind flags, arrow glyphs, number glyphs and freak wave glyphs. During the next stepwork was put into expanding the underlying data flow to store data over time and usethis to enable the visualization of time dimension in the world map view. Next a readerfor voyage data (section 2.3.2) was created followed by map visualizations of ships androutes. With the voyage data available weather data could be linked to voyages andthe voyage time graph was implemented followed by the PCP. For the last part of thethesis work focus where on data flow and linking between views, interaction and interfacedesign.

2.2 GeoAnalytics Visualization (GAV)

GAV [24] is a component toolkit for dynamically exploring time-varying, geographicallyreferenced and multivariate attributes simultaneously and to enable the capture of theinteractive visual process into information packages that allow the analysts to commu-nicate their discovery and decision recommendations. GAV includes components basedon a synergy of technologies from information visualization (InfoViz), geovisualization(GeoViz) and scientific visualization (SciViz). The GAV components are developed inC# based on Microsoft’s low-level DirectX graphics library and fulfill many generic re-quirements for a Visual Analytics application design framework.

2.3 Data

2.3.1 Weather

The input weather data used by SWIM is in GRIB (Gridded Binary) [25] format whichis a mathematically concise data format commonly used in meteorology to store weatherdata. It is developed and standardized by the World Meteorological Organization WMO.The GRIB format is used world-wide by meteorological centers for numerical weatherprediction output and serves as an efficient solution of transferring large volumes ofdigital gridded weather data.

In our work focus lies on GRIB files originated from the European Centre for Medium-Range Weather Forecasts (ECMWF) [5]. The first part of our thesis work consisted ofcreating a reader for GRIB files. By methodically reading segment by segment using theGRIB API [26] as reference, the GRIB reader developed from a simple read and printfunctionality to a fully developed binary reader being able to collect and store data forspecific parameters as well as supporting different types of GRIB files.

In a GRIB file the parameters are positioned sequentially in a binary data stream.Each parameter segment consists of the standard GRIB file fields; including a parameterdescription, measurement level, grid description and the actual data. The GRIB filesused by SWIM are of data grid resolution 1,5 degrees in both latitude and longitude

2.4. VISUAL REPRESENTATION AND INTERACTION 7

which results in 240 horizontal and 120 vertical points summed up to a total of 29040data points in the world map.

SWIM allows for exploration of weather forecasts over a ten days period. Each GRIBfile consists of a selection of weather parameters with corresponding weather data for acertain date and time. SWIM are using a total of 31 GRIB files with a forecast timeinterval of 6 hours the first 5 days and an interval of 12 hours the later 5 days. When aweather parameter is selected in the weather parameter menu the parameter is locatedand read from each of the 31 GRIB files and saved into a data array which enablesthe time dimension to be explored. Because of the large amount of binary data beinginterpreted during reading the process has been assigned its own thread in the systemwhich enables simultaneously interaction with SWIM.

2.3.2 Voyage

Voyage data is collected from a database provided by SMHI where extensive informationregarding ships and their corresponding routes are available. During start up SWIMautomatically reads all currently active voyages, which allow ships and routes to bevisualized in the world map.

Data retrieved for each voyage includes ship information such as operator, ship nameand ship type. Voyage specific information includes estimated time of departure and es-timated time of arrival, port of departure and port of arrival as well as a list of waypointpositions presenting the planned route of the voyage. Lastly, reports delivered by ac-tive are retrieved including current positions, speed and performance as well as weatherobservations at reported positions.

Reported positions are combined with planned waypoints to form a list, with latitudeand longitude positions describing voyage routes, used to visualize voyages routes inthe world map view. The position of ships at different time steps are estimated viainterpolation where the time of last reported position, the time at selected time step andthe estimated time of arrival are used to estimate the distance traveled between the lastreported position and the position of the arrival port.

2.4 Visual Representation and Interaction

In SWIM the users are given the possibility to study weather data cohering with voy-age information in different views. A view represents a visual representation displayingselectable data with possibilities of interaction and filtering.

