An evaluation of lake and river levels as indicators of climate variability in tropical Southern Africa

Abstract
Southern Africa has only a few.large lakes, one of which is Lake Malawi. It forms part of the lower Zambezi River catchment and is part of the Great Rift Valley. Lake Malawi plays a significant role in socio- economic development particularly in food production, health, energy, transport, recreation, and environment. Changes in its level are related to variations in rainfall and the surrounding atmospheric circulation. Runoff is greatest in late summer following an active-spell of the Inter Tropical Convergence Zone (ITCZ). This study looks at differences between years when the inflow to Lake Malawi is above and below normal using CEP reanalysis composites. An understanding offactors governing variations ofwater resources will be useful to hydrologists and resource managers who deal with seasonal and inter-annual fluctuations. To this end monthly lake level and rainfall data were analysed for variability in the period 1915 to 1995. From the lake level time series, differences in the minimum and maximum level from beginning to and of summer were calculated, then years with high inflow and low inflow were identified in the period overlapping with the NCEP reanalysis data 1958 - 1998. The Zambezi annual flows have been included to study regional coherence ofclimatic variability. Composites were generated during and prior to the season for sea surface temperature, outgoing long-wave radiation, wind, pressure, and circulation derivatives in the area 50S to 10N and 30W to 100E. The mean annual cycle oflake levels contains a peak in April-May and low point in November-December. The inflow index ( Maximum - Minimum) is well correlated with adjacent station rainfall (r =0.51 to 0.66). Spectral analysis on the inflow index indicates a cycle of 9.58 yrs in the period 1938-1976; and a lower frequency oscillation of 19.16 yrs in recent decades 1976-1995. Secondary oscillations occur at periods of 14.5 yrs, 8.3 yrs, 5.8 yrs (ENSO) and 2.05 yrs (QBO). The 9.58 yrs cycle is consistent with that of Abu-Zeid et al (1992) for the Nile River. Composite years include: Wet 1962,1963,1974,1978,1979,1989; and Dry 1959,1967,1972, 1977,1992,1994. Surface press.ure and wind show significant anomaly patterns for wet minus mean composites. The sea surface temperature patterns alternate over the Atlantic and Indian Oceans particularly in the 30S - 40S belt south east of Africa. Patterns in the precursor seasons: Sept - Nov and lun - Aug enable development of predictive models for Lake Malawi inflow at 3 to 6 months lead time.
Description
A thesis submitted in fulfilment of the degree of Master of Science in the Department of Geography at the University of Zululand, 1999.
Keywords
Lake and river levels
Citation