Interannual memory for spring NDVI in semi-arid South Africa

• Published in: Geophysical research letters 2008 (35)
• Main Authors: Richard, Yves, Martiny, Nadège, Fauchereau, Nicolas, Reason, C., Rouault, Mathieu, Vigaud, Nicolas, Tracol, Yann
• Format: Article
• Language: English
• Subjects: Climatology; domain_sde.mcg.cg; Earth Sciences; Environmental Sciences; Global Changes; Sciences of the Universe
• ISSN: 0094-8276; 1944-8007
• DOI: 10.1029/2008GL034119

Almost 20 years of Normalized Difference Vegetative Index (NDVI) and precipitation (PPT) data are analysed to better understand the interannual memory effects on vegetation dynamics observed at regional scales in Southern Africa (SA). The study focuses on a semi-arid region (25°S–31°S; 21°E–26°E) during the austral early summer (September–December). The memory effects are examined using simple statistical approaches (linear correlations and regressions) which require the definition of an early summer vegetation predictand (December NDVI minus September NDVI) and a consistent set of potential predictors (rainfall amount, number of rainy days, rainfall intensity, NDVI and Rain-Use-Efficiency) considered with 4 to 15-month time-lag. An analysis over six SA sub-regions, corresponding to the six major land-cover types of the area reveals two distinct memory effects. A “negative” memory effect (with both rainfall and vegetation) is detected at 7 to 10-month time-lag while a “positive” memory effect (with vegetation only) is observed at 12 to 14-month time-lag. These results suggest that interannual memory effects in early summer vegetation dynamics of semi-arid South Africa may preferably be driven by biological rather than hydrological mechanisms.