2.4.1 World Map

The main view is the world map where all weather parameters originating from cor-responding GRIB files can be presented according to selection in the weather menu.Options are provided on how to visualize a parameter, e.g. iso-line, iso-surface, glyphor numerical values. Routes with their corresponding ships are also visualized in theworld map. The ships are represented by ship glyphs which are positioned and rotatedaccording to an interpolation based on their pre-planned waypoints, reported positionsand present time (figure 2.2). For these features to be useful it is necessary to look atnot only the spatial variables but also temporal aspects. SWIM reads and store a tendays forecast consisting of 31 binary GRIB files which enables the temporal aspect to be

2.4. VISUAL REPRESENTATION AND INTERACTION 8

Figure 2.2: World map view visualizing significant wave height and wind. The sig-nificant wave height is visualized using an iso-surface where warmer color indicateshigher waves. The wind is represented by wind arrow glyphs where the orientationshows the wind direction and the glyph shows the wind strength. Ships are shownusing a ship glyph and their routes are drawn with lines.

taken into account and is represented by a forecast time slider. This slider is positionedat the first forecast prediction and with a sliding interval of six hours the first five daysand a sliding interval of twelve hours the later five days it effectively enables the userto view weather changes and ship movement over time (figure 2.3). Routes with corre-sponding ships that at a specific time index has reached its final destination is markedas inactive and grayed out. The world map view with selected weather data visualizedand chosen voyages plotted will work as an effective support to monitor several vesselssimultaneously over a period of time.

To facilitate usage of the world map view and maximize performance a series ofgraphics optimizations has been implemented. With a resolution of 240 longitude and120 latitude grid points it is necessary to use a level of detail functionality to keep theworld map useful and not cluttered when changing the map zoom. With a maximum of29 040 data points on the data grid we use eight levels of detail based on a mean valuecalculation with respect to all interesting neighbors of each visible grid point (figure 2.4).

Another important optimization feature is the active rectangular window we use tokeep track of the current visible part of the world map. Using this functionality gives aperformance advantage where weather parameter visualizations are only calculated andrendered if inside the active rectangular window. When zoomed in this optimizationsaves computational power and speeds up the interaction.

2.4. VISUAL REPRESENTATION AND INTERACTION 9

Figure 2.3: The left side shows wind wave height as an iso-surface, ships visualizedusing ship glyphs and their routes as lines. The right side shows the same map areawith the time slider advanced 48 hours.

Figure 2.4: The left side shows significant wave height as an iso-surface and meanwave direction as a regularly spaced arrow glyph. The right side shows the greaterdetail of the glyph representation received when zoomed in.

2.4.2 Parallel Coordinates Plot

General Description

PCPs consists of N equidistant vertical axes which corresponds to the selected variables(attributes). The axes are scaled to the [minimum, maximum] - range of the correspond-ing attribute and every data item corresponds to a line which intersects each of the axesat the point which corresponds to the value for the attribute. The main purpose ofPCPs is to view relations in multivariate data and suitable data types supported arequantitative data and ordinal data. Highly interactive visualization applications can bedeveloped using PCPs with techniques such as; brushing, highlighting (picking), dynamicfiltering and color coding.

Implementation

The PCP view displays all the voyages with loaded parameters for the selected time step.Each voyage is represented as a line in the PCP where each axis corresponds to a specific

2.4. VISUAL REPRESENTATION AND INTERACTION 10

weather parameter (figure 2.5). This view gives the user an overview of the weather forall the voyages at a specific time step. Advancing the time slider will update the PCPwith new data, namely weather data for each voyage at the new time step. Lines arecolored according to the parameter value of the selected axis. Which parameter to usefor coloring can be chosen by clicking the header the parameter axes. The colors of thelines are linked to the color of the ships in the world map view.

Figure 2.5: Time step parallel coordinates plot. The lines can be selected, filtered bymoving the sliders on each axis and colored by clicking the header of the parameter.

