LOCUS Tutorial Research Project Three:
Monthly Patterns and Influence of Missing Data Near Monterey Bay, California

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 Pacific Ocean adjacent to the central coast of California and Monterey Bay.

The primary oceanographic feature in this area is the California Current, a cold-water current that flows from north to south along the coast of California. The current interacts extensively with the coast, causing the occurrence of periodic upwelling events that are observed as "jets" that originate near the shore and extend a large distance offshore in just a few days. Because of the highly variable nature of the California Current, it has a fairly complex structure in ocean color images.

This SeaWiFS chlorophyll concentration image, provided by the University of California - Santa Cruz, shows a "jet" south of Monterey Bay, the semi-circular bay just south of 37 degrees North. The southern end of San Francisco Bay can be seen at the top of this image.

2. Primary Research Question

In this tutorial, we are examining an area that has moderate to heavy cloud cover at times, especially in the winter, to determine what effects, (if any) can be perceived in the monthly average SeaWiFS ocean color data that is used in Giovanni. (Note: in the future, Giovanni may include data products with higher temporal resolution; see the final section, "Questions for Further Research", for more on this possibility.) This area also has a large amount of variability in ocean color, as noted above. So the primary research question can be phrased:

Does the persistent presence of cloud cover affect the quality of ocean color data in an area which has a large amount of variability?

3. Investigation Plan

For this investigation, we will utilize 9km monthly SeaWiFS chlorophyll concentration data that are available in Giovanni. We will examine patterns over several years, one year, and one month, to see how the variability of ocean color data in this region appear in the chlorophyll concentration data. We will also look at some of the daily images of the area from one month to get an idea of how cloud cover may influence the data.

4. Data Access and Visualization Methods

The SeaWiFS 9 km chlorophyll data are processed into monthly files containing the average chlorophyll concentration for each 9 x 9 km "square" area over the world's oceans. Giovanni accesses these files and provides the capability of selecting areas of interest for examination. Giovanni can be used to create a map of the chlorophyll concentrations for the area averaged over selected time intervals, concentration vs. time plots for chlorophyll concentration (averaged over the entire selected area), month-by-month animations of the data for the selected area, or Hovmoller plots for the area.

Hovmoller plots can be particularly useful data visualizations to detect variations over time and space. Hovmoller plots display data in a time vs. longitude or time vs. latitude format. Thus, for the same area, a comparison of the spatial variability over time is easy to comprehend.

5. Preliminary Analysis

The first thing to do is to define the area and then see what the data look like averaged over a single year. So we will define the area around Monterey Bay using a box with a northern latitude boundary of 37.5 degrees North, a southern latitude boundary of 36.0 degrees North, a western longitude boundary of 124.0 degrees West, and an eastern longitude boundary of 121.0 degrees West. Then we request Giovanni to plot the chlorophyll concentrations for the year 2003 for this region:

Monterey Bay is the semi-circular bay approximately in the center of the image. This image is a good start, but looking at it, only the concentrations close to the shore are higher than 2.5 milligrams per cubic meter. So now we can customize the color palette to see what the same plot looks like with a palette ranging from 0.1 to 2.5 milligrams per cubic meter:

This palette choice gives a good indication of the variability at a distance offshore, but because the concentrations are higher near the shore, they are all swallowed up in the magenta color that indicates concentrations higher than 2.5 mg per cubic meter. So let's try an alternative palette that is better suited for the higher concentrations nearer to the coast:

In this image, the variability of the lower concentrations offshore is lost in the purple haze, but the variability nearer to the coast in the higher concentrations is clearer. These images illustrate the importance of tuning the color palette to see features and patterns of interest. We'll return to this point a bit later.

