Earth Science with Digital Data

Earth Science Instruction with Digital Data

J. D. Hays, S. Pfirman, B. Blumenthal, K. Kastens and W. Menke

Abstract

A new holistic view of our planet as a single complex system and rapidly advancing ways to access manipulate and visualize large data sets, challenges our conventional methods of presenting the Earth to undergraduates. The Lamont data viewer has been developed to allow students to focus on learning from data visualizations and calculations and lets them ignore the complexities of accessing, formating and making figures from, and calculations on the data. The capacity to access a variety of data sets, integrate the information, manipulate it and display it and in common formats adds a new dimension to the way earth science is taught. Accessing the primary data from which great discoveries were made and that underlie the paradigms of our time, allows students to explore the data and recreate the discovery process. Learning from data is learner centered, encourages student inquiry and the practice of the scientific method. However learning directly from data, at the undergraduate level, is an unexplored arena and much work needs to be done to maximize the gain from this form of pedagogy.

Introduction

As this century draws to a close conventional methods of undergraduate earth science instruction are being challenged by new mental models of the Earth and new ways of accessing, processing and presenting information. Many Earth Scientists now view the Earth as a single system, including the atmosphere, hydrosphere and biosphere as well as the solid earth. To understand these system components and the complex interactions amongst them requires the integration of information from many sources. The computer's capacity to access, integrate, organize, process and display information is certain to be central to efforts, by Earth Science instructors, to present this perspective.

Learning, in the Earth Sciences, need no longer be confined to, or even centered on words. Most earth science information is now in digital form and assembled in large databases. Although publicly available, these databases until recently have had no value for undergraduates. Raw numerical data has little meaning and most student computers lack the capacity to process large data sets. Even with access to powerful workstations, students are overwhelmed by the difficulties of handling multiple formats and using sophisticated software to manipulate large data sets. Network bandwidth and server capacity limit the feasibility of processing and simultaneously transmitting large data sets to multiple users.

To help overcome these problems one of us, (Benno Blumenthal), developed a versatile data processing tool, the Lamont WWW Data Library. The educational advantage of this tool is that it allows students to focus on learning directly from data, which is an important challenge in its own right. The Lamont Data Library has a number of important attributes:

• it brings data documentation and numerical variables together, so users can easily move between them
• accessing data sets, frees students from the restriction of viewing a few pre-selected, pre-plotted data slices, making true data exploration possible
• it is built upon software that manipulates datasets, including rendering data as illustrated hypertext with the resulting figure documented, so users can know how the figure was created (datasets used and manipulations performed). The user can then apply the same analysis to a different data set, or a different analysis to the original data set.
• it has a variety of data file and table formats making it possible to transfer data to other programs
• it can perform a range of calculations on data including arbitrary sub-setting, averaging and computing correlations
• it provides a graphical interface allowing the user to plot data in a variety of ways without needing programming skills

Users need not think about the mechanics of making a figure, they merely make choices of variable, location, and time, all properties of the data set, not the software. This allows students to explore how specific properties of data, vary in time and space and permits a sophisticated selection of data in order to render a single picture. Once focused on a region, the user can flip between variables, or change the perspective from a map to a cross section. The user picks properties of the data set or variable, the computational details of preparing the figure are hidden. The same is true for calculations. All computational work is done on a Lamont workstation so students can remotely access and manipulate these data sets with small desk top or lap top machines.

Using the Data Library

The TESY home page http://rainbow.ldeo.columbia.edu/ees/. provides course material used in a three semester, Columbia/Barnard Earth System Science course and a number of large data sets. Clicking on the "DATA" button, near the bottom of the page, accesses a menu of data sets (Table I) and a split screen tutorial. Some aspects of the Lamont Data Viewer's versatility can be illustrated by exploring a nested data set such as the Levitus94 data set of oceanographic properties (Levitus and Boyer 1994). This data set has three space dimensions, latitude, longitude, depth and can be viewed as monthly, seasonal or annual data. It is time averaged e.g. monthly January data is an average of all January data, gathered between 1902 and 1992 (Conkright M et al . 1994, Levitus et al. 1994). These data are grouped in a series of boxes by latitude, longitude and depth. Boxes with little or no data are filled by interpolation from adjacent boxes. Clicking on the hypertext, Annual LEVITUS Temperature (0 meters depth), brings up a screen containing a map of sea surface temperature. A highlighted "references" label leads to data documentation, including publications describing how the data was assembled and processed. Clicking anywhere on this map, leads to another map through which the data can be manipulated (fig 1). Pop-up menus on this page, such as "colors" allow the modification of the data presentation (contoured, colored, colored and labeled contours etc.), or "temp", links to other data sets, nested within the Levitus94 data set.