2.4.3 Voyage Time Graph

General Description

Time graphs consists of N equidistant vertical axes which corresponds to sequential timesteps in a given interval. Each attribute item is represented by a line which intersectseach of the time axes at a point corresponding to the value at that time step. Using timegraphs it is easy to visualize multiple attributes over time and observe temporal changesin data.

Implementation

When in the voyage information mode the voyage time graph gives the user an overviewof how the weather will develop along the planned route of the selected voyage withoutthe need of moving the time slider. Each parameter is represented as a line in thevoyage time graph and each vertical axis represents a different time index accordingto additional hours added to the forecast creation date (figure 6). Each parameter isnormalized according to predefined values in order to enable the different parameters tobe plotted together. Straight lines near the bottom of the plot indicate stable and calmweather along the route for the forecast period. Curvy lines however indicate varyingweather conditions and high peaks indicate extreme weather parameter values.

2.4.4 Linked Views

At start up all voyages are displayed in the world map which provides an overview ofroutes and ship positions. The ability to highlight one or several voyages is importantwhen exploring data and is implemented. In the world map view all the voyages exceptthose highlighted will be slightly transparent preventing interference with highlightedvoyages (figure 2.7). In the PCP lines of highlighted voyages will be selected. In the

2.4. VISUAL REPRESENTATION AND INTERACTION 11

Figure 2.6: Ship time graph visualizing weather parameters along the selectedroute. Each line represents a weather parameter where the x axes represents timeand the y axes represents parameter value. The static coloring of the lines arepredefined for each weather parameter.

voyage information tab text information will be presented for highlighted voyages andthe voyage time graph will display data for one of the highlighted voyages.

Figure 2.7: Highlighting in the World Map view. The left side shows the mapbefore selection and the right side after.

Different approaches for how to highlight voyages are presented to meet different userneeds. An interesting voyage can for example be selected by clicking on it in the worldmap. To highlight a voyage according to name the ship list with its search function canbe used.

The PCP which presents weather for all voyages for a specific time step can be usedto spot voyages experiencing harsh weather conditions and by clicking on a line thecorresponding voyage will be highlighted.

When a specific geographic area is of interest the selection area tool can be used.When the selection area tool is activated the user is presented with a circular area on theworld map positioned according to mouse position. The size of the selection area can bemanually set and when clicking a position in the world map all voyages traveling throughthat area at any point during the forecast period will be highlighted (figure 2.8).

Hiding uninteresting voyages is achieved by using the filter sliders in the PCP. Filteredlines will be completely hidden as well as the corresponding voyages in the world mapview. For example, the user can adjust the filters to only show voyages whose shipsare exposed to a significant wave height above 5 meters (figure 2.9). When advancingthe time slider the filter value is saved and only voyages whose ships are experiencing asignificant wave height above 5 meters will be displayed at each time step.

2.5. TASK FLOW 12

Figure 2.8: Selection area tool. The left side show the world map view with shipsand routes where the grey circle is the selection area tool. When using the selectionarea tool ships that travel through the selected area at any time during the forecastperiod will be highlighted as shown on the right side.

Figure 2.9: Filtering using time step parallel coordinates plot. In the world mapsignificant wave height is visualized using an iso-surface. The left side shows parallelcoordinates plot with corresponding map before filtering. The right side showsfiltering of ships according to significant wave height.

2.5 Task Flow

A typical usage scenario in SWIM could be that the user is interested in ships experiencingharsh weather in the Atlantic outside of the English Channel.

At startup the user will be presented with all voyages from the first time step displayedon the world map and the PCP will be empty since no weather data has been read. Thefirst user action would be to read weather data of interest which for example could besignificant wave height and wind. If the weather data is selected using the weather

2.5. TASK FLOW 13

menu data will be visualized on the map and displayed in the PCP. Focusing on shipsexperiencing harsh weather would be done by filtering out ships according to weathercriteria using the PCP where ships not fulfilling the selected weather criteria disappearsfrom the map. To further narrow the search according to geographic area the selectionarea tool would be used to highlight the area outside the English Channel. Ships onthe world map not travelling through the selected geographical area during the forecastperiod will be grayed out. Detailed information about the filtered ships experiencingharsh weather in the Atlantic outside of the English Channel can now be retrieved byclicking on each preferred ship.

