15 July 2016 Genetically optimizing weather predictions
Author Affiliations +
Abstract
humidity, air pressure, wind speed and wind direction) into a database. Built upon this database, we have developed a remarkably simple approach to derive a functional weather predictor. The aim is provide up to the minute local weather predictions in order to e.g. prepare dome environment conditions ready for night time operations or plan, prioritize and update weather dependent observing queues.

In order to predict the weather for the next 24 hours, we take the current live weather readings and search the entire archive for similar conditions. Predictions are made against an averaged, subsequent 24 hours of the closest matches for the current readings. We use an Evolutionary Algorithm to optimize our formula through weighted parameters.

The accuracy of the predictor is routinely tested and tuned against the full, updated archive to account for seasonal trends and total, climate shifts. The live (updated every 5 minutes) SALT weather predictor can be viewed here: http://www.saao.ac.za/~sbp/suthweather_predict.html
© (2016) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
S. B. Potter, S. B. Potter, Kai Staats, Kai Staats, Encarni Romero-Colmenero, Encarni Romero-Colmenero, } "Genetically optimizing weather predictions", Proc. SPIE 9910, Observatory Operations: Strategies, Processes, and Systems VI, 99102H (15 July 2016); doi: 10.1117/12.2232306; https://doi.org/10.1117/12.2232306
PROCEEDINGS
11 PAGES


SHARE
RELATED CONTENT


Back to Top