Foliar phenology of vegetation in south-central Africa and its relevance to climatic change.

摘要

The dynamics of leaf display (foliar phenology) in south-central Africa are examined and their relationship with climatic variation from 1981-1990. Regional-scale (7.6-km resolution) foliar phenology was observed using time series of the Normalized Difference Vegetation Index (NDVI) derived from multitemporal Advanced Very High Resolution Radiometer (AVHRR) data measured from NOAA Earth orbiting satellites. Analysis of these data revealed positive and significant correlations, with lags, between monthly rainfall totals and leaf display at sites with average annual rainfall totals less than 600 mm. At more mesic sites, the relationship between monthly NDVI and rainfall was not significant. NDVI images also revealed extensive increases in leaf display before the accumulation of significant rainfall each spring. Several explanations of early greening are offered, including the possibility that this phenomenon is unrelated to the current climate. Whatever explanation is accepted, it is clear that this phenological type is particularly vulnerable to delayed spring rainfall, associated with global climate change.; The regional-scale phenological patterns were consistent with higher-resolution observations of savanna-woodland phenology in eastern Zambia. Spatial and temporal patterns shown in regional-scale observations were also observed in 1-km AVHRR and Landsat MSS observations. The 1-km observations revealed a fine-scale pattern of coalescence of insular patches of high NDVI during the spring transition. A model of canopy radiative transfer was developed to describe the regional-scale reflectance in terms of canopy optics and tree cover, measured with hand-held radiometers and cameras. The model explained a significant amount of variation in NDVI for the four vegetation types studied (Colophospermum mopane woodland, plateau and scrub miombo woodland, and Terminalia woodland). The model was used to estimate the variable influence of the two canopy layers (field and tree) on AVHRR observations. Model simulations suggested that the tree layer accounted for a relatively modest proportion ({dollar}<{dollar}40%) of reflectance during the wet and dry seasons. The model's predictive utility and simplicity suggest that it may be applied more broadly to obtain estimates of tree cover from AVHRR observations of savanna communities.