Figure 2.10: Which ships will experience harsh weather in the Atlantic outside theEnglish Channel? A typical usage scenario in SWIM described using a task flow.

3 Results

Since the product of our thesis work is a software tool we chose to evaluate the result usinga usability study. Allowing users to give feedback on usability, design and techniques,conclusions could be drawn about both individual parts and the complete work.

3.1 Usability Evaluation

The purpose of the evaluation was to explore qualitative usability issues with respect tovisual representation and interactive representation [27] and to gather opinions regardingpotential use and future development without intentions to do measurements.

3.1.1 Participants

A selection of marine meteorologists and master students from the Media Technology andEngineering program [28] formed a group of six participants. The marine meteorologistshad knowledge and experience in meteorology and ship routing and the MSc studentshad knowledge in information visualization, usability and design. Their occupationsand respective domain knowledge was assumed to give insightful information regardingusability issues of SWIM.

3.1.2 Procedure

Each session lasted nearly an hour. First the participant was given a demonstration ofthe tool and its functionalities. This was done to minimize the risk of misinterpretationas well as first time user misses since the intended users are a small number of peopleusing the tool regularly. For the same reason the participants were given answers andexplanations to all their questions about SWIM and its functionality during the test.

After the demonstration the participant was given the control of the tool and wasencouraged to explore some of the functionality earlier demonstrated. A number oftasks that was prepared beforehand were presented to the user in random order. Theintention was not to evaluate the participant’s success of the specific tasks but simply toguide the user and encourage exploration. Throughout the evaluation a guide sat besidethe participant actively discussing thoughts and opinions regarding the tool and a thirdperson documented the important parts of the discussion.

The last part of the evaluation consisted of a general discussion based around thecriteria stated by Freitas et al. [27].

Visual Representations were discussed according to the following selection of criteria:Data density, limitations, visual/spatial orientation, data dimensions, relevant informa-tion, reference context, occlusion, display of detail and logical order.

14

3.1. USABILITY EVALUATION 15

Interactive Representations were discussed according to the following selection ofcriteria: filtering, selection of objects, search and querying, control of level of detail andviewpoint manipulation.

3.1.3 Results

The marine meteorologists had limited previous experience working with visualizationtools in comparison with the MSc students which were skilled using these tools. On theother hand the marine meteorologists had significant experience with weather data androute optimization tools and these differences were observed when studying the resultsof the usability evaluation.

All participants completed the prepared tasks without major complications and theoverall opinions describe SWIM as a tool with many possibilities. Direct data manipu-lation and filtering based on dynamic queries immediately changing the displayed viewand additional linked views was greatly appreciated. All participants and especially themarine meteorologists described the possibility to filter voyages in the world map viewaccording to criteria in the PCP as useful. During the evaluation the participants alsopointed out both known and previously unknown issues and discussions about certaincriteria regarding visual and interactive representation brought to light strengths andweaknesses of the tool.

With regard to the visual representations, participants experienced the built in level-of-detail functionality in the map view as smooth and natural. However, comments wheremade about not being able to connect weather parameters to a specific route instead ofregularly spaced grid-points. All found the spatial organization of visual representationsas well thought-out and easy to interpret. The ability to choose different visualizationmethods for certain weather parameters in the world map and therefore being able tocombine different representations was pointed out as well designed and important. Themarine meteorologists requested functionality where the user would be automaticallywarned about voyages whose ships are affected by extreme weather situations ratherthan having to gain that insight during manual exploration. All participants agreed thatthe information displayed in the map view was relevant but additional information wasrequested in specific situations such as more detailed voyage information in the ship tooltip.

