G. Law - Environmental Variability

Interdisciplinary Fisheries and Coastal Ecology Research at Rutgers University, Cook College Campus, NJAES

C. Grant Law • Environmental variability

A significant part of my dissertation focuses on environmental variability, which has required two tough tasks -- finding enough environmental data to be able to see and characterize variability, and finding some intuitive way to express it in both a temporal and spatial context. Understanding spatial differences in a given variable is most easily accomplished with a simple map (at least for me), so I tend to focus on map building. For temporal variability I simply stack maps, building a three-dimensional matrix. I can then perform statistical analyzes on each column vector (essentially a time-series for the variable at a given point on the map), and plot the results as a new map.

The first dataset I began working with is the NCEP/NCAR Reanalysis Package. Specifically, the surface pressure values and the 10 meter, 6-hour projected winds. The 10 meter winds were chosen for their compatibility with the physical circulation model (Quoddy) I am using for larval transport studies. Both data sets have a temporal resolution of 6 hours. In an effort to familiarize myself with wind patterns, I built animations of the winds displayed as vectors. These animations (as well as common experience) demonstrate clear spatial, inter-annual and intra-annular variability in winds. The animation linked from the image to the right displays Reanalysis winds in red. The blue vectors are the Reanalysis winds interpolated to the domain grid used in Quoddy. The colored background indicates surface air pressure (sorry for neglecting a colorbar).
Visualizing wind variability has been tough. Progressive vectors do a pretty good job. This is a progressive vector diagram of the winds in 2000. Note the significant shift in the westerlies to the northeast, then the south in the summertime through the Mid-Atlantic Bight.
This is another animation showing progressive vectors for months 1 through 12 (January-December). If the sequence moves too fast, try using the slider. To generate these figures, each month's mean wind vectors from 1948 to 2004 are layed end to end. While watching the sequence, notice the year to year consistency in the pattern. Interannual variability in this pattern is visible as "wiggliness"in the vectors.
I have also been interested in the correlation between winds and the North Atlantic Oscillation index (NAO), the anomaly in the air pressure difference between Iceland and Lisbon (or the Azores). This figure shows the correlation between NAO and wind magnitude from 1950 to 1975 -- a period of generally low NAO indices.
This figure shows the correlation between NAO and wind magnitude from 1975 to 2004 -- a period of generally high NAO indices.
This figure shows the correlation between NAO and wind direction from 1950 to 1975. Red colors mean that winds rotate counterclockwise from the mean wind direction (small black vectors) with increases in NAO. Blue colors indicate clockwise rotation.
This figure shows the correlation between NAO and wind direction from 1975 to 2004.
I have also compiled a large collection of hydrographic data for analyzing hydrographic variability. Unlike the wind data, which comes on a regular grid, the hydrographic measurements occur at irregular points in space and time - whenever and wherever an oceanographer happened to have dropped gear into the sea. Any spatial or temporal analysis must then attempt to fill in the spaces/times in between measurements (this is done with the wind data as well, but the resolution of the Reanalysis data is high enough that linear and squared-distance interpolations do a reasonably good job). To interpolate the hydrographic data in three dimensions, I am using OAX 5. The method used by this objective-analysis software takes into account the physical environment when filling in the holes between data points, so it produces more realistic results than a straight linear interpolation. To do these interpolations I must choose a collection of data points to be integrated in the analysis. Because hydrographic data is fairly limited in coverage, most researchers build climatologies by incorporating all data points collected in a given span of time (all June data on record, all winter data, data from the entire year of 1993, etc). The resultant output is essentially an average for those data. Because I am interested in temporal variability, I must try and limit the time span of data to incorporate in the objective analysis. This gets more and more difficult as you go back in time, when less hydrographic data was being collected. There's also significant differences in coverage from month to month (apparently, oceanographers get more work done in the summer months relative to the winter -- I wonder why?) I've had reasonable success using a time span of two months, but I'll monkey around a bit with that timespan in the future. Because I like to work with 2 dimensional maps, and because the species I am interested in is benthic (the sea scallop Placopecten magellanicus), I have focused on the bottom conditions of temperature and salinity. Again, to get a sense for how things vary spatially and temporally, I assembled an animation of the seafloor hydrographic data. While watching the animation, pay attention to the summer development and fall destruction of the cold pool on the Mid-Atlantic Bight shelf. It has long been believed that this process is an important component of the natural history of the shelf ecosystem, but direct mechanisms are only now beginning to be understood.
I am also interested in understanding how North Atlantic shelf water varies from year to year. I can use my objective analyses to address this question, but I need some way to differentiate shelf water from all the rest. Traditionally, this is done by evaluating how temperature and salinity vary relative to one another. In this set of figures (again this is a quicktime movie you can scroll through) I have plotted temperature vs. salinity for each reliable point in the domain for every objective analysis in the set (1980 through 2004). I also include a color key to identify where the plotted points come from, and spatial plots of temperature and salinity for the broader context. The red line is a rough estimate of the T-S flavor of shelf water. Shelf water can be identified in the objective analyses by finding those points in the data sets lying within some threshold distance from this line.
These figures display the average salinity and temperature for the top 10 meters of the water column.
These figures display the average salinity and temperature for the top 10 meters of the water column.
These figures display the salinity and temperature anomalies for the top 10 meters of the water column.
These figures display the stratification index, defined as the densities at 50 meters (or the seafloor where the bathymetry is less than 50 meters) minus the surface densities.
These figures display the stratification anomalies.

RU COOL • Cook College • NJAES • RUMFS • HSRL