Forecasting Rainfall Anomalies in Arid Landscapes: A Machine Learning Approach for Enhanced Environmental Management in Lubbock, Texas

Abstract

Climate change is exerting profound effects on precipitation patterns, particularly in arid regions, leading to prolonged droughts and water scarcity. This research delves into the climatic complexities of Lubbock, Texas, a semi-arid locale susceptible to climate change impacts. Employing machine learning techniques, the study aims to predict precipitation anomalies, addressing the limitations inherent in traditional statistical analysis by machine learning predictions. Various models, including linear and ridge regression, autoregression, ARIMA, SARIMA, Multilayer Perceptron (MLP), and Long Short-Term Memory (LSTM), are assessed for their efficacy in precipitation prediction. Results showcase the promising performance of MLP, underscoring the importance of integrating variables such as temperature and snowfall into predictive models. However, variables can predict more accurately with larger training data. Moreover, the research explores the influence of additional factors, like thunderstorms and snowfall, on precipitation. The findings not only highlight the potential of machine learning for refining weather predictions but also offer valuable insights for regions grappling with environmental challenges. The study contributes to the understanding of regional precipitation variations, emphasizing the significance of comprehensive strategies for weather data processing in Lubbock, Texas. The research outcome aids in developing adaptive environmental management strategies and mitigating environmental inequality, addressing the evolving impacts of climate change over time.