With regard to the interactive representation a difference was noticed between themarine meteorologists and the MSc students in discovering interaction possibilities. TheMSc students which are skilled using visualization techniques based on brushing, pickingand filtering found it easier to discover and use these features. The marine meteorologistsrequested additional descriptions of the visual representations. However, even thoughthe chosen target group is a small number of people using the tool regularly we foundit possible for people without previous experience of similar techniques to embrace thefundamental ideas with a short introduction.

All participants pointed out the difficulty of comparing weather values for differenttime steps. One participant suggested maximum and minimum axis values of the TimeStep Coordinates Plot spanning over the entire forecast period as opposed to being timestep specific. Participants also described the variety of selection methods to highlight aunique voyage as well implemented and the different approaches as good complementsto each other. The Selection Area tool was especially appreciated by the marine meteo-

3.1. USABILITY EVALUATION 16

rologists who saw advantages with the tool enabling exploration of geographical areas.The voyage time graph was described by all as a powerful technique to quickly gain

insight of weather development along a specific route. Comments were however expressedregarding carefully choosing normalization values for each parameter to avoid misinter-pretation. Participants also requested an explanation of the colors of weather lines inthe voyage time graph. The time slider was described by all participants as natural andappeasing to use. A more distinct representation of elapsed hours of the weather forecastwas pointed out as a valuable addition. When selecting a single voyage requests about ago-to-ship functionality was expressed, automatically focusing and positioning the viewover the ship corresponding to the selected voyage in the world map.

Suggestions of improvements and additions to the GUI were given. One of those, afeature pointed out missing by the majority of the participants, was the ability to undosteps in the interaction.

4 Discussion

4.1 Conclusions

This report presents SWIM which is a tool to explore multi-source, time-varying andgeospatial data. SWIM allows weather forecast data and voyage information to be an-alyzed using an interactive map view linked to a PCP and a time graph. Using SWIManalysts at shipping companies can monitor ship voyages and weather development alongplanned routes. Voyages whose ships are being exposed to harsh weather can easily bebrought to focus and detailed route specific weather and voyage information can beretrieved.

A qualitative usability evaluation was performed with a selection of six participants.Discussing and exploring the tool with each participant resulted in useful feedback re-garding functionality and design issues. All participants completed the prepared taskswithout major complications and the overall opinions describes SWIM as a tool withmany advantages and interesting functionalities compared to similar tools in the samedomain.

In summary, especially appreciated features were:

• The world map view with interaction possibilities such as highlighting, smoothzooming, brushing and weather parameter visualization options.

• The PCP with the possibility to filter and distinguish voyages according to weathercriteria, presenting results immediately in the world map view.

• The voyage time graph which gives an overview of weather development alongplanned routes of selected voyages.

• The ability to view and interact with the geospatial data in linked views withseveral highlighting possibilities.

Using a PCP proved to be a powerful tool to filter and explore data in the world mapview. Discovering voyages which fulfills certain weather criteria is an important featurewhich is made possible using the filtering and picking techniques of the PCP. The synergybetween geovisualization (map) and information visualization (PCP) was demonstratedand evaluated as useful.

The both known and previously unknown issues regarding visual and interactiverepresentation that brought to light strengths and weaknesses during the evaluation willconstitute the foundation for further development of SWIM.

17

4.2. FUTURE WORK 18

4.2 Future Work

For future development of SWIM more automation could be built into the tool. The toolcould present a list with voyages meeting certain criteria for harsh weather conditions.Using this feature investigation of risky voyages could start immediately without theneed of first filtering according to criteria. To easily compare weather values of differenttime steps for all voyages a second time graph could be implemented where weatherdevelopment over time could be shown displaying one weather parameter and all or aselection of voyages. This would facilitate discovering of voyages experiencing the mostextreme weather parameter value during the entire forecast period.

