Detection of large-scale climate signals in spring vegetation index (normalized difference vegetation index) over the Northern Hemisphere



[1] Climate is one of the determinants driving ecosystems on both local and global scales. During the last two decades, there has occurred a dramatic temperature increase in the northern midlatitudes to high latitudes. The rapid warming has resulted in the promotion of bioactivity and an early growing season. However, the temperature and vegetation changes are not uniform in geographical distribution. In the present study, we analyze the spatial features in the normalized difference vegetation index (NDVI)-temperature relationship over Eurasia and North America in spring for the period 1982–2000. The NDVI data are derived from the Earth Observing System Pathfinder advanced very high resolution radiometer data sets. A singular value decomposition analysis is applied to the covariance matrix of the NDVI and temperature. Most of the squared covariance, 91.6%, is captured by the first seven paired modes. The result clearly indicates that the temperature is a focal factor influencing vegetation activity. Furthermore, those seven paired modes show large-scale features and well-defined patterns. The atmospheric circulation systems, such as the Southern Oscillation, North Atlantic Oscillation/Arctic Oscillation, Pacific/North American pattern, Eurasian pattern, western Pacific pattern, western Atlantic pattern, eastern Atlantic pattern, and North Pacific index, are found to be associated with that. The time coefficient corresponding to the first paired modes, centered on western Siberia, is correlated significantly with the Eurasian teleconnection pattern. Their correlation coefficients are 0.72 and 0.78 for vegetation and temperature, respectively, for the data period. Other modes are also correlated with one or more circulation indices. This implies that the large-scale circulation is essential for understanding the regional response of vegetation to global climate change. Taking all nine circulation indices and time lags into account, a large portion (71%) of the satellite-sensed variance in NDVI could be explained. The temperature-NDVI relationships did not change significantly when the NDVI was rescaled from 1 to 5 degrees, indicating that the singular value decomposition analysis is an appropriate technique for detecting the degree of coupling between vegetation and climate and that the vegetation-temperature relationship presented in this study is robust.