Exploring data sets

Nested three-dimensional data sets, such as the Levitus94 data set of ocean properties, allow students to draw together diverse measurements and view them in common formats (maps or cross-sections) or perform calculations on them. A map of any data set, at any depth, can be made by merely selecting the data set and typing the desired depth in the box immediately above the map (Fig. 1) and click "redraw". Look for instance at maps of temperature at 200m and 1000m. At two hundred meters the subtropics of all oceans show pools of relatively warm water, a result of convergence and sinking of warm surface near the centers of the great subtropical gyres (fig 2). These warm water lenses have a topographic expression, caused by their relative buoyancy (IGOSS sea surface height data set fig 3). By a thousand meters most of the ocean is filled with cold water, except for some marginal seas where shallow sills prevent exchange with the open ocean. Ocean cross sections clearly shows the connection between this cold deep water, filling most of the ocean and high latitude surface waters (fig. 4). A map view can be transformed, into a north-south cross section by changing the pop-up menu labeled "latitude", to "depth" and the one labeled "longitude", to "latitude" (fig 1). Click anywhere on the longitude line, along which you want the cross section to run (try 150W) (Fig 4). To move any north-south cross section east or west, type a new longitudinal coordinate in the box above the cross section and hit "redraw".

The study of the distribution of chemical species concentrations, in the deep-sea, can lead students to an understanding of the deep circulation of the ocean, as it has professional oceanographers. Nutrients such as phosphate (PO4) and nitrate (NO3), also nested in the Levitus94 data set, are stripped from ocean surface waters by phytoplankton and later returned to deepwater through the oxidation of sinking organic matter. The concentration of dissolved oxygen is high in water recently at the sea surface, but reduced at depth, through the oxidation of this sinking organic matter. Studying the spatial variation of these nutrient concentration's, in maps, at a variety of depths and cross-sections, allows students to follow the movement of individual water masses leading to an understanding of the deep oceans circulation patterns. Phosphate concentrations at 2500 meters (fig.5), for example, show a general increase from the North Atlantic to the North Pacific via the Antarctic reflecting the aging and direction of water movement at this depth. As one would anticipate, the reverse is true for oxygen concentration at the same depth.

The buttons across the bottom of Figure 1 allow a student to find the source of the data: "documented page". Produce a page without buttons or other documentation: "plain page". Link the figure so that if you click on it you will go back to the Viewer thus not only is the figure included in the page, but information/data behind the figure is only a link away. This figure and link can be included in your document by using the HTML code provided.

Any selected data can be downloaded, by clicking the "more options" button and following directions. The "more options" button also leads to ways to change the aspect ratio of the page, change fonts or animate a time or space series. To exercise the animation option, select a time or space series, e.g. the Levitus Monthly Temperature data set (table I). Type the time period, over which you would like the data animated, say Jan to Dec, in the "time" box at the top of the screen (fig 1), and click on the map or "redraw". A movie of seasonal changes in sea surface temperature results. An instructor might want students to determine the depth to which seasonal variations occur. Animating the monthly temperature data through a year at several depths illustrates the answer. The answer obtained in this way is at best a rough approximation. A better estimate can be made by using the Viewer's powerful calculating ability and calculating the regional convergence with depth, of monthly temperature extremes. To do this the student must select a sub set of the global monthly temperature data set. This is done by selecting "Dataset" at the bottom of the page (fig.1). On the page that appears select the measurement of interest, in this case monthly temperature, then choose "select data" and specify the aerial coordinates of the data sub set.

On the next screen select "filters" then from the list of possible calculations select "max XY" or "min XY". On the next screen select "Tables" and a table of the average max or min temperature from a series of depths within the area selected will be displayed.* The same procedure can be followed to obtain the minimum temperatures.

