CUSF Standalone Predictor - Version 2

Go to file

Daniel Richman be69f3d439 Change prune time to 14 days		2012-07-12 17:05:27 +00:00
cron	Change prune time to 14 days	2012-07-12 17:05:27 +00:00
gfs	New directory layout and predict.py script	2010-05-29 12:42:02 +01:00
pred_src	Also use alarm to clean up hung pred processes	2011-12-03 11:46:02 +00:00
predict	Add python logging, catch unhandled exceptions, configurable virtualenv	2012-07-12 17:04:15 +00:00
.gitignore	New directory layout and predict.py script	2010-05-29 12:42:02 +01:00
AUTHORS	Added AUTHORS, CREDITS and LICENSE	2011-01-13 15:30:16 +00:00
CREDITS	Added AUTHORS, CREDITS and LICENSE	2011-01-13 15:30:16 +00:00
LICENSE	Added AUTHORS, CREDITS and LICENSE	2011-01-13 15:30:16 +00:00
README.md	updated readme information	2011-01-10 22:32:09 +00:00
predict.py	Add python logging, catch unhandled exceptions, configurable virtualenv	2012-07-12 17:04:15 +00:00

README.md

CUSF Standalone Predictor - Version 2

Cambridge University Spaceflight landing predictor - a web-based tool for predicting the flight path and landing location of latex meteorological sounding balloons.

Install

The source for the predictor itself is in pred_src/ and instructions for building it can be found there.

The following items need to be executable (chmod +x ./predict.py) by the user under which the predictor runs:

predict.py
pred_src/pred (once compiled)
cron/clear-pydap-cache-cronjob.sh
cron/purge-predictions-cronjob.sh

The predict/preds/ and gfs/ directories need to have rwx access by the PHP interpreter and the predict.py python script. You will need to install the following python packages: pydap, numpy, json, simple-json. We use at to automatically background the predictor, so you will need that installed.

Other than that, just clone this repo to a non web-accessible folder and create symlinks to the predict/ directory in the repo.

There are useful configuration options in predict/includes/config.inc.php.

Information

The two shell scripts in the cron/ directory should both be run daily. clear-pydap-cache-cronjob.sh clears the cache used by pydap so that old data does not build up. purge-predictions-cronjob.sh deletes scenarios and predictions not accessed or modified within the last 7 days. Re-running a prediction for a scenario will therefore reset its time to live to 7 more days.

The directory names are UUIDs comprised of an SHA1 hash of the launch parameters, and re-running predictions will overwrite data in the existing directory, rather than create a new one.

We use GFS data provided by the NOAA, accessed via NDAP and their NOMADS distribution system. The 1.0x1.0 degree data (26 vertical pressure levels) is used for standard predictions, and the 0.5x0.5 degree data (47 vertical pressure levels) is used for the high definition (HD) predictions.

License

This work is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or any later version. This work is distributed in the hope that it will be useful, but without any warranty; without even the implied warranty of merchantability or fitness for a particular purpose.

Credits & Acknowledgments

Credit as detailed in individual files, but notably:

Rich Wareham - The new predictor and the hourly predictor system
Fergus Noble, Ed Moore and many others

Adam Greig - http://www.randomskk.net - random@randomskk.net
Jon Sowman - http://www.hexoc.com - jon@hexoc.com