The ability to use presets of user customizations, filtering criteria and standard pa-rameters to read would increase the dynamics and speed when using the tool.

Shipping companies regularly evaluates the performance of their ships by evaluatingreports along the routes, weather factors and other variables of interest to determineif the voyage was performed as planned [18]. To facilitate this work SWIM could beextended to support post voyage analysis. Using the time slider to view past time stepsand presenting reported data and way point specific weather SWIM could be a usefultool for this purpose.

4.2.1 Forecast Certainty

Weather forecasts always come with a bit of uncertainty. The further into the future,the higher level of uncertainty. This is natural since exactly predicting and simulatingthe earth’s atmosphere is an extremely difficult task. Applications similar to SWIM haveused only data from the latest available forecast which of course provide the highest levelsof accuracy but still produces errors, especially with long predictions.

A time-lagged forecast [29] is the result of weighing earlier forecasts according tospecified values and using these lagged forecasts the user would be able to estimate theaccuracy of the current forecast. This lagged forecast would behave with inertia becauseof its dependence on earlier forecasts and would often be a bit too conservative whenevaluating the forecast. However, comparing the lagged forecast with the latest forecastwould give the user a sense of the certainty of the forecast. If the lagged forecast presentsimilar values as the latest single forecast it indicates that the prediction is reliable. Onthe other hand, if the two values differ greatly it is an indication of uncertainty in theforecast.

This feature would fulfill the need of a complementary indicator describing the fore-cast accuracy and visualizing time-lagged forecast data would further facilitate decision-making.

Bibliography

[1] National Center for Visual Analytics NCVA url: http://ncva.itn.liu.se. [cited at p. iii]

[2] L. A. Treinish, ”Visual data fusion for applications of high-resolution numeri-cal weather prediction,” in 2000 IEEE Visualization Conference, 2000, pp. 477-480+594. [cited at p. 1]

[3] M. Mandiak, P. Shah, Y. Kim and T. Kesavadas, ”Development of an integratedGUI framework for post-disaster data fusion visualization,” in 2005 7th InternationalConference on Information Fusion, FUSION, 2005, pp. 1131-1137. [cited at p. 1]

[4] Swedish Meteorological and Hydrological Institute (SMHI) url: http://www.smhi.se.[cited at p. 1, 3]

[5] European Centre for Medium-Range Weather Forecasts (ECMWF) url:http://www.ecmwf.int/. [cited at p. 3, 6]

[6] W. Mller and H. Schumann, ”Visualization methods for time-dependent data - anoverview,” in Proceedings of the 2003 Winter Simulation Conference: Driving Inno-vation, 2003, pp. 737-745. [cited at p. 4]

[7] GeoVista Studio url: http://www.geovistastudio.psu.edu. [cited at p. 4]

[8] CommonGIS url: http://www.commongis.de. [cited at p. 4]

[9] G. Andrienko and N. Andrienko, ”Visual exploration of the spatial distribution oftemporal behaviors,” in 9th International Conference on Information Visualization,iV05, 2005, pp. 799-806. [cited at p. 4]

[10] N. Andrienko and G. Andrienko, ”Interactive visual tools to explore spatio-temporalvariation,” Proceedings of the Working Conference on Advanced Visual InterfacesAVI 2004, pp. 417-420, 2004. [cited at p. 4]

[11] M. Jern and J. Franzn, ””GeoAnalytics” - exploring spatio-temporal and multivari-ate data,” in Information Visualization 2006, IV06, 2006, pp. 25-31. [cited at p. 4]

[12] M. Jern and J. Franzn, ”Integrating InfoVis and geo vis components,” in 11thInternational Conference Information Visualization, IV 2007, 2007, pp. 511-518.[cited at p. 4]

[13] G. Grinstein, M. Trutschl and U. Cvek, ”High-dimensional visualizations,” Proceed-ings of the 7 SIGKDD Workshop on Visual Data Mining (KDD-2001)Th, pp. 1-14,2001. [cited at p. 4]

