tomasrojasc / correlation-module-nowcast Goto Github PK

This module is part of the NOWCAST proyect by ESO that aims to do short-term-forecast of the seeing parameter in order to improve astronomical observations.

This module makes the discrete cross correlation of two sites' seeing measurements using the module pydcf by Damien Robertson.

In order to execute this module, you have to run the file main.py.

The first thing this module does is taking the two files in the folder files2cross_corr. This files have to be named file1.csv and file2.csv and they have to have the structure specified in files2cross_corr/README.md. Once those files are in place, it separates the time series in UT time intervals nights, for this it considers a day being from 22ºº hrs to 11ºº of the next day, and the label of the day is given by the first day.

Example: Suppose you have the data of 2009-06-03 at 12ºº hrs to 200-06-04 at 14ºº hrs, then a UT day would be from 2009-06-03 22ºº to 200-06-04 11ºº and the date_keyassociated with this day will be 2009-06-03.

Once That is done, it saves each day for each site in a different folder named UT_data/file1 and UT_data/file2 respectively. Once it has those files, it passes the common dates for the two sites to Damien Robertson's correlation module, and it saves all the outputs to the output folder.

Then it takes all those outputs that are in csv form and merges them into a pandas dataframe that is saved in the folder final_output as a binary file.

Finally it makes another binary file containing the maximum correlations and the corresponding shifts, also as a binary file un the final_output folder.

Basically the only thing you have to do is set up file1.csv and file2.csv and run the python file main.py

All the intermediate files are deleted after the execution is done. This is to avoid confusions. This can be ignored by deleting the line in main.py that executes the clean.sh bash file.

• The pydcf bby Damien Robertson can be found here.