grain size analysis

I started my grain size analysis today.. I’ll be running 20 samples from Portsmouth Harbor that were acquired during the Feb-March 2008 R/V Gulf Challenger cruise. However, since 20 samples are a bit much to work with at once, I’m running them in two batches of 10 samples each. My procedure will follow Larry Ward’s Coastal Geology Laboratory (CGL) methodology, which is based on the tried-and-true sieve and pipette analysis techniques of Folk (1980). The CGL methodology comprises the following:

• removal of organics
• washing/removal of salt
• dispersal (breaking up particle agglomerates)
• wet sieving (separates gravel, sand, and mud fractions)
• dry sieving of sand (>62 microns) and gravel (>2 mm) fractions
• pipette analysis of mud (<62 micron) fraction
• computation of grain size distribution and statistics (using Gradistat; see Blott and Pye (2001)).

Since I’m interested in percentages of carbonate and organic carbon, I’m going to run these analyses as well. The carbonate fraction is determined by dissolving carbonate out of the sample with an HCL solution.. and the organic carbon fraction is determined by loss on ignition (LOI). Normally I wouldn’t run the carbonate analysis, but several samples contain a LOT of shell hash.

For my work, which involves modeling bedload transport, it’s particularly important to know (a) median grain diameter, (b) mud content, and (c) organic content… in addition, of course, to the full grain size distribution. Median grain diameter is used to predict entrainment in many sediment transport models, and both mud and organic content can cause cohesion in sediments, which affects particle entrainment as well.

sed samples

We spent a few days on R/V Gulf Challenger (25-26 February and 12 March 2008) collecting bottom samples and bottom videography in Portsmouth Harbor. There are 26 bottom sample + video stations, and 3 stations with bottom sample only.

I plotted up GPS locations from the bottom samples to see how they looked. Two GPS positions were recorded for each sample (one when the Shipek sampler started descending, and another when it came to surface) using a Garmin WAAS-enabled handheld GPS from the stern of the boat. The plot looks something like this:

Obviously, there is sort of a large difference between the deploy and retrieve positions (represented by green and red dots, respectively). Hmmmm. I consulted with Shachak and he recommended that I interpolate between the two, reasoning that the mean position is probably closer to the true sample location. Here’s what I did:

• convert deploy + retrieve positions from dd.dddd to UTM coordinates
• find midpoint between deploy + retrieve positions (interpolation on XY grid)
• calculate XY distance between deploy + retrieve positions
• calculate approximate uncertainty of midpoint position, choosing the greater of the following values:
  • 1/2 of XY distance (e.g. somewhere between deploy + retrieve positions)
  • 5m
• PLOT midpoint with buffer distance equal to (approximate) uncertainty

And here’s the result: