Comprehensive review of detrending methods for crop yields: Approaches, applications, and future directions

• Provides a comprehensive review tracing the 60-year evolution of detrending methods, from classical statistical to contemporary machine learning techniques. • Systematically compares the theoretical foundations, inherent trade-offs, and practical performance of diverse approaches, offering a clear guide for method selection. • Demonstrates with concrete examples how the choice of detrending method directly impacts critical conclusions in climate risk assessment, agricultural policy, and food security planning. • Identifies key knowledge gaps and emerging challenges, such as the need for adaptive methods to handle technological shocks and non-linear trends. • Proposes a strategic roadmap and actionable guidelines for future research, emphasizing the integration of hybrid models and domain-specific knowledge. Against the backdrop of the severe challenges posed by global climate change and food insecurity, accurate yield forecasting is critically important for agricultural risk management, policymaking, and resource allocation. Crop yields result from the combined effects of meteorological conditions and advances in agricultural technology. A key scientific challenge in yield forecasting is accurately distinguishing short-term yield fluctuations caused by weather variability from long-term trends driven by technological progress. This separation is essential for producing reliable datasets that support the analysis of yield variability and predictive modeling. For decades, the detrending of crop yields has been a central focus in fields such as agricultural meteorology and crop science. This paper systematically reviews the development of detrending techniques in yield forecasting over the past six decades, from traditional linear and polynomial methods to advanced machine learning algorithms and multi-model integration approaches in recent years. It thoroughly examines the theoretical foundations, advantages and limitations, application scenarios, performance variations, and empirical outcomes of different detrending methods. The analysis reveals that the choice of method can significantly influence research outcomes, with important implications for climate change impact assessments, agricultural policymaking, crop yield forecasting, and food security planning. Furthermore, the paper highlights current research hotspots and challenges while outlining future directions and development trends in the field. This paper offers a systematic perspective on understanding the evolving trends in crop yields and proposes that future research should focus on adaptive and dynamic detrending algorithms, uncertainty quantification, integration of external variables, standardization of methods, and the use of big data resources. This comprehensive assessment provides both methodological guidance for researchers and a strategic roadmap for advancing the study of detrending techniques in agricultural yield analysis.