Table of Contents
- 1. Research Setting
- 2. Primary Research Question
- 3. Investigation Plan
- 4. Data Access and Visualization Methods
- 5. Preliminary Analysis
- 6. Refinement of Analysis
- 7. Statement of Results
- 8. Discussion of Results
- 9. Statement of Conclusions
- 10. Questions for Further Investigation
1. Research Setting
The research setting for this tutorial is the Gulf of Mexico. The Gulf of Mexico is a semi-enclosed oceanic basin that fits nicely between latitudes 18-31 degrees North and longitudes 82-98 degrees West. Texas, Louisiana, Mississippi, Alabama, and Florida are the “Gulf Coast” states, and (as might be guessed from its name) Mexico possesses most of the remaining coastline.
The Gulf of Mexico has one major current system, the Loop Current. The Loop Current flows roughly northward from the Caribbean Sea (where it is called the Yucatan Current) between the Yucatan Peninsula and the western end of Cuba, and somewhere in the middle of the eastern Gulf it “loops” to the east, then flows southward parallel to the west coast of Florida. After it passes the Dry Tortugas, the Loop Current turns sharply to the east, becoming the southern end of the Gulf Stream as it passes through the Florida Straits. The warm water of the Loop Current, and its summertime location, is an important factor influencing the strength of hurricanes. Like the Gulf Stream, the loop of the Loop Current can occasionally form pinch-off eddies that can move around the western Gulf for months after they are formed.
The waters of the southern Gulf are very warm, allowing the growth of coral reefs in many areas. The clear, warm waters of the Loop Current even allow the existence of corals in relatively deep water off of the southwestern coast of Florida.
2. Primary Research Question
The primary research question that is the subject of this tutorial concerns the use of scatter plots generated by Giovanni to examine remote sensing data for the Gulf of Mexico. Therefore, the basic question to be answered is:
How can scatter plots be used to determine what relationships may exist between chlorophyll (chl) and sea surface temperature (SST) in the Gulf of Mexico?
It will soon be apparent that this is a complex subject, and that is why this tutorial is only Part I for this subject.
3. Investigation Plan
The initial investigation plan is straightforward. We will generate SST vs. chl scatter plots for the Gulf of Mexico for two months, January and July 2003. These plots will be analyzed to see what information they can provide.
4. Data Access and Visualization Methods
The system we will be using for this tutorial is Giovanni’s “Ocean Color Multi-Parameter Intercomparison System”. This system allows comparison of several data parameters in a variety of different ways. The system utilizes SeaWiFS data and MODIS-Aqua data, so at this time only a couple of years of MODIS-Aqua data are available for intercomparison. (Future plans include the addition of MODIS-Terra SST data, extending the SST data in the system back to the year 2000.)
Some of you may be asking, “So what exactly is a scatter plot?”
Another term for scatter plot is an X-Y plot. Either way, two variables that are measured simultaneously at the same location are plotted on graph consisting of two axes, X and Y; meaning that the value (y) of one variable is plotted with the value (x) of the other variable. The “x-y” pairs determine a point on the graph. Chlorophyll has a fairly wide range of concentrations in the ocean, ranging from 0.01 milligrams per cubic meter (mg m-3) to over 50 mg m-3 (and sometimes even higher). For that reason, rather than plotting the actual value of the chlorophyll concentration, the base-10 logarithm of the concentration is used. As examples, a concentration of 10 mg m-3 has a logarithm of 1 (10=101) and a concentration of 0.5 mg m-3 has a logarithm of -0.3 (0.5 = 10-0.3).
Below is a randomly-generated example of a SST vs. chl scatter plot. One point (somewhat removed from the main conglomeration, which is sometimes called an outlier) is used to demonstrate x-y plotting. The x value is 0.4, which corresponds to about 2.5 mg m-3 chl. The y value is about 9.5 degrees Centigrade (°C).
5. Preliminary Analysis
Knowing what a SST vs. chl scatter plot looks like, we can now begin our initial examination of the Gulf of Mexico. The first thing to do is define the research area, as shown below for January 2003. [All of the other figures in this tutorial are linked to full-size figures; click on the figure to see the full-size version.] The contours shown are SST contours (another feature of Giovanni's intercomparison capability). The general shape of the Loop Current this month is well-defined by the 24 °C SST contour line in the eastern Gulf.
This is a somewhat familiar presentation of the data. So what does the scatter plot look like?
In the scatter plot above, three data clusters have been circled. Cluster 1 (light blue) is the main cluster containing most of the data points. Cluster 2 (light green) and Cluster 3 (light pink) are other groups of data that are a bit separated from the main cluster.
6. Refinement of Analysis
Viewing the scatter plot above should invoke an immediate question: what is the significance of these data clusters to the Gulf of Mexico? We will surmise that the clusters represents different water environments in the Gulf. Cluster 1 shows a fairly significant relationship between SST and chlorophyll, with higher chlorophyll concentrations corresponding to lower SST values, where SST ranges from 12-24 °C. Cluster 2 is characterized by the highest SST values and the lowest chl concentrations, while in Cluster 3, the SST is relatively constant between 21-24 °C, with chlorophyll values varying over a significant range.
Looking at the area plot of the Gulf of Mexico shows that the lowest SST values occur along the northern Gulf Coast. So now we will create an area plot of this region, and a corresponding scatter plot. The area plot is centered on the Mississippi River delta area.