Now that the area and the data have been visualized, let's look at the patterns of productivity using Hovmoller plots. Since we've just looked at the average concentrations over 2003 in an area plot, we'll use a Hovmoller longitude vs. time plot for 2003 next:

Because the California coastline is aligned north to south, the longitude vs. time plot is a little easier to interpret. The angle of the coastline actually allows us to examine the temporal dynamics of two different areas in the same plot! The area from 122.1 W to 121.8 W is primarily the Monterey Bay area; the area from 122.4 to 122.7 W is further up the coast, just south of the mouth of San Francisco Bay. Looking at the plot above, and the previous area plot for 2003, it appears that there's more productivity near the mouth of San Francisco Bay than around Monterey Bay.

Now let's look at the Hovmoller plot for several years, from January 1998 to December 2003:

This plot indicates that 2003 was a bit strange; there is actually a lot more "action" in the Monterey Bay region than the region south of the San Francisco Bay outlet. There appears to be a basic pattern of lower productivity in the winter, and higher productivity in the spring, summer, and fall, generally highest in the summer. This could be due to several factors, such as warmer temperatures, more sunlight, less clouds, and possibly some strong wind-mixing events similar to what occurs in the Gulf of Panama. Examining the possible causes of this pattern would be interesting, but that is not the main goal of this tutorial.

6. Refinement of Analysis

In this section, a closer examination of data from one month - December 2003 - will be performed, and this will also include a closer look at the type of data Giovanni is using for analysis, a mapped monthly average data product of SeaWiFS chlorophyll concentration data.

Below is an image of the data for December 2003, produced using the palette that was employed above for the Hovmoller plots.

Not very interesting!! So now we'll try the default palette to see if a better choice can be determined:

This plot is considerably better. And it shows that there are only some small areas with concentrations above 2.5 mg per cubic meter, so now we can try the first color palette that we used earlier:

It's hard to tell whether the default palette or the customized palette provides more information in this case. In either case, it can be seen that there is a pattern of higher concentrations near the coast, and lower concentrations offshore, with some small patches of higher concentrations. And this plot looks a bit scattered, doesn't it?

There is a very good reason for the scattered appearance of the December 2003 plots. And the reason is: CLOUDS. December is a particularly foggy and cloudy month off the coast of California, so SeaWiFS does not observe the surface of the ocean every day. Because of the clouds, SeaWiFS may not observe any of the region on a given day, or it may only observe parts of the region.

The images below are called "browse" images: they allow researchers to take a look at the data for a given day at low resolution, to determine if they would like to get the data file for that day. This type of data file can be processed into the highest possible spatial resolution available for SeaWiFS data, which is 1 kilometer at the center of the instrument's scanning swath (also called the nadir point of the swath). Although SeaWiFS observes a region almost every day, on some days the region is near the center of the scanning swath, and on other days the same region may be near the edge of the scanning swath. All of these factors affect the quality of the data.

Not every browse image for December 2003 was chosen for this tutorial; these images provide an overview of the types of cloud conditions that were occurring during the month. The letters "MB" were placed on each image close to the location of Monterey Bay. In some of these images Vancouver Island can be seen at the top (north); in other images the Baja Peninsula can be seen at the bottom (south).

December 3, 6, 7, and 8 (left to right): Fairly clear views of the region on the 7th and 8th, but on the 3rd only the area south of the bay was visible, and on the 6th only a small area offshore.

December 12, 14, 15, and 17 (left to right): December 14 and 15 were pretty good days to see this region, while December 12 and 17 were rather poor.

December 18, 25, 27, 29 (left to right): Only a small region was observable south of Monterey Bay on the 18th. Broken clouds were present on the 25th, and clear observations were made on the 27th. Two days later on the 29th, the region was entirely clouded over.

So what does this mean for the quality of the data? The SeaWiFS Project produces monthly average data products used by Giovanni through a process called "binning". A simple explanation for binning is that any observation for a particular spot on the ocean surface will be placed in the bin for that spot: the bins are 9 km by 9 km, but the observations are at a higher spatial resolution. A daily "bin" will thus hold a few observations (note that SeaWiFS observes some places more than once a day at high latitudes). An 8-day bin will hold all of the observations acquired over a period of 8 days, and a monthly bin will hold all of the observations acquired in a month. To produce the data in the monthly products, all of the observations in the bin for a given month are averaged together.

