Visualization of geochemical data for rocks and sediments in Google Earth: Development of a data converter application for geochemical and isotopic data sets in database systems

Abstract

We developed a Keyhole Markup Language (KML) generator for converting geochemical and isotopic data sets for rocks and sediments stored in database systems into KML. Our program allows users to visualize geochemical or isotopic data easily in Google Earth. The generator accepts data files produced by the database systems PetDB, SedDB, GEOROC, and GANSEKI. The data are plotted three-dimensionally as a bar graph on the surface of the virtual Earth at the sampling site. This type of visual presentation, including information on sample localities, directly shows the distribution of isotopic or compositional anomalies of specific samples on the Earth's surface. We provide a Web application for the generator, so anyone can set the parameters for visualization over the Internet. With other KML generators we developed earlier, geochemical data can be overlain on a seismic tomographic model. This overlay image can provide information on the origin of samples in the tomographic model.

Key Points

• Visualization of geochemical data for rocks and sediments in Google Earth
• Development of KML generator for geochemical and isotopic data set
• Overlay image of geochemical data on other geoscience data

1. Introduction

Rocks and sediments have been sampled from both the land and the ocean floor at various locations and then analyzed. As a result, a vast amount of geochemical data has been produced and published. Researchers have constructed database systems, e.g., Petrological Database of the Ocean Floor (PetDB; http://www.petdb.org/) [Lehnert et al., 2000] and Geochemistry of Rocks of the Oceans and Continents (GEOROC; http://georoc.mpch-mainz.gwdg.de/georoc/) [e.g., Sarbas et al., 2005] to accommodate the full range of analytical values published in scientific articles and to provide them through the Internet. Researchers can quickly obtain geochemical data on rocks and sediments sampled from various areas on the Earth from these database systems. Most geochemists use the graphic functions provided in spreadsheet software, such as Microsoft Excel, or data analysis applications to plot the data. The database systems usually provide data files in a format accepted by most of these applications, and therefore each geochemist may plot the data with relative ease using familiar applications. Although information on the sampling site is important in interpreting the obtained research results, common graphic presentations cannot simultaneously show both information on sample localities and geochemical data. The geochemical data are generally plotted in two-dimensional (2-D) space, which does not include geographical information.

The appearance of virtual globes, for example, Google Earth, completely changed the visual presentation of geographic data. By exploiting the layer system embedded in a virtual globe, researchers can display geographic data directly on a map, i.e., on the Earth's surface, with various types of other data. Keyhole Markup Language (KML) is an XML-based language schema for visualizing geographic data in Google Earth. Various research institutes and projects, e.g., the National Geophysical Data Center (NGDC), U.S. Geological Survey (USGS), and National Institute of Advanced Industrial Science and Technology (AIST), currently provide their own observational or analytical data in KML. As a result, Google Earth is becoming a commonly used data browser in all fields of the geosciences [e.g., Butler, 2006; Nourbakhsh et al., 2006; Yuan et al., 2008; Chen et al., 2008, 2009].

In this study, we also adopt Google Earth as a browser for geochemical data on rocks and sediments available in several online database systems and provide a tool for visualizing the data in Google Earth. We developed the KML generator, a program that converts data files obtained from these database systems into a KML file. It enables us to easily visualize geochemical data in Google Earth without any processing of the original data file. The data are plotted as three-dimensional (3-D) bar graphs at the associated sampling sites on the surface of the virtual Earth. Such visual presentations also provide the information on the sampling localities and can facilitate the interpretation of geochemical data. There are several Geographic Information System (GIS) tools for visualizing geochemical data. Virtual Ocean (http://www.virtualocean.org/) and GeoMapApp (http://www.geomapapp.org/) are representative efficient tools. They can also show the data directly on map. They are integrated tools with various functions not only for visualization but also data analysis. Our system with Google Earth is specialized in visualization. By selecting Google Earth as the visualization platform, we suppose that users need no extensive training to visualize the data because we aim to make the KML generator as simple to use as possible for easy handling of the tool and the generated images in Google Earth.