2) The virtual hydro-caster

The virtual hydro-caster, accessible under "tools" in the oceanographic section of the data menu (Table I), lets students sample a vertical profile through any of the data sets nested in the Levitus94 data set. A student can think about where and what data to sample to solve a particular problem just as an oceanographer

can think of where and what to measure, in the real ocean, to test an hypothesis. For example, select a location south of England near the Straits of Gibralter by clicking on the virtual hydro-caster's map (fig 6). Here salinity shows an intermediate maximum at less that one thousand meters (Fig. 6). Typing other coordinates, in the boxes to the right of "Data at.." below the "zoom" button, moves the cast to a new location. A student might want to determine the source of this intermediate salinity maximum.

*These are temperature data averaged over all grid boxes. Since a one degree by one degree grid covers a different area in high than in low latitudes the original data must be multiplied by the cosine of the latitude to correct for this. (Benno would you write how you do this in expert mode just to let the reader know there is an expert mode?)

A series of virtual hydro casts from different locations, within the region, will show that the intermediate salinity maximum strengthens and shallows toward the Straits of Gibralter, suggesting the Mediterranean as its source. The hydro-caster can calculate density from the salinity and temperature data and display the some times complex relationship between these variables (fig. 6). The pop-up menus below "vertical" and "horizontal" allow access to other data sets. Oxygen, for example, in this part of the Atlantic, shows a minimum near one thousand meters and a deep maximum between two and thee thousand meters. This deep oxygen maximum is in the southward flowing North Atlantic Deep Water, formed through deep winter convection of Norwegian and Greenland Sea surface water. The data that lie behind these views can be recovered as tables by clicking the highlighted text "Data in current view" at the bottom of the screen (fig. 6). These data can then be downloaded to an EXEL file, for further manipulation and study.

3) Integration of data sets

Multiple data sets from a given region can be used to gain insights about interactions between regional meteorology, oceanography, biology and geology. The wind patterns of the Eastern Equatorial Pacific (COADS wind vectors Table I) for example, show the tropical wind circulation (northeast and southwest trade winds)1. These wind patterns (fig. 7a) suggest the equatorial surface water should be pushed westward by the prevailing easterly winds.

A north south section of Levitus annual phosphate data shows phosphate isopleths outcrop at the equator (fig 7b) indicating upwelling along an equatorial divergence caused by zero coriolis effect at the equator. Ocean Sediment Thickness data, from Geological Data (Table I) show a band of thickened sediments lying along the equator2. These thickened sediments (fig 7c), composed primarily of the microscopic remains of organisms, are the result of relatively high Equatorial Pacific productivity caused by the return of nutrients to the surface through equatorial upwelling over millions of years.

A map of surface phosphate concentration, in this region (Table I) shows a phosphate maximum along the equator with phosphate isopleths paralleling the equator. The sediment isopachs show a similar pattern except for a neck of thin sediments centered at about 102 degrees west. A look at the ETOPO5 World Topography and Bathymetry data set (Table I) reveals that this neck of thin sediments is centered on the crest of the East Pacific Rise.

1. To look more closely at a given map view or cross-section, a student can use the "zoom" pop up menu (fig 1). To enlarge or reduce the view of interest, select the change desired and either click on "redraw", the center of the map is unchanged, or click on the map to define a new center at the click point. The borders of the new map can be more precisely defined by typing in the desired coordinates in the small boxes near the left-hand border of the screen and to the right of the "HELP" button (fig. 1).

2. The range of a selected variable may be different in different map and cross-section views. The range can be altered by typing in new values in the boxes, below and near the ends, of the variable color bar (fig 1). Here the range of sediment thickness needs to be adjusted to bring out thickness differences in the relatively thin sediments of the open ocean so set the top of the range at 1000 meters works well.

Constructing an east-west cross section of sediment thickness data, centered on the thin sediments at the equator and 102 degrees west, shows a beautifully symmetrical increase in sediment thickness both to the east and west (fig 7d). This symmetrical increase in sediment thickness is due to the symmetrical increase in ocean floor age as the lithosphere moves away from the spreading center. Knowing either the average spreading rate or average sedimentation rate allows a student to use these data to approximate the other.

The thickness pattern, of Eastern Equatorial Pacific deep-sea sediments, is a consequence of the tropical wind system's effect on surface ocean currents, which in turn influence regional surface water nutrient distributions and biological productivity. The near match between high productivity today and thick sediments, speaks to the continuity of these processes over millions of years. The sediment thickness anomaly at 102 degrees West has another cause &endash; plate tectonics. Clearly, the capacity to draw on a variety of data sets and view them as information rich displays in common formats has important educational advantages. How best to exploit these advantages in the classroom is now the major challenge. We provide two examples of what we have done.

