SEARCH

SEARCH BY CITATION

Keywords:

  • weather;
  • wave;
  • hydrodynamics;
  • forecasting system;
  • GIS;
  • integration;
  • interactive maps

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

The prediction and presentation of weather and marine conditions in an integrated way is of paramount importance in many areas of interest, including offshore operations. Huge datasets which are generated from running weather and marine forecasting systems present a challenge in terms of processing and presenting these data efficiently. This paper presents a method of optimizing huge meteorological and marine datasets for effective use by GIS, creation of automatic meteorological and marine maps, and an interactive map for querying any necessary integrated meteorological and marine data at selected geographical locations. This application significantly reduces the data processing time and the complexities in presenting such huge datasets. Copyright © 2012 Royal Meteorological Society


1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

The weather and marine predictions play an important role, not only for an individual person who would like to know the daily forecasts but also to companies who have large-scale operations both onshore and offshore in order to plan the day to day operations. Offshore facilities require not only general weather information such as temperature and winds, they also need marine information such as wave height and frequency and water currents. Accurate predictions in and surrounding the facilities are very important for safe and efficient operations. This means that the weather and marine models need to be configured in the area with very high resolution.

Configuring and running these high resolution weather and marine models poses a great challenge. However, once these models are put in place and are running, a large amount of data is generated. Displaying such a large dataset and converting it into meaningful information is demanding. Creation of an optimal database to store the data, automatic creation of high resolution meteorological and marine maps, interactive querying and visualization of the data is extremely challenging and is seldom addressed.

From the past, where the media of communication was printed newspaper, to this digital age of television and the internet, the ultimate aim has been to convey information to users in a way which can be easily interpreted and understood. Keeling (2010) has outlined the visualization of the weather during the past and present and has emphasized that the challenge still exists for better and faster communication of such information.

The strides that the field of geographical information systems (GIS) is making as an application in almost every field are incredible. New GIS tools enable users to process and better portray the forecast weather and marine data in an efficient and rapid manner.

GIS is being increasingly used in several aspects of meteorological and marine applications. Dyras and Serafin-Rek (2005) demonstrated the use of GIS for precipitation mapping: spatialization techniques were used to process the data from meteorological measurements into web images. However, the procedure was not fully automatic and lacked interactivity. Barszczynska et al. (2005) used several GIS tools to enhance hydro-meteorological services, although the tools were not truly integrated. Ustrnul and Czekierda (2005) applied GIS for the development of climatological air temperature maps. Although GIS spatialization techniques were used, the approach was restricted just to the production of air temperature maps. Luna et al. (2006) used GIS for evaluating maps of extreme temperatures. Franke et al. (2008) used GIS to regionalize precipitation data using a trigonometric approach, whilst Matyas (2008) applied GIS to shape the measures of rain shields in a landfilling tropical storm. Dou and Zhao (2011) applied GIS to study the climate change and its human dimension using meteorological statistics.

The present paper presents a method of optimization of huge meteorological and marine datasets for effective use in GIS, creation of automatic meteorological and marine maps, and description of an interactive map for querying the predicted meteorological and marine data precisely at a required geographical location. The examples are given from the Saudi Arabian and the Arabian Gulf regions.

2. System and data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

2.1. The system

The developed system consists of weather, wave and hydrodynamics models. The domain of the weather model covers the entire Middle East and marine models encompass the Arabian Gulf, as shown in Figure 1. Since the requirement of the system is to have accurate data, the models are configured to predict information at an interval of approximately 3 km which is avery high resolution forecast. The relationships among the models are illustrated in Figure 2.

thumbnail image

Figure 1. Weather, wave and hydrodynamics model domains. This figure is available in colour online at wileyonlinelibrary.com/journal/met

Download figure to PowerPoint

thumbnail image

Figure 2. Integrated system

Download figure to PowerPoint

The primary model in the system is the weather model which also the most computing intensive. The system starts executing by obtaining the analysis and forecasts gridded data from National Center for Environmental Prediction (http://www.ncep.noaa.gov/) The outputs of the model, which are weather parameters (temperature, winds, humidity, pressure, rain and dust), are processed and stored in a centralized database. The wave model subsequently starts its simulations by reading the necessary input data (e.g. temperature, wind speed and direction) from the weather output, which is stored in the database. The outputs of the wave model, after completion, are stored in the same centralized database. Similarly, the hydrodynamic model takes the necessary parameters (e.g. wind speed and direction) from the database, simulates the forecast and stores it back to the database. The database acts as an integrating tool for all the models. The configuration of the individual models in the system, the parameters involved and the volume of data generated as a result of running these models are elaborated in the following sections.

