The start of the growing season (SOS) is essential to track the responses of vegetation to climate change. However, recent findings on whether the SOS in the middle-high latitudes of the Northern Hemisphere (NH) continued to advance or reversed during the global warming hiatus were not consistent. It is necessary to investigate the causes of this controversy and to examine the relationship between the SOS and preseason temperature trends. To this end, we first applied four widely used phenology extraction methods to derive the SOS from the GIMMS NDVI3g dataset and then used the ensemble empirical modal decomposition (EEMD) method to extract the nonlinear trends of the SOS and preseason temperature. Our results clarify, for the first time, that the limitations of the linear assumption-based trend analysis methods are an important but overlooked cause of the discrepancies among existing studies on whether the SOS was advanced or delayed in the NH (>30° N) during the global warming hiatus. We further revealed the range of the mismatches between the SOS and preseason temperature trends at the latitude, altitude and biome levels. Specifically, we discovered that the SOS in the NH (>30° N) obtained by the four phenology extraction methods showed a significant reversal from advance to delay during the global warming hiatus, and the corresponding average rate of change was very small. The area showing increasing preseason temperatures decreased during the global warming hiatus, but it always occupied most of the NH (>30° N). However, delayed SOS trends were dominant in the NH from 50° N to 60° N, above 3000 m and in biomes other than TBMF and BF. Accordingly, using an EEMD-like approach to evaluate the changes in the SOS and preseason temperature is necessary for improving our understanding of the changes in the SOS and their association with climate.
Keywords: Northern Hemisphere; ensemble empirical mode decomposition method; spring phenology; trend change.
© 2023 John Wiley & Sons Ltd.