Learning from Data Sets

We use a three-semester course to introduce Earth's Environmental Systems to our majors. The semester offerings are; Climate (atmosphere and ocean), Solid Earth and Biosphere (Pfirman et al. in prep). These courses increasingly use the Data Library as a source of information and a vehicle for learning. Interactions with computer databases in class, laboratories and at home, are complemented with lectures, "wet" laboratories, discussion sessions, and field trips. All three semesters are served on the web, complete with lecture notes, laboratory descriptions and homework assignments (the url is http://rainbow.ldeo.columbia.edu/ees/). We will give two examples of how the data Library is used in these courses.

An oceanographic example

After an introduction to atmospheric circulation, we turn to the general circulation of the ocean and its role in climate processes. Armed with knowledge of buoyancy, pressure gradients and the Coriolis effect aspects of the general circulation can be "discovered" by examining relationships between a variety of oceanographic measurements. We do this via a group exercise where students identify features common to each of the major ocean basins (spatial variability), and then discuss the reasons for these similarities

The distribution of sea surface temperature is fundamental to the circulation of both the atmosphere and ocean and is directly related to the basic energy balance of the Earth. This distribution is not a surprise to students; it is part of their experience. Spatial salinity variations are more complicated and students are initially surprised to see relatively low salinity's along the equator. However when encouraged to examine the distribution of atmospheric pressure and/or global precipitation data and think about atmospheric circulation, they can relate high equatorial precipitation to low equatorial salinity values.

In the lab on the following day, students explore exceptions to the general pattern of sea surface temperature and salinity. They generally discover areas of upwelling, or isolated seas that communicate in a restricted way with the open ocean, e.g. the Mediterranean Sea. They are asked to explain why these "anomalies" occur. Using the viewer, they examine deeper map views or cross sections of anomalous areas. In upwelling areas, cross sections show outcropping isotherms indicative of upwelling cool water. If students are then asked to examine nutrient concentrations in the same area, they find that nutrients are brought to the surface by upwelling. The isotherms reveal the depth from which the up-welled water originates indicating that wind induced upwelling is drawn from relatively shallow depths. In this example, each student observation leads to another question, another observation, and the understanding of another process.

The same concepts can be approached in another way. Students can first be taught upwelling processes, then examine the distribution of global winds and make predictions about where upwelling will occur, using sea surface temperature and nutrient data to test their hypotheses.

A Solid Earth example

Here the Viewer is used to explore the distribution of three global data sets; earthquakes epicenters, active volcanoes and topography. The students are asked to use these data sets to identify Earth's major plate boundaries and classify them as either convergent (subduction zones), divergent (spreading centers) or strike slip (transform faults). They then draw these plate boundaries on an inflatable transparent globe, using standard symbols. Upon completion of this exercise, the student is asked to pick a region of the earth to explore in detail. They are asked to let the data speak to them and not be mislead by their preconceived ideas about how the Earth works. They then write a report that must include (1) a summary of their observations, (2) their interpretations and (3) a question for further research.

This assignment includes a closely directed section and an exploratory section. The "find the plate boundaries" section lets students experience both the descriptive power and the ambiguities involved in splitting the globe into mobile plate boundaries and stable plate interiors. Plotting the data on an inflatable globe rather than a 2-D map emphasizes the inter- connectedness of plate boundaries. The figures are from a students' exploration of the Tonga-Kermadec Trench region in the SW Pacific. In the figure above, the students' have reproduced the textbook result that subduction zone earthquakes get deeper and deeper with distance from the bathymetric trench.

Knowing that different mechanisms cause earthquakes at different depths, the students then speculated that earthquake magnitude should vary systematically with depth. The figure at the right shows this is not markedly so and why not? The students propose this as a question for further research.

Conclusions

The ability to easily access and view earth science data can provoke a major change in the way earth science is taught. These data sets contain the data underlying the great scientific discoveries of our field during the last half century, including plate tectonics. Using the viewer, students can explore the primary measurements underlying these discoveries and recreate the discovery process. Learning directly from data has important pedagogical advantages; it is constructivist, learner centered, encourages student inquiry and the practice of the scientific method. It provides a way of learning about the earth that parallels both the history and practice of earth science, aligning the methods of the student with those of the researcher/instructor.

