Selecting the optimal gridded climate dataset for Nigeria using advanced time series similarity algorithms

Area of Science:

• Hydrology and Climate Science
• Geospatial Data Analysis
• Environmental Monitoring

Background:

• Gridded climate datasets are crucial for hydro-climatic research, especially in data-scarce regions.
• Selecting the most reliable dataset is challenging due to varying data quality and spatial resolutions.
• Nigeria's hydro-climatic research is hindered by a lack of long-term, dense observational data.

Purpose of the Study:

• To evaluate, compare, and rank five prominent gridded climate datasets (CRU, TERRA, CPC, ERA5, CHELSA) in Nigeria.
• To assess dataset performance in replicating in situ rainfall and temperature data.
• To introduce and validate advanced time series similarity algorithms (STS, CCD) for gridded data evaluation.

Main Methods:

• Applied Perkins skill score (PSS), short time series distance (STS), and cross-correlation distance (CCD) to compare datasets.
• Utilized compromise programming (CP) with four statistical criteria for performance evaluation.
• Validated performance against in situ rainfall, maximum, and minimum temperature data at 26 Nigerian locations.

Main Results:

• The Climate Research Unit (CRU) dataset was ranked as the best for Nigeria, followed by CHELSA, TERRA, ERA5, and CPC.
• STS and CCD methods confirmed CRU's superior performance, showing integrated values of 17 (rainfall), 10.1 (max temp), and 20.8 (min temp).
• CRU demonstrated excellent statistical performance with Pbias (0.5-1), KGE (0.5-0.9), NSE (0.3-0.8), and NRMSE (-30 to 68.2).

Conclusions:

• STS and CCD are effective and efficient alternatives for evaluating gridded climate data performance.
• CRU is the recommended gridded climate dataset for hydro-climatic research in Nigeria.
• The study highlights the utility of advanced algorithms in overcoming limitations of traditional multi-criteria decision methods.