2.2. Weather data

The core of the whole system is the weather simulation model. SKIRON (Kallos et al., 1997, 2006) was chosen for this system and customized and configured to the local environment. It is based on the Eta/NCEP limited area weather forecast model (Mesinger et al., 1988; Janjic, 1994). The selection of SKRION as the weather forecast system was based on several reasons:

  • the model includes several advanced characteristics, such as data assimilation scheme for soil temperature and soil wetness, that make it appropriate for use in regions with varying physiographic characteristics;

  • it includes geological and land use categories that fit better in the landscape of Saudi Arabia and the surrounding area (sand and rocky soil characterization with the appropriate thermophysical and hydraulic properties);

  • the atmospheric model may easily interface with wave, hydrodynamic and ecological models (e.g. WAM model), through the production of specific outputs, and,

  • the system incorporates the desert dust cycle module that predicts dust, so as to simulate realistically all the phases of the dust cycle; the algorithms related to the dust cycle include the state of the art parameterizations of all the major phases of the desert dust life such as production, diffusion, advection and removal processes.

This model was implemented for the large domain covering the whole of the Middle East, with a very high resolution grid size of about 3 km. The model simulates the data for 120 h and for about 16 weather parameters (Table 1) resulting in the generation of huge sets of data of about 34 million records per day. Some of these parameters (such as temperature, wind speed and direction) are necessary to run the other models (the wave and hydrodynamic models). The weather model configuration is summarized in Table 2 and the model domains are illustrated in Figure 1.

Table 1. Model and the forecasted parameters
ModelParameters
WeatherMean sea level pressure, temperature fields at 2 m, 850, 700, 500 hPa, specific/relative humidity at 2 m, wind speed and direction at 10 m, 850, 700, 500 hPa, latent/sensible heat flux, cloud cover, accumulated precipitation, dust concentration near the ground, dust load, dry and wet deposition of dust particles
WaveSignificant wave height, mean wave direction, peak frequency, mean frequency, friction velocity, wind direction, drag co-efficient, wind speed at 10 m, mean wave period, mean square slope
HydrodynamicsWater current speed and direction
Table 2. Model characteristics and volume of data produced
ModelNo of receptor pointsGrid size (KM)Forecasting horizon (hours)No of parametersRecords generated day−1
Weather289 19831201634 703 760
Wave44 3413120105 320 944
Hydrodynamics12 138312061 456 672

2.3. Wave data

The wave model chosen for this purpose is WAM (Hasselmann et al., 1988) and is configured for the Arabian Gulf. This model reads its required input from the output of the weather model which is stored in the database. The configuration of the model is summarized in Table 1. The model runs for about 120 h, generating about 5.3 million records per day for 10 parameters (Table 2).

2.4. Hydrodynamics data

The hydrodynamics model chosen is HYDRO2 (Al Rabeh et al., 1990), which predicts the ocean currents. The domain of this model is coastal, partially covering the Arabian Gulf as illustrated in Figure 2. The model is relatively less computing intensive and generates about 1 million records per day which are stored in the database. Compared to the above two models the volume of data generated is low, as shown in Table 1.

2.5. GIS data

The presentation of information is enriched when overlaid on different GIS layers. GIS data used for generating the maps are the base map, land use map and other important structures. The term geodatabase is used for the container holding the collection of geographic datasets. One of most popular geodatabase formats is the Environmental Systems Research Institute (ESRI) File geodatabase, which is used to store the data in this particular study (http://www.esri.com/software/arcgis/geodatabase/index.html).

3. Development of the meteorological and marine database

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

The integrated system of weather and marine models produces a large amount of data. Traditionally these data are retained as standard ASCII files. These files, however, are inconvenient in terms of storing, retrieving and manipulating the contents, in addition to the absence of the functionalities of a typical relational database. The database has the ability to access relational data sources, retrieving data for computational and easy storage using its powerful queries, and was used as a central point of communication among various models in the integrated structure (Figure 2).

