June 9, 2004
Caryn House, WHOI

Broad Scale Proposal Gridding Discussion

Attending: Peter Wiebe, Ted Durbin, Maria Casas, Dicky Allison, Nancy Copley, Robert Groman, and David Mountain (Rapporteur).

The broad-scale gridding group met in Woods Hole, Ma at the Caryn House from 1000 until ~1330. The purpose of the meeting was to review progress in gridding the data sets from the 30 GLOBEC Georges Bank Program broad-scale cruises to a standard grid. Topics that were discussed included:

  1. Extraction of data from the Database. What is most efficient way to do this?
  2. What data sets to focus on out of a rather large list of possible variables?
  3. What spatial persistence or correlation space scales to use in the kriging of the data?
  4. How will produced mapped data be stored/served?
  5. How will mapped products be displayed/served?
  6. Establish protocol for doing above that others can use?

The standard grid now has 2385 points distributed so that it encompases all of the broad-scale stations. Information about the grid and a file containing the data points is available on the U.S. GLOBEC website at: http://globec.whoi.edu/globec-dir/phase4doc/grid/grid_report.html.

The following is a summary of the discussion.

1. Extracting data from the GLOBEC database for doing gridding:

This process is encountering some snags. The download routines needed some revision. This is being worked on. A complicating factor was that the JGOFS/GLOBEC data management system had recently been switched to a new faster machine, but that systems software generated new problem in serving data with the result that the older machine was returned to service.

2. List of parameters to be mapped:

A list had been developed earlier. We went through that list to review the status of mapping the various data sets.

Bathymetry - The group is currently using the 0.05 lat/lon gridded data set (from Greenberg or from USGS need to determine if these are the same). The modelers are trying to develop a finer resolution bathymetery data set for their needs. Since we are using the same grid points for our broad-scale mapping grid, we will stay with the 0.05 bathymetry.

Temperature - Nancy doing vertically averaged from CTD and surface from along track survey by survey. Ted has done a 5-year average of the 0-30m temperature by month. Nancy will do the whole water column average, 0-15m average and 50-100m average for each survey. The same will be done for salinity and density. With the density values, the difference between the two layers also will be calculated and mapped as a stratification index.

Chlorophyl -Ted has done fluorescence to-chl a regressions and then used the fluorometer profiles to develop a vertical chlorophyll profiles. He will calculate vertically averaged chl a for 0-50m and 0-bottom. These could then be mapped.

Nancy has been mapping the original, actual chlorophyll data at the 3 sampled depths. Nancy also did the along track fluorescence data. The Greene Bomber fluorescence data shows some different features than the ships along track. That data set also should be looked at as well.

Nutrients - We need to determine what would be the most useful way to display this data. Probably mapping the values at the discrete depths sampled would be best. We need to talk to David Townsend about this and about what he has already done.

Carbon (for 1999 only - POC data collected by David Townsend) - since this was collected in only one year, this will have a lower priority for mapping activity.

Currents - Charlie Flagg has calculated detided currents for some, if not all, cruises. These will be good data sets to map. We need to get Charlie involved in this. Chen may have modeled current fields for each survey in 1999 (and 1995?) that can be mapped. To do the other years we may need to look for support under Phase IVb.

Backscatter - Peter is doing this and should have things to show in a couple of weeks.

Satellite data - since the imagery has finer data than our grid, the issue is to extract values for our grid points. We need to talk with Jim Bisagni about this.

MET - the meteorology data sets are not well suited for the survey by survey mapping, given the high temporal variability (e.g., day-night, storms), so at this time no effort will be spent trying to map them.

Zoo biomass - Peter is doing the bongo displacement volumes for each survey. Ted will do volumes from the 0 net on the MOCNESS.

Species counts -

For zooplankton Maria has done 5-year means by month for a number of species, and survey by survey for a number of stages, as well. Maria will send a list of what has been done. Ann Bucklin is doing Pseudocalanus by species.

For ichthyoplankton, David has done the cod and haddock eggs and larvae for all surveys. The eggs have been done for both the early and the late stages. The larvae have been done as total abundance and abundance by 10 day age cohorts.

For predators, Ted is looking at the predators caught in the 150 mesh MOC-1 nets. The predator groups MOC-10 data has been received by the Program Service Office, which is putting it into a format consistent with the GLOBEC database.

Derived rates - Jeff Runge has done egg production estimates for Calanus, which should be mapped. Mark Ohman has estimated mortality rates for different stages of Calanus based on a vertical life table approach. The mortality group also has estimated consumption rates by different predators. Each of these estimates is done on a spatially explicit form that could be mapped.

3. Using average spatial statistics (i.e., an average variogram) is felt to be the best approach to krigging the survey data sets. Whether to include anisotropy or not is another issue to which more thought should be given. We should write up a couple of paragraphs on what we have done and why in mapping our data. As a test of how sensitive the gridding is to individual choices, we will all analyze a common data set to see how variable our results are. Ted will send a couple of zoo data sets to do the test.

4. The format for storing the gridded data is needed. An animated discussion took place about how best to store the gridded data in the GLOBEC database. We decided to let Bob think on this for a while. Ted will send him an example set of gridded data.

5. For storing all of these outputs, we would have both the gridded data files and plots that have been made from them. The plots can be stored as jpg or png files, and be served by the GLOBEC website. However, we also will want to have an on-demand capability to generate a plot from the grid files. This capability is in development by Bob Groman and the Program Service Office. These will be more consistent between data sets, since all of the plot settings (lat, long, scale, etc) will be the same.

6. Some poster presentation should be made at the upcoming workshop of what this gridding activity has been doing. Peter will do a draft layout of what we could do.

Additional point: Ted suggested revising the current grid being used to assign the nodes in sector 3 (northeast peak region) that are >150m depth to the deep offshore region (sector 5). Everyone agreed that this should be done.