One of the most useful features of a virtual globe is its ability to present different types of geographic data simultaneously. With other KML files obtained from various research institutes via the Internet, researchers can overlay geochemical data on other observational/analytical data without using any complicated procedures. We have developed various KML generators, in addition to that for geochemical data on rocks and sediments, for compiling results obtained from different research fields in Google Earth [Yamagishi et al., 2009a, 2009b, 2010; Nagao et al., 2008]. Overlaying images by using multiple KML generators and existing KML files may give us a new look into the Earth's interior and could play an important role in constructing a new structure model.

2. Converting Geochemical Data for Rocks and Sediments to KML

We developed our KML generator for visualizing a set of geochemical and isotopic data as a Java application (Java SE 5.0). We also developed a Web-based graphical user interface (GUI) for the KML generator so that anyone may use it through the Internet. The application is currently accessible at http://www.jamstec.go.jp/pacific21/google_earth/. It produces KML files written in accordance with KML v2.0, which is compatible with the latest version (KML v2.2).

2.1. Data File Formats Accepted by the KML Generator

Our KML generator supports the data files provided by PetDB, GEOROC, Integrated Data Management for Sediment Geochemistry (SedDB; http://www.seddb.org/), and Geochemistry and Archives of ocean floor rocks on Networks for Solid Earth Knowledge Integration (GANSEKI; http://www.jamstec.go.jp/ganseki/). PetDB and SedDB provide data as Microsoft Excel files, and GEOROC and GANSEKI provide data as comma-separated variable (CSV) files. The types of column in the data file depend on the database system and on the parameters that users choose when they create the data file in the system.

The KML generator requires columns for the sampling position (latitude and longitude) and data (the analytical value). The database systems supported here always provide all of these columns. The contents of additional metadata columns will appear in Google Earth; these columns list the sample name or unique ID for the analysis, the citation for the corresponding paper, and the analytical method used to process the data (see section 2.2.4). The first row in the data file must show the titles of the columns; the titles of the required columns are listed in Table 1. The KML generator can automatically select the columns for the analytical values to be plotted in Google Earth because it includes a built-in exclusion list for each database system depending on the column titles. Therefore, neither the input data file nor the generator itself must be reconfigured for a particular database system. The KML generator will also accept a user's own data file if it has the same format as files obtained from the database systems mentioned above and then plot the data in Google Earth.

Table 1. Column Titles Required by the KML Generator in the Original Data File Provided by a Database System

Value of the Column	Title of the Column	Database Systems
Longitude	Longitude^a	PetDB, GANSEKI
	Long^a	SedDB
	LONGITUDE MIN^a	GEOROC
Latitude	Latitude^b	PetDB, GANSEKI
	Lat^b	SedDB
	LATITUDE MIN^b	GEOROC
ID for the analysis	Sample_ID^c	PetDB
	Sample Name^c	GANSEKI
	Sample ID^c	SedDB
	UNIQUE_ID^c	GEOROC
Analytical method	Method^d	All
Citation for corresponding paper	Reference^e	PetDB, GANSEKI, SedDB
	CITATIONS^e	GEOROC

• a If there are columns whose first row is the same this column title, the generator will accept the first of them as the “Longitude” of the sampling point. If not, the generator will stop to make a KML file and show an error message.
• b If there are columns whose first row is the same this column title, the generator will accept the first of them as the “Latitude” of the sampling point. If not, the generator will stop to make a KML file and show an error message.
• c If there are columns whose first row is the same as this column title, the generator will accept the first of them as “ID for the analysis.” If not, in the balloon and the left panel in Google Earth, they will appear as a dash. If a column appears but no values are given, they will be shown as blank.
• d If there are columns whose first row is the same as this column title, the generator will accept the first of them as “Analytical method.” If not, in the balloon and the left panel in Google Earth, they will appear as a dash. If a column appears but no values are given, they will be shown as blank.
• e If there are columns whose first row is the same as this column title, the generator will accept the first of them as “Citation for corresponding paper.” If not, in the balloon and the left panel in Google Earth, they will appear as a dash. If a column appears but no values are given, they will be shown as blank.

