Vegetation Dynamics Over the Northeast Region of Brazil and Their Connections With Climate Variability During the Last Two Decades of the Twentieth Century.

Abstract

The spatial and temporal responses of natural landscape ecosystems in the Northeast region of Brazil (NEB) to changes in rainfall conditions over the last two decades of the 20th century have not been fully examined. The NEB ecosystems are highly dynamic landscape responding to weather conditions, of which rainfall is the most important variable. The research described in this dissertation was conducted to test whether or not the impact of rainfall fluctuations on the vegetation dynamics can be spatially and temporally differentiated within the NEB ecosystems. This was achieved in two phases: In phase one, the spatial and temporal consistency of the response of monthly normalized difference vegetation index (NDVI) derived from Advanced Very High Resolution Radiometer (AVHRR) data records to monthly rain gauge data records for the period 1982-1993 was assessed. This assessment of the spatial and temporal responses of the NDVI time series data to rainfall time series data tested the hypotheses that responses of NEB ecosystems to rainfall conditions can occur with different time lags and can be spatially heterogeneous. This was achieved by using correlation coefficients with different time lags and by applying factor analysis through the use of Varimax rotation. In phase two, the spatial and temporal variability of the NDVI responses to NEB land surface conditions for the period 1982-2001 was quantified. This quantification of temporal and spatial variability in NDVI across the NEB ecosystems tested the hypothesis that responses of NDVI variability to land surface conditions of the NEB can be temporally and spatially heterogeneous by combining statistical parameters of monthly series of NDVI (minimum, mean, maximum, anomalies and coefficient of variation) and by using annual NDVI images. The results of phase one of this research showed that the direct response (i.e., no lagged response) of the NEB ecosystems to rainfall was associated with different land cover types. A strong positive relation was found in rainfall-limited NEB ecosystems that are associated with Caatinga biome. Particularly, the disturbance of the landscape reduced the significance of the relationship. Over the Cerrado biome and Atlantic rain forest (evergreen tropical forest), significant negative relations between NDVI and rainfall were found where there are no moisture availability constraints. However, the NDVI responses of the NEB ecosystems to changing rainfall conditions were found to be strongly positive for the correlation of NDVI with rainfall in the concurrent plus one previous month. This was demonstrated for most of the NEB ecosystems, except for the southeastern NEB that is formed by the Atlantic rain forest that, in certain areas, has been converted to cropland. In general, there is a good spatial agreement between the NEB ecosystems and the patterns of NDVI variability on both the annual basis and the interannual basis as the temporal responses of the NEB ecosystems are affected by both positive and negative correlations with rainfall. In phase two, the results show clear indications that, for the inter-annual NDVI variability in the interval between 20% and 45% variability a positive response of ND VI to changes in rainfall exist for most of NEB ecosystems where main annual rainfall amounts vary from 300 to 750 mm. A strong decrease of response was found in the monthly NDVI anomalies trend over the period of September 1988 to August 1997. The decrease might be interpreted as a response of NEB ecosystems caused by the impact of enhanced aridity over the 1990s. Overall, in the first phase, it was observed that responses of different land cover types of the NEB to rainfall conditions appeared to occur with different time lags and appeared to be spatially heterogeneous. In the second phase, the inter-annual NDVI variability in the NEB ecosystems to land surface variations appeared to be temporally and spatially heterogeneous. These results suggest that vegetation responses over the NEB ecosystems to changing rainfall conditions, as measured by changes in NDVI to rainfall fluctuations, may have a potential for predicting the impact of rainfall conditions on the NEB landscape dynamics.