This is Part 2 of a series of notebooks showing how to plot GINI-formatted satellite data from a THREDDS server using MetPy and Siphon. In Part 1 we covered how to access and parse the data file. In this part, we cover:

• Grabbing the data from the file
• Making sense of the projection information
• Plotting with CartoPy and matplotlib

This is the first of what we hope will be a series of posts showing how to use Python for weather analysis and create graphics for a variety of purposes. In this two-part post, we demonstrate plotting a water vapor satellite image, specifically using GINI formatted data. GINI is the format currently used for satellite data transmitted across NOAAPORT, and is available on Unidata's demonstration THREDDS server. This first part focuses on accessing the data using Siphon and MetPy; the second part will introduce plotting using CartoPy.

By Joshua Clark

Editor's note: Josh Clark was a Unidata Summer Intern in 2015. He graduated from the University of Northern Colorado (UNC) in the spring of 2015 with a B.S. in Meteorology, and is headed to the graduate program at San Jose State's Fire Weather Research Lab starting in the Fall of 2015.

An example EdexPy interface (click to enlarge)

It is hard to believe my time here at Unidata has come and gone so quickly! Next week, I imagine it will be back to the “harsh” reality of being a student — sitting on a beach somewhere near Monterey or perhaps fly fishing the Sierras over the next twenty days awaiting the start of my first year of graduate school at San Jose State. What a terrible reality that will be!

This experience at Unidata and UCAR has been an incredible opportunity and I am privileged to have been afforded these ten weeks. When I started here, I envisioned an entirely different internship than what previous interns had completed. Rather than developing one particular project, I focused my time on gaining a greater understanding of software engineering as a whole and contributing to existing Unidata projects. I found a comfortable spot working with Unidata Python developers Ryan May and Sean Arms, and within one week I had learned a great deal about unit testing, code health, automated testing, and version control. Later, I would implement these principles in my first Python library, MesoPy.

A more realistic test is to ask for successive slices of some data, and see if multithreading speeds things up. So I made a test that makes data requests like send:

https://motherlode.ucar.edu:8081/thredds/dodsC/fmrc/NCEP/GFS/Global_0p5deg/files/GFS_Global_0p5deg_20090115_0000.grib2.dods?

Absolute_vorticity[0][0][0:1:360][0:1:719]
Absolute_vorticity[1][0][0:1:360][0:1:719]
...

Each request is 1 Mbyte. These are  grib2 files, which may be jpeg2000 encoded (im not sure in this particular file).

So each thread makes 20 requests of successive slices, and I try this with 1-14 simultaneous threads. The plot below is the speedup factor. Its pretty impressively linear, up to about 9 threads, then things get slower but then recover again. Not sure if that's meaningful - I need to run this 100 times and average.