19

BIBLIOGRAPHY 20

[14] J. Blaas, C. P. Botha and F. H. Post, ”Extensions of parallel coordinates for interac-tive exploration of large multi-timepoint data sets,” IEEE Trans. Visual. Comput.Graphics, vol. 14, pp. 1436-1443, 2008. [cited at p. 4]

[15] A. Inselberg, ”The plane with parallel coordinates,” The Visual Computer, vol. 1,no. 4, pp. 69-91, 1985. [cited at p. 4]

[16] E. J. Wegman, ”Hyperdimensional data analysis using parallel coordinates,” Jour-nal of the American Statistical Association, vol. 85, no. 411, pp. 664-675, 1990.[cited at p. 4]

[17] ! G. Andrienko, N. Andrienko, ”Construction Parallel Coordinates Plot for ProblemSolving,” Smart Graphics ’01 (Pergamon), 2001. [cited at p. 4]

[18] P. Lundblad, M. Jern and C. Forsell, ”Voyage analysis applied to geovisual analyt-ics,” in 12th International Conference Information Visualization, IV08, 2008, pp.381-388. [cited at p. 4, 18]

[19] J. C. Roberts, ”Exploratory visualization with multiple linked views,” ExploringGeovisualization, pp. 159-180, 2004. [cited at p. 4]

[20] B. Shneiderman, ”Dynamic queries for visual information seeking,” IEEE Software,vol. 11, pp. 70-77, 1994. [cited at p. 4]

[21] L. Spirkovska and S. K. Lodha, ”AWE: Aviation weather data visualization envi-ronment,” Comput Graphics (Pergamon), vol. 26, pp. 169-191, 2002. [cited at p. 4]

[22] Y. Song, J. Ye, N. Svakhine, S. Lasher-Trapp, M. Baldwin and D. S. Ebert, ”Anatmospheric visual analysis and exploration system,” IEEE Trans. Visual. Comput.Graphics, vol. 12, pp. 1156-1164, 2006. [cited at p. 4]

[23] M. C. Dorneich, O. Olofinboba, S. Pratt, I. Wilson and C. Herbster, ”An experi-mental evaluation of weather avoidance using route optimization as a decision aid,”in 2002 IEEE International Conference on Systems, Man and Cybernetics, 2002,pp. 608-612. [cited at p. 4]

[24] GAV Framework url: http://vita.itn.liu.se/GAV. [cited at p. 5, 6]

[25] National Center for Environmental Prediction url:http://www.nco.ncep.noaa.gov/pmb/docs/on388/. [cited at p. 6]

[26] A Guide to WMO GRIB url: http://dss.ucar.edu/docs/formats/grib/gribdoc/.[cited at p. 6]

[27] C. M. D. S. Freitas, P. R. G. Luzzardi, R. A. Cava, M. A. Winckler, M. Pimentaand L. P. Nedel, ”On Evaluating Usability of Information Visualization Techniques,”AVI’02: Proceedings of Advanced Visual Interfaces, 2002. [cited at p. 14]

[28] Media Technology and Engineering, Linkping University, Campus Norrkping url:http://www.medieteknik.nu. [cited at p. 14]

[29] Lu, C. Yuan, H. Schwartz, B. E. Benjamin, S. G. , ”Short-Range Numerical WeatherPrediction Using Time-Lagged Ensembles” Weather and Forecasting, vol. 22; NUMB3, pages 580-595, 2007 [cited at p. 18]

List of Figures

1.1 Wind Flags . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2.1 SWIM overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52.2 Weather and voyage visualization in the world map . . . . . . . . . . . . . . . 82.3 Change of time step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4 World map level of detail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.5 Time step parallel coordinates plot . . . . . . . . . . . . . . . . . . . . . . . . 102.6 Ship time graph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.7 Highlighting in world map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.8 Selection area tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122.9 Filtering using time step parallel coordinates plot . . . . . . . . . . . . . . . . 122.10 Task flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

21