Does any phenological event defined by remote sensing deserve particular attention? An examination of spring phenology of winter wheat in Northern China

Phenology is often considered the “leading indicator” of ecological responses to climate change, and therefore it is important that researchers have accurate methods to track phenological changes. Remote sensing has been widely used to study phenological responses to climate change. However, land surface phenology observed by remote sensing is fundamentally different from that observed in the field, which raises the difficulty in understanding and validating phenological change observed using remote sensing. In this study, we revisited the criteria of “good” phenological events and argued that the relationship between phenology and climate factors is one of the most important meanings of phenological studies. Instead of validating remotely sensed phenology by its consistency with field observations, this study aims to judge different possible definitions of phenological events based on remote sensing by their temperature sensitivity and correlation. Using the winter wheat zone in northern China as the study area, we compared the temperature correlation and sensitivities of winter wheat phenology date derived from different methods: the relative threshold method with different thresholds, and the curvature method, based on remotely sensed data. Our results show that there is no distinct phenological event that is overwhelmingly more sensitive or correlative than any others. Therefore, there are no particular phenological events that deserve emphasis when exploring the relationship between phenology date and the pre-season temperature. Instead, the phenological stage (i.e. the threshold of relative threshold method) that is most sensitive or correlative to pre-season temperature varies spatially, showing a good latitude gradient. On an average, the thresholds of the most correlative and sensitive phenological stages to pre-season temperature decreased by 9.92% and 14.69% per latitudinal degree, respectively. The results indicate that the traditional emphasis on discrete phenological events could miss the phenological stages that are most sensitive and correlative to pre-season temperature, thereby resulting in a limited understanding of phenological responses to climate change.