Exploring the optimal climate conditions for a maximum maize production in Ghana: Implications for food security

Abstract

In Sub-Saharan African (SSA) countries like Ghana, where a significant portion of the population relies on agriculture for their livelihoods and sustenance, climate variability poses a substantial threat to crop productivity and food security. Therefore, it is crucial to employ advanced methodologies to study the intricate relationship between climate change and crop yield. This study therefore aims to assess the impact of different climatic variables on the variation of maize yield in Ghana from 1992 to 2018 and the pivotal role of machine learning techniques in predicting the variations in maize yield, considering the complex climate-crop yield interactions. The machine learning techniques utilized include the Random Forest (RF) Model, the Extreme Gradient Boosting (XGBoost) model, and the Artificial Neural Network (ANN) model for prediction. The results demonstrate that rising temperatures and precipitation have a positive impact on Ghana's maize yield. Additionally, the study identified a critical range of climatic conditions that maximized maize production during the study period. Specifically, an average temperature between 27.9∘C and 28.1∘C, coupled with a precipitation range of 1290 mm to 1390 mm, corresponds to the optimal conditions for achieving maize yields exceeding 2.0 MT/ha. Among the machine learning models utilized for the prediction, the ANN emerged as the optimal model with an approximate mean squared error of 1%. Ultimately, our results provide a comprehensive and actionable framework for stakeholders in the agricultural sector, equipping them with the knowledge and tools needed to adapt to climate change and optimize maize production in Ghana.

Keywords

• Climate change
• Cereal production
• Machine learning techniques
• Artificial neural network
• Sustainable development goals