LDMOJ4/extreme-rainfall-sahel

A repository with some of the scripts that were used to analyse rainfall in Niamey, Niger. This includes some exploratory data analysis, maximum likelihood estimation to fit extreme value distributions and gamma distributions, and building a Markov chain of rainfall existence to simulate 100's of years of rainfall. # Extreme Rainfall in Niamey A repository with some of the scripts that were used to analyse rainfall in Niamey, Niger for my dissertation. This includes some exploratory data analysis, maximum likelihood estimation to fit extreme value distributions and gamma distributions, and building a Markov chain of rainfall existence to simulate 100 years of rainfall. ## Project Abstract This paper analyses the effectiveness of convection permitting and parameterisation models in replicating the dynamics of Sahelian storms in Niamey. It then uses future runs of those models to try to predict how the climate will change in the future. This is done by comparing the intensity, duration, and meteorological dynamics of rainfall events in model runs with observational datasets. The focus of this paper is on extreme precipitation events and their frequency and likelihood. Rainfall data for 100-year periods is simulated using discrete time Markov chainsto allow more accurate analysis of the return periods of extreme rainfall. It is hypothesized that an increase in global sea temperature, an expected result of global warming, will increase the likelihood and intensity of extreme rainfall events. ## Context This project was completed in spring 2024 and includes all of the code to support my dissertation, please read the dissertation for full context. This was one of my first data science projects. The code is poorly annotated as it was not marked further and has lots of innefficiencies but it produced strong findings. The code can't run independently as the source data is not accessible. ## Data This analysis uses two climate models: The Met Office Unified Model Convection Permitting for Africa (CP4-A) and the R25-Africa Regional Model. The major difference between the two models is the resolution. The CP4 model as it is referred to in this paper has a resolution of 4.5km whereas the R25 model has a 25km resolution. The resolution determines how convection is modelled. The smaller r …