Replace/supplement fetchdata script with swathe-level processing tool

[danielmwatkins]

The current version of IFT-pipeline accesses the near-real-time Earthdata Snapshots. This has the advantage of being simple for data access, but has the downside that the merged images add ambiguity to the time stamps (i.e., if an image has sections from two images in it, which image does the time refer to?). The ideal method would be to access the swathe level data directly. This data is available at specific bandwidths, and differing resolution, so an intermediate step is to merge the desired bandwidths into truecolor and falsecolor images.

Laura Crews at University of Washington has demonstrated that a combination of two Python libraries allows a straightforward and extendible framework to access NASA and NOAA satellite imagery at the swathe level. She was able to construct close to hourly resolution true color and false color images from VIIRS and run portions of the IceFloeTracker code on it. Her work provides a path to achieving 3 major goals:

• Extend Ice Floe Tracker to other optical satellite products
• Decrease the frequency of image artifacts interfering with floe identification
• Increase the time resolution to enable observation of a broader range of ice dynamics

The key Python libraries are the following:

• NSIDC's earthaccess library link
• PyTroll's satpy library link

We'll have to think about the choices of workflow regarding scene choice, subsetting, and so on. Open to ideas on where this would best fit!