Painting a picture with data isn’t simple

The series of 21 paired maps offers a great year-by-year comparison, but I also wanted to see how the data from the entire time series matches up. I created interpolation maps using ArcGIS to view spatial trends on a map. I used a method called kriging to fill in the empty areas where I did not have data points. This assumes auto-correlation, the idea that things close to one another are most similar. The result is a best guess of the average Tmax and salp density for the entire WAP sampling area, not just the exact locations where researchers have collected data. This certainly has errors, particularly further away from sampling stations, but it can offer insight into general trends.

Screen Shot 2015-08-28 at 5.59.44 PM

So what does the figure above tell us about Tmax? The black dots are the shelf sampling stations. These really are the focus of my question, because these locations are where UCDW is occasionally present. The map shows that Tmax values are higher to the west. This makes sense because we know that warmer UCDW is always present over the continental slope. The highest temperatures on the shelf are found on the 300 line, where the 1.67 to 1.84 degree Celsius contour covers two sampling stations.

And how about the salp map? Warmer colors are focused in the north and to the west. These areas have the most favorable conditions for Salpa thompsoni: warmer water and limited sea ice. The most obvious deviation from this trend is on the 300 line shelf stations where predicted salp densities are higher than those immediately to the north and south. Therefore, the area where UCDW intrusions appear to be the most common also seems to be where salps are more abundant than would be generally expected.

This has me excited, but I’m not totally sold yet. First off, there are many ways to display this data. I asked for help and read up on kriging methods but by no means am I an expert on the many choices you make when running this kind of analysis. And then I had to decide how to set the breaks for the contours on the maps. Every time you change the cut-off mark between two different colors you get a slightly different product. Once again I turned to help sections as well as people who know what they are talking about, and did my best to follow their guidelines. The trends I mentioned above did seem to be consistent with multiple different methods of presenting the data.

These types of interpretation and presentation questions are issues I had never considered before. It takes much more work than I realized to communicate a message about your data. I am beginning to see how all of the decisions that you make when manipulating numbers are extremely important. Understanding the methods section of a paper and comprehending figure captions can be challenging, and now I’ve had a glimpse at why they need to be extremely detailed.