Comparing this scatter plot to the scatter plot for the entire Gulf of Mexico indicates a strong similarity to the data distribution in Cluster 1. As SST decreases closer to the coast (closer to the continental land mass that is cooled by winter weather), chlorophyll concentrations increase. This is due to the greater availability of nutrients near the coast, and also due to the effect of sediments and turbidity, which can cause remotely-sensed chl concentrations to be measured higher than they actually are. Despite this problem, the scatter plot for the northern Gulf Coast indicates that most the data distribution in Cluster 1 is primarily a “coastal” data cluster. As the northern Gulf Coast and the west coast of Florida are large, broad areas, they contribute a large number of data points to the scatter plot for the entire Gulf.
The next data to be examined is Cluster 2, the cluster with the highest SST values and the lowest chlorophyll concentrations. A quick look at the first area plot for the entire Gulf indicates that the lowest chlorophyll concentrations occur in the middle of the Gulf. So now we select a big square area in the central western Gulf of Mexico, which provides a fairly uniform area plot of chlorophyll values, and an interesting scatter plot.
It isn’t hard to tell that the data distribution for this scatter plot is very similar to the data distribution in Cluster 2. Thus, the warm waters in the central Gulf of Mexico have characteristically low chlorophyll concentrations, and a distinctly different X-Y scatter plot than the coastal zone.
Now we will turn our attention to Cluster 3. In this cluster, the SST is fairly warm, though not quite as warm as in the central Gulf. The chlorophyll concentrations range from 1.0 to nearly 10.0 mg m-3. The increased concentrations are very likely not the low values found in the central Gulf of Mexico; they are more like coastal concentrations. So where in the Gulf of Mexico is there a combination of warm waters and high coastal chlorophyll values?
Our thoughts, like the asteroid that might have hastened the demise of the dinosaurs, should quickly slam onto the Yucatan Peninsula. For the third time, an area plot and scatter plot for this region can be created.
This scatter plot contains the same data distribution as Cluster 3, and also an area of uniform SST and low chlorophyll values. The area plot shows why this occurs. The Yucatan Current area to the east of the Yucatan Peninsula has low chlorophyll values and warm SST, similar to the central Gulf. Along the northern and western coast of the peninsula, increased chlorophyll concentrations are observed, in an area that also has warm SST. Thus, the northern and western coast of the Yucatan Peninsula appear to be the main source of the Cluster 3 data. The water is slightly cooler due to its proximity to land, which cools off in winter a bit despite being in the tropics. The northward extension of elevated chl values into the Gulf also indicates the possibility that winds may have caused local upwelling or mixing, bringing cooler, deeper waters to the surface and increasing productivity with a dose of nutrients.
7. Statement of Results
This use of SST vs. chl scatter plots indicates that they can be used to define different water environments in the Gulf of Mexico. The northern Gulf coast, the central Gulf, and the coast of the Yucatan Peninsula possess differing SST and chlorophyll concentrations that can be distinguished in the scatter plots. During the month of January 2003, these scatter plots provide a simple way to visualize relationships between SST and chlorophyll concentrations in this region.
8. Discussion of Results
Attentive readers may remember that the “Investigation Plan” indicated that two months would be examined, namely January and July 2003. So far, the discussion has only examined January 2003. There is a good reason for this.
As the waters of the Gulf of Mexico get warmer during the summer, it becomes much more difficult to distinguish different water environments on the basis of temperature, because there is very little variation in SST. Both the area plot and the scatter plot shown below demonstrate this fact.
The scatter plot shows that summer SST values are in a very narrow range, between 25 and 32 °C. These warm SSTs certainly contribute to the power and destructiveness of hurricanes, but they erase the ability to distinguish coastal areas from open ocean areas. Chlorophyll concentrations are still clearly low in the central Gulf and higher in the coastal zone, but that’s about all that can be said. So the usefulness of SST vs. chl scatter plots is definitely influenced by the location and the season.
However, when oceanic environments can be defined on the basis of co-varying values of chlorophyll and SST, scatter plots are a very useful and rapid way to determine the characteristics of a given environment. Because SST and chlorophyll are both related to other vital oceanic variables, the relationships found in scatter plots can be indicative of the dynamics of physical and biological ocean processes. Scatter plots provide a graphical representation of these dynamics. In the Gulf of Mexico, the seasonal interaction of land and sea can be perceived in the scatter plots.
9. Statement of Conclusions
In this tutorial, it has been determined that SST vs. chl scatter plots can be effective diagnostic tools to distinguish various oceanic environments. For the Gulf of Mexico, these scatter plots indicate:
- In the winter, the Gulf of Mexico central region is characterized by warm SST and very low chlorophyll concentrations.
- In the winter, the northern Gulf of Mexico coastal region is characterized by an inverse relationship between SST and chlorophyll; i.e., lower SST is correlated with increased chlorophyll concentrations, due primarily to the occurrence of colder water temperatures near the coast.
- In the winter, the Yucatan Coast of Mexico has elevated chlorophyll concentrations and warm SST on the northern and western coasts, and low chlorophyll concentrations with warm SST on the eastern coast.
- In the summer, the warm SST values found in the entire Gulf of Mexico region make SST vs. chlorophyll scatter plots less effective for distinguishing various oceanic environments.
10. Questions for Further Investigation
This tutorial may have provoked a lot of questions, some of which will be addressed in Part II. Below are three questions that can provide starting points for independent application of Giovanni’s SST vs. chlorophyll scatter plot function.
1. How do SST vs. chl scatter plots for a selected region vary over the seasons (where there are significant seasonal changes, of course)?
2. Where are the strongest positive correlations between SST and chlorophyll concentrations found?
3. Where are the strongest negative correlations between SST and chlorophyll concentrations found?