Research

Changes in ice floe size distribution in the western Weddell sea

The project will seek to determine changes in the sea ice floe size distribution along the axes of a floating buoy array deployed in the western Weddell Sea during summer 2004/05, and investigate possible interactions between floe size and observed deformation of the buoy array.

Floe size distributions will be obtained from digital photographs taken from helicopter overflights during the period that the buoy array was deployed, using digital image analysis software. An algorithm developed by Matt Paget [IASOS honours, 1995] is the primary candidate for image analysis, since it was developed for, and has been successfully applied to, similar images. However - given that the software intended for use was written in 1995, part of the initial research for this project will be an investigation into the capabilities of more modern GIS software for the task at hand.

As part of the image assessment, the project will attempt to refine the method used to calculate ice floe size distributions to account for incomplete floes at image edges. Using present methods, partial floes are discarded in analysis, giving an incomplete picture of floe size distribution. Investigations into probabilistic solutions will be made, along with image combination and scaling. For example, using a larger-scale satellite image where floe size exceeds the digital image capture area. The possibility of applying the floe size distribution algorithm to satellite imagery will also be explored.

Once a reliable estimate of floe size distribution for each series of images is obtained, the project will explore interactions between floe size distribution, buoy array deformation, meteorological and oceanographic conditions at the time.

Possible questions include:

what relationships exist between floe size distributions and array deformation?
how do meteorological conditions affect the floe size distribution along each axis of the array?
how do changes in the floe size distribution fit with observed changes in the abundance and distribution of ice types and open water?
what conclusions can be drawn about the summer decay of sea ice?

Overall, the work will aim to describe and explain some of the processes that drive changes in pack ice conditions, in particular changes in ice type and concentration, open water fraction and floe size. In doing so, the project aims to improve some of the tools available for future research in the area.

Progress to date

With two months to go, life is full of late nights. Here's what's taken up my time so far:

Getting the FSD software running took over two months, during which time I learned to use IDL and then Matlab to dice up my weather data. I've now got 10-minute weather transmuted into hourly samples, after going overboard with a 3-hour running mean, and have daily samples to boot.
The first step after the software was running was using new images, with different properties. Some parts of the program were rehashed, and now it can use any RGB or greyscale TIFF. I also had to confim that the program got the same results, and had to dig up the old data. Everything chips away...
Now, I have half of three chapters underway, and I'm getting through analysing some images! Painfully, some data will be thrown away since the FSD software can't tell ridge shadows from leads on sunny days. Hmm.
...and lets not forget drift buoys - I just made my first map of the buoy drift. Now, I have to make daily markers. Back off to GMT to learn how!

Results

I think it will be fair to say that we will find some differences in floe size distribution, but no more on that yet. The main results are, to date:

Adam learns IDL, matlab, GMT and a spot of R.
FSD software is running and will be a tad easier to use in future
The ISPOL weather data is in shape and plotted up for eyeballing