In an area with high variability in chlorophyll concentrations and a considerable amount of cloud cover, this process means that a high concentration "feature" could either dominate the average, or be nearly lost in the average. If a high concentration feature was observed on one or two days, and then clouds covered that area for the rest of the binning period, then the high concentrations would dominate the average. Alternatively, the high concentration feature might only be observed once, and then a lot of other days in which the feature was absent could have provided data, which would substantially "dilute" the contribution of the high concentration feature in the monthly data. In most cases, the binning process will reduce the contribution of a short-lived feature, thus reducing the apparent variability in that region; in a few cases, the binning process could be dominated by observations of a feature if very little data were acquired when the feature was absent.

For December 2003 in the Monterey Bay region, it appears that the data acquired on December 7 and 8, December 14 and 15, and December 27 -- only five days in the month -- provides most of the data that Giovanni analyzed for this month. (There may be data from a few of the other days, of course.) But the substantial gaps in the data caused by cloudiness mean that most of the connectedness of features offshore of Monterey Bay will be lost, which probably accounts for the scattered appearance of the area plots.

If it was possible to see the region clearly every day, we might see the progression of a "jet" such as the one shown at the begining of the tutorial, where a productive area develops near the coast and then moves offshore. SeaWiFS may only observe such a jet on one or two days, so it won't capture the movement of the jet, only where it was at a particular time.

7. Statement of Results

This tutorial has demonstrated that the monthly data products used by Giovanni provide an averaged view of chlorophyll concentrations in a given region. If the region has considerable variability in chlorophyll concentration and is affected by cloud cover, the averaging (binning) process will tend to reduce the spatial variability in the data. This means that features which only last a few days to a week may be difficult to discern in the monthly data.

The monthly data are therefore more suitable for the examination of patterns in chlorophyll concentration that occur over seasonal, annual, or multi-year periods.

This study determined that the Monterey Bay region exhibits a fairly regular pattern of variability, particularly showing reduced chlorophyll concentrations during the winter months and elevated chlorophyll concentrations in the summer months. The spring and fall months had intermediate chlorophyll concentrations between the low concentrations of winter and the high concentrations of most summers.

8. Discussion of Results

While this tutorial showed that cloud cover can substantially reduce the number of observations of a given region in a month, it also shows that a cloudy month still provides enough data to give a general characterization of the chlorophyll concentrations in that region. Therefore, use of the data for the observation of patterns over seasons and years is still quite possible, even for regions with a lot of cloud cover.

This study also indicates that in order to see features and "events" in chlorophyll concentration data that last for periods of time considerably shorter than a month, data products with higher temporal resolution should be used. Section 10 provides a few questions related to this topic.

9. Statement of Conclusions

• The Monterey Bay region shows a consistent seasonal pattern in chlorophyll concentration, with higher concentrations in the spring, summer, and fall (highest in the summer), and lower concentrations in the winter.
• Cloud cover likely affects the resolution of short-lived features in this region, but is does not appear substantial enough to invalidate the use of this data for the examination of seasonal and multi-year patterns of variability.
• Color palettes must be chosen with care to allow the best visualization of chlorophyll concentration data.

10. Questions for Further Research

A few questions related to this tutorial:

• The Hovmoller plot for January 1998 to December 2003 shows that the year 1998-1999 looks significantly different than the other years. What event was occurring at this time that could have influenced this region?
• What is the primary cause of the increased chlorophyll concentrations, i.e., higher phytoplankton productivity, in the Monterey Bay region? What data indicate that this is the primary cause?
• Will higher resolution SeaWiFS averaged data products, such as the 8-day data products, show the occurrence of short-lived features in the California Current system?
• Are jets in the California Current system caused by variability in the flow of the current, strong winds coming off the coast, or both? What data can be used to determine the relationship between winds and jets, if a relationship exists?