This respository contains data and code used to construct a "clearsky dictionary". A full description of the algorithm and data can be found here. Files are sorted by type into code and data.

The code directory contains all code used in the project. Code is further broken up by matlab, python, and bash. See the python subdirectory for further description of the code contained there, as well as instructions for compiling the CSD. Scripts in bash are to be used in conjunction with some scripts in python, and so description of the bash scripts can also be found in python.

The data directory contains all data used in the project. Several data sources can be found here, including the raw TSIs (found in TSI_images) as well as pyranometer measurements provided by NREL's SRRL group (found in clearsky_index). The TSI_images directory contains not only the structured image files (contained in jpegs and ordered so that the path to each TSI indicates the date on which it was taken, as well as the camera which took it), but also a zip file containing all TSIs, an unstuctured directory containing the originally-named copies of the TSIs (cloud_images), and some python scripts which be used to unpack the TSIs from cloud_images (found in utils).

If you are accessing this code through GitHub then the image files will not be included, as they are too large to push through Git. Instead the utils directory is included, and a zip file of the images can be made available upon request. A description of how to do this can be found in the utils README file.