Turning Pixels of Antarctica into Numbers (WEEK 2)

On my last blog post, I discussed how I went about collecting data. Now that I have a set of satellite images, I will discuss how I have pulled out quantitative data from them.

First of all, I downloaded each MODIS satellite image as a GeoTIFF file. I then pulled up each image on ArcMap.

Rift Width Measurements

Below is the satellite image from November 6th, 2012.

Image 1

By measuring the “rift width”, I specifically focused on the “tail” on the left, or west, of the rift. The width measurement above is outlined in yellow, which I added on using the draw toolbar and line feature.

Upon double-clicking on the line, a pop-up box appeared with the line’s length, start point, and end point. The lengths are available in metric units, English units, and even decimal degrees. The start and end points are available in metric units, English units, decimal degrees, and DMS degrees. For consistency, I recorded each measurement in meters and decimal degrees, as appropriate.

Image 2

Although I would have also liked to obtain data about the measurement’s angle, I was unable to do so because the map data is projected from a 3D surface onto a 2D surface, and the line is not straight.

Additionally, although it may be hard to see when the image is that far zoomed out, there is indeed some room for error when doing these measurements. As I mentioned on the last blog post, a challenge is that each pixel represents an area of 250m-by-250m. In other words, if within a certain area of 250m by 250m, 80% of the area represents rifted area and 20% represents non-rifted area, then the pixel appears to represent rifted area. Thus, although I drew each width line going to the corner of a pixel (see below), it is entirely possible that the width line should stop at any other point within that pixel (see below).

Image 4

Rift Length Measurements

For the following examples, I will continue to show the image from November 6th, 2012. By measuring the “rift length”, I drew lines approximately from the midpoint of the width to the end of the rift, which I again added on using the draw toolbar and line feature.

Image 5

I used the same approach as for the rift width measurements to obtain the line’s length, start point, and end point in meters and decimal degrees, respectively.

Image 6

As can be seen, however, the rift does not propagate in a straight line. While much of the beginning of it does (at least when projected on a 2D surface), it then becomes much more sinuous. A more accurate measurement could be obtained by tracing out the entire rift with many straight lines, as is shown below in red.

Image 7

Similar to the rift width measurements, the length measurements have the same issues with pixel sizes, and I am unable to obtain angle measurements due to the curved nature of the Earth.

In the following blog post, I will be showing and describing some of the trends and patterns I have observed from these images.