2.2. Plotting Geochemical Data for Rock and Sediment Samples in Google Earth Using the KML Generator

The KML generator produces a KML file for plotting geochemical data for rock and sediment samples as 3-D bar graphs that appear in Google Earth at the sampling sites.

The generator can handle various geochemical values that can be classified into five categories: (1) concentration, (2) isotope ratio, (3) ratio of category 1 or 2, (4) ratio between an index value and derivatives from the index value, such as the delta value for stable isotopes and analogs for rare earth elements (REEs) or other metals, and (5) compositions. Categories 1, 2, and 3 are single value. The precalculated values of category 4 are also single. For this mode, the tool supplies the single-value plot mode, in which each value is represented by a bar graph height. For categories 3 and 4, the generator provides a function for calculating the ratio between two analytical values. For category 5, the tool supplies a multivalue stack mode, in which the values of a data set are presented in a stacked bar graph. Major element compositions are a specialized application of this mode, and REE compositions, noble gas relative abundances, etc., can also be displayed. Examples of how these functions of the generator are used will be given in sections 2.2.1–2.2.4.

2.2.1. Plotting a Single Value Produced by One Type of Analytical Value

Figure 1 shows examples of the bar graphs constructed for a selected type of the analytical data. The bars on the 3-D bar graphs become taller and thinner as the values increase. Even if there are data sets for multiple samples taken from the same location, users can distinguish the bar graphs because of the variation of their thickness shown in Google Earth. The color of a bar graph varies gradually between the selected colors over the range between the maximum and minimum values or changes at threshold(s) between two or three selected colors (see Figure 1).

Figure 1

(a) The ³He/⁴He distribution on the North-Atlantic Ridge. The original isotope data were obtained from PetDB (glass and whole rock samples of MORB). The ³He/⁴He values commonly exist in the range 8 ± 1Ra except around hot spot regions. The green bar graphs indicate the values of ³He/⁴He in the range from 7 to 9Ra. Blue and red indicate values of less than 7Ra and greater than 9Ra, respectively. Over a wide area around Iceland, ³He/⁴He ratios are high because of the high mobility of He. (b) Stacked bar graph showing the legend.

2.2.2. Plotting the Ratio Between Two Types of Analytical Value

Chemical compositions or isotopic ratios are usually plotted on a scatter diagram, typically in a 2-D space representing different isotopic ratios or chemical compositions. The overall trend of the evolution of rocks or the border of a rock group is indicated by an area or a line, respectively, in this diagram. To exhibit anomalies in data obtained for rock samples by using the evolution curve, users can select two types of analytical value for isotope ratios or chemical compositions from the list automatically displayed in the KML generator's GUI and make a KML file to plot the ratio between the selected values in Google Earth. By setting the evolution or border line value to “threshold” and defining colors for values greater than or less than the threshold, users can clearly see the group to which the rock samples belong (see Figures 2 and 3).

Figure 2

Schematic diagram illustrating how the colors and threshold are set when the ratios between two types of analytical value, value1 and value2, are plotted. When a threshold and two colors, color1 and color2, are set in the generator, bar graphs showing a ratio larger than the threshold will appear in color2, and other bar graphs will appear in color1. The criterion is represented by a line: value1 = threshold × value2. When this line has an intercept at value1, or a vertical deviation from the line is to be shown, the value of the intercept or the deviation can be set to “offset.” The criterion is given by the line: value1 = threshold × value2 + offset. In Google Earth, the calculated values of [(value1 – offset)/value2] will be plotted as bar graphs.

