Quantifying the Influence of Remote Climate Indices on Key Climate Variables in Northern Ghana: A Comprehensive Multivariate Approach

This research investigates the influence of remote climate indices, such as ENSO, NAO, IOD, and others on rainfall and temperature variability in northern Ghana, where climate fluctuations impact agriculture, socio-economic and livelihood activities. Using data from 1960–2016, multivariate regression models were applied to analyse correlations between these indices and local weather patterns. A hybrid outlier detection method–combining the Interquartile Range, Bayesian Change Point analysis, and Hampel Identifier–was used to manage data anomalies. Multicollinearity was addressed through iterative variance inflation factor (VIF) reduction, yielding final VIF values below 5, thereby enhancing model stability. The results show that the AMO exhibits strong positive correlations with maximum and minimum temperatures at key locations, including Tamale (tmax: 0.533, tmin: 0.677), Wa (tmax: 0.637, tmin: 0.702), and Yendi (tmax: 0.701, tmin: 0.732). The OLS regression models highlight the significant role of indices such as NAO, SOI, PDO, and TSA in influencing rainfall, although the explanatory power, indicated by R2 values, varies by location. Temperature models, in contrast, showed stronger and more consistent relationships, particularly with AMO, TSA, and PNA. These findings underscore the importance of integrating remote climate indices with local knowledge to enhance climate prediction and adaptation strategies in northern Ghana. This study contributes to a deeper understanding of teleconnections in the area and offers a framework for future research and policy interventions that aim to improve the resilience and adaptive capacity of vulnerable communities.