The development and implementation of the database has been a challenging task due to the complex structure of the different models and the huge amount of data being produced each day. The database model was created taking into consideration the different requirements such as the data types of each parameter, the relationship among the parameters and the volume of output produced by the models participating in the integrated system. The relational database is typically illustrated in the form of entity relationship (ER) diagrams which depict the column names of each table and the relationship among the different tables. The ER diagram of a simple version of the meteorological database is shown in Figure 3.

thumbnail image

Figure 3. Database ER diagram

Download figure to PowerPoint

The complete data generated by each model from the whole domain can be stored in the single table. However, storing and retrieving the large volumes of data through a single table is highly time and resource consuming and is a major bottleneck for the smooth and timely operation of the process. In order to address this issue, a unique spatial optimization technique has been implemented. The tables storing the model output i.e. TAB_SKIRON, TAB_WAM and TAB_HYDRO (Figure 3) have been duplicated into several tables and data are divided based on the range of geographical locations. The number of duplicate tables and the division of data depends on the required data storage and retrieval efficiency. The shorter the range of the geographical co-ordinates of the tables the greater the efficiency of the data storage and retrieval. The geographical size of the table may be based on the area of interest, such as population centres or installations. The size of the box also depends on the creation of GIS maps (such as temperature, winds, rain, dust) of the required area of interest. For example, the data from the SKIRON model was divided into about 40 tables, as shown in Figure 4. The geographical division of the tables was based on the requirements of GIS maps.

thumbnail image

Figure 4. Optimization of database tables based on geographical location. This figure is available in colour online at wileyonlinelibrary.com/journal/met

Download figure to PowerPoint

The models execute and produce data for the whole domain together. In order to divide the data and store them in the respective table an algorithm is created. This algorithm (illustrated in Figure 5) processes the data for the range of geographical co-ordinates for the corresponding table and stores them into respective tables. This algorithm is also used to retrieve the data: whenever the query for the data is initiated, the algorithm processes for the corresponding table and directs the user to the table where the data are stored corresponding to the selected geographical co-ordinate. This spatial optimization of the database tables and the algorithm reduces the data storage and retrieval time significantly. For example, the data processing time was reduced by approximately 40 times for the SKIRON model as the single table was divided into 40 geographically based tables.

thumbnail image

Figure 5. Data query process

Download figure to PowerPoint

4. Creation of GIS maps

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

GIS tools depict the geographical information as detailed maps and images more effectively than other presentation techniques. The digital maps, which are easily constructed, allow the presentation at different geographical locations with changeable (dynamic) scales. These maps open virtually unlimited avenues for data processing and spatial analysis. With the advent of powerful computers at desk level and the availability of data and networks, the production of digital maps is cheaper as well as more efficient than other methods. With proper methods such maps can be created almost automatically. The process of the creation of GIS maps in this study is completely automatic and time efficient. This process runs into several steps as illustrated in Figure 6. Each step is described in the following paragraphs.

  • Filter vector data. The central repository, i.e. the database, contains a large amount of data for all the parameters for different dates and times of the weather, wave and hydrodynamics models. This step involves querying necessary data for the creation of the images. Data need to be filtered out based on the geographical location of the image, the date and time of the image and the parameter for which the image is being created. Optimizing the database tables geographically, as described in Section '3. Development of the meteorological and marine database', helped significantly in reducing the time of creating the vector data file. Figure 7 illustrates this process.

  • Convert vector data to raster. Once the required dataset is filtered out it needs to be converted to a raster layer. Raster data are an abstraction of the real world where spatial data are expressed as a matrix of cells or pixels, with the spatial position implicit in the ordering of the pixels. With the raster data model, spatial data are not continuous but are divided into discrete units. This makes raster data particularly suitable for visualization of meteorological and marine datasets. The current step generates raster data in the format of a GRID as a GIS layer. In order to convert the vector data to raster, an appropriate interpolation technique has to be chosen. This was based on two primary factors: first, the suitability to the current datasets, (weather, wave and hydrodynamics datasets) and, second, the efficiency of the processing time, as the interpolation process is time consuming.

    ESRI ArcMap exposes several spatial interpolation techniques, the most prominent being the inverse distance weighted (IDW) and spline methods which are deterministic methods, while kriging methods are based on statistical models (http://webapps.fundp.ac.be/geotp/SIG/interpolating.pdf) These models include autocorrelation, i.e. assessing the statistical relationships among the measured points. Because of this, not only do geo-statistical techniques have the capability of producing a prediction surface, they also provide some measure of the certainty or accuracy of the predictions.

    The above mentioned interpolation techniques were employed to create the raster meteorological and marine maps. It was observed that the maps created using the kriging technique were found to be more accurate and appropriate for the images with the regular boundary (square or rectangle). This was also suggested by Ustrnul and Czekierda (2005). However, when the boundary of the images was irregular (such as island, country or sea boundaries), the nearest neighbour technique was found to be more appropriate.

    A test in processing times of different interpolation techniques for the same dataset and computing resource showed that the nearest neighbour (2 s) and IDW (3 s) techniques took the lowest processing time followed by the spline (6 s) and kriging (26 s) approaches. Nevertheless, the processing time is more for the kriging technique; it was chosen as the interpolation technique in order to produce better predicted surface maps where the boundary was regular, while the nearest neighbour technique was used for irregular boundaries. The output of this step is raster in ESRI GRID format as shown in the Figure 8.