Figure 3

Distribution of isotope anomalies of Pb (a) on the Indian Ocean and (b) on the Pacific Ocean. The original isotope data were obtained from PetDB (glass and whole rock samples of MORB). The Northern Hemisphere Reference Line (NHRL) in ²⁰⁸Pb/²⁰⁴Pb is given by (²⁰⁸Pb/²⁰⁴Pb)_NHRL = 1.209(²⁰⁶Pb/²⁰⁴Pb) + 15.627 [Hart, 1984]. The magnitude of the isotopic anomaly in ²⁰⁸Pb/²⁰⁴Pb is expressed as the vertical deviation from the reference line, that is, Δ8/4 = [(²⁰⁸Pb/²⁰⁴Pb) – (²⁰⁸Pb/²⁰⁴Pb)_NHRL] × 100. The Dupal anomaly appears as anomalous Sr, Nd, and Pb isotopic signatures in mid-ocean ridge basalts. The Dupal anomaly in ²⁰⁸Pb/²⁰⁴Pb is defined as a vertical deviation of more than 40. Therefore, the values of [(²⁰⁸Pb/²⁰⁴Pb) – 16.027]/(²⁰⁶Pb/²⁰⁴Pb) are plotted. Plotted values of more than 1.209 are represented as the Dupal anomaly, as shown in the red bar graphs. The rocks sampled from beneath the Indian Ocean exhibit a large-scale isotope anomaly in Pb in the southern hemisphere mantle. On the other hand, few rock samples from the Pacific Ocean show the anomaly.

The generator also plots the deviation from the threshold in this mode. When two types of analytical value are selected, the “offset value” can be set as the index in the GUI, and the calculated values of [(analytical value – offset value)/other analytical value] will be plotted (see Figure 2). Figure 3 shows an example of the DUPAL anomaly. Figure 3 clearly shows that the values of ²⁰⁸Pb/²⁰⁴Pb in mid-ocean ridge basalts sampled from the Indian Ocean are higher than the Northern Hemisphere Reference Line (NHRL), which is the Dupal anomaly [Hart, 1984; Rollinson, 1993]. However, in the Pacific Ocean, few rocks show this anomaly.

The visualization method proposed here, in which data are plotted directly on the sampling site, can help users to intuitively understand the distribution of isotopic anomalies on the Earth's surface (see Figures 1 and 3). With a huge amount of isotopic data stored in the database systems, users easily and rapidly examine where and what kind of isotopic anomalies exist in volcanic rocks on the Earth's surface, which allows us to investigate the origin of the anomalies. By using one of the KML generators we previously developed, the global map of isotopic anomalies can be overlain on seismic tomographic model in Google Earth. Seismic tomographic model is useful to discuss spatial extent of temperature variation in the mantle. The overlaying image of isotopic data on seismic tomographic model possibly provides a clear image of relationship between mantle temperature variations and isotopic anomalies of volcanic rocks derived from such mantle (see section 3.1).

2.2.3. Plotting Multiple Values and Merging Them as a Stacked Bar Graph

The KML generator can also make a KML file to plot all the major element compositions at once as stacked bar graphs in Google Earth (see Figure 4). The KML generator has a list of the elements in data files obtained from the database systems (see Table 2) and automatically selects and uses the composition values existing in the processed data file to produce bar graphs, even if the data for certain samples do not include all the elements on the list. Users can choose to replace selected elements with other elements from the list of analytical values. By default, each element has a different color on the stacked bar graphs, but users can change the color for each element.

Figure 4

(a) Major element compositions of rock sampled from Karoo Province. Original geochemical data were obtained from GEOROC as a precompiled data file (KAROO PROVINCE.csv) including various volcanic rocks, e.g., basalt, andesite, and rhyolite. Major element compositions are shown as a stacked bar graph. Graphs are constructed even if the composition data in the data file are incomplete. The remainder of the total abundance of major element compositions making up 100% is expressed as “residual” and appears as a black bar in the graph. (b) Stacked bar graph showing the legend.

