FREQUENCY ANALYSIS OF EXTREME RAINFALL IN THE CONTEXT OF NON-STATIONARITY

FREQUENCY ANALYSIS OF EXTREME RAINFALL IN THE CONTEXT OF NON-STATIONARITY

STEPS :

1. threshold selection
•  since generalized pareto distribution (GPD) is used in this frequency analysis, first we have to select the threshold which also known as peaks over threshold
2. parameters estimation
•  there are many ways to estimate the parameter. maximum likelihood, method of moments, probability weighted moments there are many research has been done to compare the estimation. in comparing those estimation, monte carlo is used. but maximum likelihood estimation is the best way to estimate the parameter in rainfall analysis.
3. model selection
•  by assuming the data is non-stationary, then have to detect trend in the data. sometimes the trend could not be seen by plotting. hence, we need to do some testing to detect the trend in data.
4. model checking
•  in making conclusion or decision whether the model is fit or not. using graphical: probability plot, quantile plot, return level plot and density plot.
5. return level
• the last step is estimate the return level. in this step, we have to ensure what unit (year, month or day). however, the unit is related with the unit of the data that we have. if the data is daily, then the unit must be in day.

will be update soon ...