Mechanisms of inter-annual rainfall variability over tropical highlands of Africa and its predictability potential
Loading...
Date
2002
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Modes of climate variability over tropical Africa and adjacent Atlantic and Indian Oceans are investigated using continental monthly averages of gridded rainfall and temperature and marine environmental data, which consists of monthly averages of sea surface temperature, sea level pressure and zonal and meridional wind components. Rainfall and temperature data are gridded at a resolution of 2° x 2° latitude, longitude and based on conventional station data reports (University of East Anglia CRU data set) in the domain 10°N-35°S over Africa except Madagascar. The environmental data is at a resolution of 10° x 10° latitude, longitude and based on ship reports (COADS) ia the domain 30QN - 40°S, 70°W - 100°E. The data sets are subjected to Principal Component Analysis (PCA) using the correlation matrix technique, resulting in the gridded data being standardized. The data are analyzed for annual cycle and the departure therefrom (i.e. Inter-annual fluctuations). The time-varying spectral energy nature of time series is investigated using wavelet analysis technique.
PCA (with annual cycle) yielded 12 and 7 homogeneous rainfall and temperature regions, over the African continent south of 10°N, which explained a total of 81 and 91 variance respectively. PCA on marine environmental parameters over the Atlantic and Indian Oceans revealed a number of homogeneous regions, some of which could be associated with synoptic systems. Similarly, PCA was performed on the same data after removing the annual cycle.-The time evolution of these PC modes or time scores were shown to oscillate with periods of 2-4, 4-6, 8-12 and 14 -16 years. Long period oscillating modes were revealed over tropics, particularly the Atlantic Ocean.
Possible causes of rainfall and temperature variability over the tropical highlands of Africa were investigated by correlation analysis between rainfall and temperature, and environmental parameters and other indices at lag 0, using monthly data. Some modes of marine environmental parameters over the Atlantic and Indian Oceans have shown good association with the climate over the tropical highlands of Africa up to three months. Results also show that modes over tropical equatorial belt of the Atlantic and Indian Oceans explain rainfall variability up to about 25% for some areas over Africa. Temperature variability over the continent is associated with SST modes over the equatorial tropical oceans and some of these explain more than 64% of temperature variability. ENSO has also shown good association with temperatures over the continent while QBO is poorly associated with temperatures over the continent. Analysis results have shown further that rainfall variability over east and southern Africa regions is associated with the variability of SST, SLP and U wind component in the box 60°E~90oE,5°N-10oS, over the Indian ocean. Time delay analyses have shown that generally environmental parameter indices lead rainfall for between 2 to 20 months. Results indicate that there is no clear picture as to which parameter leads the other, as oscillations are not always in phase or exhibiting constant lag. The un-harmonious nature of these features could sometimes undermine predictability in
statistical modelling. So the idea of reviewing predictor-predictant relationships in a time varying manner is well justified.
Some parameters extracted from key areas show persistency and stability up to three months while the Nino3 index has shown persistency and stability of up to 6 months. Results have also revealed good associations between some environmental parameters over some key areas of the Atlantic and Indian Oceans and rainfall over some areas of tropical Africa, at 0 to 3 month lags. Analyses results have shown that the potential of forecasting rainfall over some areas of tropical highlands of Africa exists.
Description
Thesis submitted to the Faculty of Science for the Degree of Doctor of Philosophy, in the Department of Geography and Environmental Studies at the University of Zululand, 2002.
Keywords
Rain and rainfall--Africa, Tropical highlands