Unlike books, databases have no beginning or end, no natural starting place and no finish line. There are many ways to move through them but no most correct way. Pathways can be tailored to the specific learning style of individual students. While books segregate information, computers integrate it and encourage systems thinking (Richmond 1993, Logan 1995). The computer can bring together, information important to a problem from a number of databases and display them in a common format.

To successfully use the viewer and gain understanding from data sets, students must sharpen their critical thinking skills, analyze spatial and temporal patterns, make and test predictions based on their hypotheses. These uses of the viewer develop computer literacy and enlighten students to the power of the computer as a tool to access and process information. In so doing students exercise their ability to analyze, classify, measure, develop conceptual models and synthesize information.

Never the less the use of real data displays has its challenges. Some students have difficulty transforming the two dimensional computer screen images into three-dimensional mental models. Ways to improve the ability to visualize in three dimensions must be sought to help these students. Many of the thought processes involved in this kind of data analysis are not part of today's curriculum. Consequently instructors must be patient and have realistic expectations about how quickly these thought processes can be learned. They must also seek new and better methods to help students learn them. The results will be well worth the effort, however, for our students will face an increasingly digital world as they enter the next century. Those that have gained the ability to learn directly from data will have placed a very useful arrow in their quiver of life enhancing skills.

References:

Conkright M., S. Levitus, and T. Boyer, World Ocean Atlas 1994 Volume 1:
Nutrients. NOAA Atlas NESDIS 1, U.S. Department of Commerce, Washington, D.C., 1994.
Levitus S. and T. Boyer, World Ocean Atlas 1994 Volume 2: Oxygen. NOAA Atlas NESDIS 2, U.S. Department of Commerce, Washington, D.C., 1994.

S. Levitus, R. Burgett, and T. Boyer, World Ocean Atlas 1994 Volume 3:

Nutrients, NOAA Atlas, NESDIS 3, U.S. Department of Commerce, Washington, D.C. 1994.

S. Levitus and T. Boyer, World Ocean Atlas 1994 Volume 4: Temperature. NOAA

Atlas NESDIS 4, U.S. Department of Commerce, Washington, D.C., 1994.

Logan, Robert K. (1995) The Fifth Language: Stoddart Publishing Co. Limited Toronto, Canada

Pfirman, S. Hays, J. D., Kastens, K., Reibel, J. (in press) Earth System science on the Web: The three semester Columbia/Barnard Experience

Richmond, Barry (1993) Systems thinking: critical thinking skills for the 1990's and beyond: Systems Dynamics Review Vol. 9, no. 2 (Summer 1993): pp 113-121.

FIGURE CAPTIONS

Figure 1

1. Data visualizations can be modified through the use of this pages pop up menus and buttons along the bottom of the page. Data can be manipulated by accessing the data "Get Data" and performing various calculations on it.

Figure2

2. Global ocean temperature map at 200 meters depth. The warm sub tropical waters are the roots of the thick warm water lenses beneath the subtropical gyres.

Figure3

3. The buoyancy of the sub tropical warm water lenses (fig. 2) result in ocean surface water topographic highs. These values are sea surface height anomalies (departures from mean sea surface height), in centimeters, for February 1995.

Figure4

4. North- south cross section from ocean surface to the bottom along 150 degrees west. This clearly shows the ocean is mostly filled with cold water that out crops in the high latitudes of both hemispheres.

Figure5

5. Phosphate concentrations at 2500 meters, showing a continual increase from the North Atlantic to the North Pacific via the Southern Indian Ocean. This increase follows the net flow of water at this depth.

Figure6

6. The virtual hydro caster displaying values for potential temperature, salinity and calculated density for a location near the Straits of Gibralter.

Figure7

7. a) Equatorial Pacific mean wind directions.

b) North-South cross section of phosphate concentration Eastern Equatorial Pacific.

c) Equatorial Pacific sediment thickness

d) East-West cross-section of sediment thickness. The crest of the East Pacific rise is located at about 102 degrees west.

8. a) Earthquake depths near the Tonga-Kermadec trench.

b) Earthquake magnitudes near the Tonga-Kermadec trench.

Table I back

Data Catalog: Datasets by Category