Table 2. Column Titles Regarded as Major Element Compositions

Major Element Compositions	Column Titles
	PetDB, GANSEKI	SedDB	GEOROC
SiO2	SiO2	SiO2 [wt%]	SIO2(WT%)
TiO2	TiO2	TiO2 [wt%]	TIO2(WT%)
Al2O3	Al2O3	Al2O3 [wt%]	AL2O3(WT%)
Cr2O3	Cr2O3	Cr2O3 [wt%]	CR2O3(WT%)
Fe2O3	Fe2O3	Fe2O3 [wt%]	FE2O3(WT%)
FeO	FeO	FeO [wt%]	FEO(WT%)
MnO	MnO	MnO [wt%]	MNO(WT%)
MgO	MgO	MgO [wt%]	MGO(WT%)
CaO	CaO	CaO [wt%]	CAO(WT%)
Na2O	Na2O	Na2O [wt%]	NA2O(WT%)
K2O	K2O	K2O [wt%]	K2O(WT%)
P2O5	P2O5	P2O5 [wt%]	P2O5(WT%)

2.2.4. Information on the Plotted Analytical Values

The coordinates of the sampling site, the citation for the corresponding paper, the analytical method, and the analytical value(s) are also shown in Google Earth (see Figures 1 and 4). A small gray sphere appears at the top of the bar graph; by clicking the sphere, users can view a balloon displaying this information. The titles of the columns for these data in the data file for each database system are listed in Table 1. Samples with no information on the analytical method or the corresponding paper in the data file can be used to generate a KML file, but the missing information is indicated in the balloon by a line.

2.3. KML Structure

The bar graphs are constructed using the element <Polygon> in Google Earth. In the KML file, separate bar graphs are constructed for each analysis, i.e., for each row in the data file, and are put into different <Folder> elements. The sample name or unique ID for an analysis is set in the coelement <name> in the <Folder>, and the corresponding paper and analytical method are written in the coelement <description>. Therefore, the bar graphs are listed in the “place” panel in Google Earth with the sample name or unique ID, reference, and method. This list allows users to select and jump to a bar graph they wish to see; when they double-click on the folder name in the panel, the view position in Google Earth will jump to just above the selected bar graph.

Other stacked bar graphs are shown floating in space in order to display the legends (see Figures 1 and 4). Users can select the position of these bar graphs by using optional parameters for the generator. We are currently improving the method of showing the legends because stacked bar graphs floating in space are not necessarily displayed on the screen, especially when users focus on a very small geologic region, as described below. When we began to develop the generator, KMZ files, which are compressed KML files including image files, were not supported by Google Earth. Therefore, we displayed the legends as stacked bar graphs by using <Polygon> in KML. We plan to modify the generator to show the legends by using <ScreenOverlay> in the next version so that the legends are always displayed on screen.

3. Uses of the Generator That Exploit Google Earth Technology

3.1. Overlaying Images of Geochemical Data for Rocks and Sediments on Other Geoscience Data Using Other KML Generators

We have developed other KML generators to display various geoscience data in Google Earth [Yamagishi et al., 2009a, 2009b, 2010; Nagao et al., 2008]. One produces a KML file for visualizing a seismic tomographic model that can be overlain on geochemical data in Google Earth. The main purpose of the seismic tomographic model is to indicate temperature anomalies in the Earth's interior, and geochemical data for rocks show the horizontal distribution of the mantle's chemical compositions. An image in which geochemical data are overlain on the seismic tomographic model correlates the thermal and compositional anomalies.

Figure 5 is an example of a visualization of geochemical data for rocks in combination with the tomographic model via the KML generators. The work of Isse et al. [2009] indicates that three patches of slow seismic velocity anomalies appear in the mantle wedge beneath the Izu-Bonin-Mariana arc, and volcanic rocks from each anomalous region show a unique signature in the Sr, Nd, and Pb isotope ratios. Figure 5 clearly shows this relationship. This could further our understanding of the origin of magma and the evolution of the Izu-Bonin-Mariana arc region.

