Determining the Effective Meteorological Parameters on Potato Yield in Niğde Province and Yield Prediction with Machine Learning and Deep Learning Models

Abstract

In this study, we explore the impact of meteorological parameters on potato yield in Nigde province, a key agricultural region in Turkey for potato production. Understanding the relationship between weather conditions and crop yield is crucial for optimizing agricultural practices and ensuring food security, especially in regions susceptible to climate variability. The study includes all meteorological parameters observed and recorded in Nigde Directorate of Meteorology, covering the period between 1990 and 2023. Through the analysis of historical weather data and potato yield records, we aim to identify the most influential meteorological factors affecting potato production. Then, artificial intelligence techniques such as Random Forest, Gradient Boost, Convolutional Neural Networks and Recurrent Neural Networks are utilized to predict the potato yield in order to provide valuable insights into how different weather patterns influence crop performance. The findings suggest that potato yield is correlated with meteorological parameters to some degree and AI techniques can predict the potato yield but lacks the precision. The reason is that potato production in Nigde is mostly done with irrigation. However, this further increases the risk that could result from the changes in groundwater levels and pollution in irrigation ponds for agricultural practices in Nigde.