S68
Linking climate to landscape: Investigating the response of vegetation to precipitation variability in a semi-arid catchment area of southern Africa

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Sunday, 17 January 2010
Exhibit Hall B2 (GWCC)
Forrest R. Stevens, University of Florida, Gainesville, FL; and A. E. Gaughan and C. Gibbes

Understanding how inter- and intra-annual precipitation affects seasonal vegetation dynamics is critical for assessing potential impacts of climate variability on vegetation structure and composition. This is especially true in semi-arid and arid ecosystems where water is a limiting resource and where varied species respond to timing, frequency, and intensity of rainfall in different ways. This study presents an analysis that correlates the response of photosynthetic activity with inter- and intra-annual precipitation variability in a regional catchment of southern Africa at a monthly time-step over the years 2000-2009. We estimate monthly precipitation using the Tropical Rainfall Monitoring Mission 3B43 dataset and monthly vegetation responses using the MODIS 13A1 normalized difference vegetation index (NDVI) product as a proxy for vegetation productivity. We present a time-series model that accounts for monthly lag and seasonal effects that estimates NDVI “greening” response to precipitation. Initial analyses show that the highest correlation between NDVI and precipitation is measured at a one month lag. However, the coupling between precipitation and vegetation varies by vegetation type. We present a comparison of model results across vegetation types in the catchment area. In addition to identifying the months of prior rainfall that are important to NDVI greening, a Geographic Weighted Regression (GWR) is used to predict NDVI after the peak of the growing season. GWR provides a useful tool to elucidate the rainfall-vegetation relationship by minimizing the effects of unmeasured, spatially-varying factors. Our research contributes to climate-land studies by examining how vegetation types respond to precipitation variation at multiple spatial and temporal scales and is a vital component to understanding shifting dynamics of dryland ecosystems.