Figure 5

Seismic tomographic model and isotope ratios of Sr, Nd, and Pb resulting from analysis of volcanic rocks obtained from the ocean floor in the Izu-Bonin-Mariana arc region [Isse et al., 2009]. (a) Purple, (b) green, and (c) blue bar graphs show isotopic ratios of ⁸⁷Sr/⁸⁶Sr, ¹⁴³Nd/¹⁴⁴Nd, and ²⁰⁶Pb/²⁰⁴Pb, respectively. By controlling the thickness and height of the bars, the user can plot (d) all of them together with the seismic tomographic model. Figure 5 clearly show that the three patches of slow seismic velocity anomalies in the mantle wedge beneath the Izu-Bonin-Mariana arc are correspond to the three areas on the Earth's surface defined by the isotopic data.

3.2. Spatial Scalability of the Generator and Google Earth

Large amounts of geochemical data are stored in the database systems mentioned above, in particular data on rocks and sediments sampled from areas of scientific interest. In these regions, many samples have been collected and analyzed, so many data are available for a small area. In these cases, it is difficult to distinguish each data set in Google Earth.

One of the advantages of Google Earth is the extremely high scalability of the display space. Google Earth can show the entire Earth or significantly enlarge a small area. The KML generator also offers high spatial scalability because users can adjust the height and thickness of the bar graphs. The scalability of Google Earth and the generator enables users to visualize either the data on a hemisphere-scale area (see Figure 3) or those on small geological scales. Figure 6 shows a geochemical map that displays the chemical compositions of the soil in Aichi prefecture in Japan [Yamamoto et al., 2007]. These geochemical data are not stored in a database system, but users visualized them in Google Earth via the generator by preparing a data file with the same format as data files provided by the database systems. The mapping area is very small, several tens of kilometers square, but many rock samples for the area exist. By selecting the appropriate height and thickness of the bar graphs, users can visualize the data via the generator without using any special procedures. With the KML file provided by the AIST [Geological Survey of Japan, 2010], geochemical data can be compared directly with the geological map in Google Earth.

Figure 6

Geochemical map of Aichi prefecture in Japan showing the distributions of SiO₂ [Yamamoto et al., 2007] and the geological map of this area using KML files provided by AIST [Geological Survey of Japan, 2010]. The color of the bar graphs varies from blue to red with increasing values. Yamamoto et al. [2007] reported that SiO₂ is depleted in the stream sediments of granitic areas. The pink area on the geological map shows granite produced in the Cretaceous period. The geochemical data can be compared directly with the geological map in Google Earth.

4. Conclusion

We have developed a KML generator to convert geochemical data for rocks and sediments to KML so data can be visualized in Google Earth. The graphical method proposed here can display not only data but also information on the sampling site, which facilitates interpretation of the analytical results. We have also developed other KML generators. The use of multiple generators can provide quick access to an image in which various geoscience data are overlain, which would be useful for investigating where or how the sampled rocks originated.

5. Future of the KML Generator

We will continue to maintain and to improve the KML generator until Google decides not to develop Google Earth anymore. We already started development of the next version of the generator for the geochemical data. We are improving the Web GUI so that users can handle the tool more intuitively. As mentioned in section 2.3, we plan to change the method showing the legend. We will, in the future, add new functions into the KML generator, e.g., new visualization mode for multiple samples taken from a core, based on users' feedbacks.

Acknowledgments

We are grateful to I. Kumagai for helpful discussions and comments. We thank T. Tanaka, K. Yamamoto, and M. Minami for permission to use their data. Google Earth is a trademark of Google Inc. This work was supported by a Grant-in-Aid for Research and Development (JAMSTEC award) from the Japan Agency for Marine-Earth Science and Technology.