Correlation Between the Interannual Variability of the Vegetation Activity and the ENSO Cycle in the Tropics

Monitoring the interannual variation of the vegetation productivity in the tropics is necessary to understand the global changes in carbon cycle and climate. Although tropical ecosystems must have been affected by ENSO cycle, the relation is not clarified yet. In this study, we analyzed the correlation between the normalized difference vegetation index (NDVI) originated from the Advanced Very High Resolution Radiometers (AVHRR) on NOAA satellites and the ENSO signal by means of Singular Spectrum Analysis (SSA). Assuming that ENSO cycle consists of the tropospheric biennial oscillation (TBO) mode (2-3 year) and the lower frequency variability (LFV) mode (3-7 year), we have extracted corresponding frequency components from time-series NDVI data from 1982 to 2000, and analyzed the relation to climate variability using the observed climate records. A certain portion of NDVI variation (around 20-30%) could be explained by the component with 2-7 years' periodicity. Seasonal analysis of NDVI and the Southern Oscillation Index (SOI) shows clear relationship between them in the most places over tropics (e.g. Amazon, Southeast Asia, and Central Africa). In addition, the strongest correlation with the SOI was found in DJF. It is suggested that the wet season (DJF) affects the vegetation growth. Correlation analyses of NDVI and climate variables (temperature, precipitation, incoming solar radiation, vapor pressure and cloud cover) were also carried out to identify the driving factor of ENSO-related NDVI variations. Precipitation and cloud cover variations shows good coincidence with NDVI variations. Since water is abundant over tropical rain forests, the cloud cover variation, which is related to incoming solar radiation, can explain the NDVI variation over tropics. Although the relationship between NDVI and incoming surface solar radiation was not detected well, one of the possibilities is due to the quality of radiation data. Our analysis indicates that tropical ecosystems are responding to climate changes due to ENSO. Further analysis will be focused on the quality of satellite data and comparison with terrestrial ecosystem modeling approaches.