Synthetic rainfall forecasting model for the Kingdom of Saudi Arabia
This project is a personal and independent effort to re-create and improve upon a proprietary Fortran model. The goal is to gain experience with Python programming and its associated data analysis libraries (numpy, pandas, scipy, matplotlib, etc).
The synthetic rainfall forecaster uses 30-years of historical rainfall data to model significant characteristics:
-
Rainfall intensity: characterized by depth of individual storm events, sampled using a bootstrap method, drawback is that it fails to capture extreme events not existent in the historical data.
-
Seasonality: characterized by frequency of rain events per month, follows a poisson distribution.
-
Clustering of storm events: characterized by a series of conditional probabilities.