A systematic approach to modeling monthly maximum temperature and total rainfall in Kenya

Area of Science:

• Climate Science
• Statistical Modeling
• Environmental Science

Background:

• Goodness-of-fit (GOF) tests are crucial for selecting probability distributions in climate modeling.
• Combined GOF test approaches offer improved precision but are underutilized.
• Erratic climatic conditions necessitate robust statistical methods for accurate climate modeling and risk assessment.

Purpose of the Study:

• To evaluate probability distributions for climatic variables using a comprehensive, combined GOF test approach.
• To identify the most suitable distribution for temperature and rainfall data.
• To assess the robustness of the best-performing distribution through sensitivity analysis.

Main Methods:

• A scoring system was developed to rank distribution performance across multiple GOF tests.
• A composite score determined the best-fit distribution.
• Sensitivity analysis was performed on the selected distribution using varying data partitioning (block sizes).

Main Results:

• The Generalized Extreme Value (GEV) distribution consistently outperformed others for temperature and rainfall data.
• Extended block sizes captured long-term patterns but increased uncertainty; shorter block sizes led to overfitting.
• Intermediate block sizes offered a balance, yielding reliable parameter estimates and stable return levels.

Conclusions:

• The Generalized Extreme Value (GEV) distribution is robust for climate modeling.
• Appropriate selection of block sizes is critical for reliable climate parameter estimation and risk assessment.
• This study enhances methodologies for climate adaptation strategies, with implications for regions like Kenya.