  • Surface analysis. Simple raster data such as the GRID layer do not visually depict much, however they open up many options for performing surface analyses such as creating contour lines and directional arrows. This GRID can be used to perform visual analysis of the data, or can be used as a background map (base map) or can be used in other analyses. Surface analysis involves identifying a specific pattern within the raster dataset. Patterns such as contours, that were not readily apparent in the original raster dataset, can be derived. Contours are the most widely used meteorological and marine data display. Both colour and line contours are created as GIS layers in this step to illustrate the data pattern. In addition to this, the surface analysis aids in depicting the directions for certain parameters such as wind and current directions using arrows. Output of the surface analyses are the vector layers in the form of lines and points as shown in Figure 9.

  • Overlay appropriate GIS data. The power of GIS is overlaying layers to depict visually the relationship between them. In this step the necessary GIS layers such as facilities and road network information are added to the raster and contour layers making it a combined group layer and giving a better understanding of the variation of the parameters such as temperature and wind speed change in the area.

  • Export as image and publish to web. When the set of layers, including the raster layer, contours and other GIS layers are ready, the view is exported as an image. ESRI ArcGIS exposes several functions to convert the current map view to image. The generated images are published to the web. Figures 10-13 illustrate the final maps created through this process.

thumbnail image

Figure 6. Map creation process

Download figure to PowerPoint

thumbnail image

Figure 7. Process of creating a vector data file

Download figure to PowerPoint

thumbnail image

Figure 8. Convert vector data to raster

Download figure to PowerPoint

thumbnail image

Figure 9. Surface analysis. This figure is available in colour online at wileyonlinelibrary.com/journal/met

Download figure to PowerPoint

thumbnail image

Figure 10. Raster map of temperature generated from the process of the whole domain

Download figure to PowerPoint

thumbnail image

Figure 11. Raster map of waves generated from the process of the whole domain

Download figure to PowerPoint

thumbnail image

Figure 12. Raster map of temperature generated from the process of the selected domain

Download figure to PowerPoint

thumbnail image

Figure 13. Raster map of waves generated from the process of the selected domain

Download figure to PowerPoint

5. Interactive map

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

Static maps are the most common maps prepared for visualizing the meteorological and marine data. Going a step further, these static maps can be played with respect to time to give an animation effect. These colour coded images, as well as contours and other GIS data are significant for giving an understanding of the data, however many times it is required to know the vector or tabular data at an exact geographical location and must be known interactively i.e. by zooming to an area and clicking on the point of interest. The necessary data at the required location can be queried by entering the geographical co-ordinates, but with the advent of powerful GIS map services, a special interactive map is provided to the users where users can zoom in to any location and interrogate the map and obtain the required meteorological and marine data with a single click as shown in Figure 14.

thumbnail image

Figure 14. A sample output from the interactive map

Download figure to PowerPoint

Complex processes are involved in the background when the user clicks on the map. Firstly, if the selected co-ordinate is not the receptor point, (it is a point at which the model predicts the result), a nearest receptor point is chosen for each of the three models. Secondly, the required data are queried from the three different models and shown to the user together as an integrated output. The process is illustrated in the Figure 15. As an example, if the user clicks at a location in the sea, the result is displayed for weather, wave and hydrodynamics forecasted information in tabular or graph format, whilst if the user clicks on a land point, only the weather information is shown. A sample output is shown in Figure 14. This is an innovative way of handling both weather and marine data and has not previously been presented in the literature.

thumbnail image

Figure 15. Interactive map process

Download figure to PowerPoint

6. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

A spatially optimized database was created not only to store and retrieve huge amount of data generated from meteorological and marine models but also to serve as an efficient integrating mechanism to the models. A fully automatic and time efficient process converts the vector data into meaningful images and maps using GIS techniques. The interpolation technique of nearest neighbour and kriging was found to produce better meteorological and marine maps. An interactive map algorithm allows the user to interrogate the map at any location to view the model output information in an integrated and efficient manner. This GIS enhanced meteorological application simplifies the processing and presenting of huge data generated from meteorological and marine models in an efficient and effective way and reduces the complexities involved in the process.

Acknowledgements

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References

The author would like to thank the Research Institute of King Fahd University of Petroleum and Minerals for the support provided during the study.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. System and data
  5. 3. Development of the meteorological and marine database
  6. 4. Creation of GIS maps
  7. 5. Interactive map
  8. 6. Conclusions
  9. Acknowledgements
  10. References