Portal de Periódicos da CAPES

Aspects of seasonal dynamics of flooding in the Okavango delta

2005; Volume: 37; Linguagem: Inglês

0525-5090

T. Masaka, Piotr Wolski, L. Raditsebe, M. Murry-Hudson,

Hydrology and Watershed Management Studies

Seasonal changes in inundation area and peak flood occurrence in the Okavango Delta are analysed using a multiple regression technique. The multiple regression models developed constitute useful and simple tools for predicting inundation area and peak flood occurrence in the Okavango Delta. The regression model for monthly inundation area achieves a coefficient of determination of 0.80 and standard error of 538 km2. Explanatory variables in the model are various expressions of long-term and short-term antecedent rainfall and inflow conditions. The model for flood peak occurrence can be used for accurate predictions only in the Jao-Boro distributary, for which it achieves a coefficient of determination of 0.85 and standard error of 15 days, with distance to Delta inlet and an expression of flood size as explanatory variables. Propagation of the flood in the two other analysed distributaries, Maunachira-Khwai and Mboroga-Santantadibe, is complex and its quantitative description appears to be beyond the capacity of a simple regression approach. Additionally, the analyses presented provide insight into the role of storage in the dynamics of flood in the system: hydrological inputs are accommodated in the large system storage, and hydrological response is strongly dependent on the factors affecting (slow!) release from that storage. Based on the analyses, the classic model of kinematic flood wave propagation has been adapted accordingly. Introduction The Okavango Delta (Figure 1) is a large wetland pulsed by the annual flood of the Okavango River. The permanent and seasonal flooding creates an ecosystem in stark contrast to the surrounding rain-fed savannah of the Kalahari. This makes the Okavango Delta the basis for subsistence livelihoods of the local population, and the main attraction of Botswana’s tourism industry. In 1997 the Okavango Delta was declared a Ramsar site – a wetland of international importance. An important hydrological feature in the Okavango Delta is the difference in timing of local rains and flooding. The flood expands several months after the end of the rainy season, during the dry cold period, which makes water practically available throughout the year. The timing of the flood in the system, i.e. when the flood arrives, achieves maximum extent and subsides, has thus profound consequences for the ecology of the system. Also, it directly affects farming (planting time) and tourism (accessibility). Predicting flood timing is thus an important issue. However, probably even more important is the understanding of the controls behind the flooding process and its natural variation. Such understanding could be used to elucidate, for example, causes of reduced flooding and late flood arrival. This becomes important in view of increasing development pressure on the Okavango River and the Delta proper, and the common perception of Delta drying due to upstream abstractions. Complex hydrological models are one of the solutions that can be applied to predict Botswana Notes & Records, Volume 37 179 ______ 1. Harry Oppenheimer Okavango Research Centre, Private Bag 285, Maun, Botswana. 2. Department of Environmental Sciences, University of Botswana, Private Bag 0022, Gaborone, Botswana. 3. Department of Water Affairs, Private Bag 0029, Gaborone, Botswana flood size and timing of flooding. However, simple methods, such as statistical regression-type relationships, are often as useful and helpful in understanding the phenomena. For example, regression models of outflows and flood size run for the annual data were reasonably successful and indicated that the system is relatively well describable using statistical procedures. This study aims at describing the relationship between the flooding, particularly the seasonal changes in the extent of the flood and timing of peak flood occurrence, and the influencing environmental variables such as inflow, precipitation and evaporation. Previous Work A qualitative understanding of processes influencing the hydrology of the Delta has been obtained in the course of past research and is presented in many publications. McCarthy et al (1991) described processes of channel–floodplain water exchange, revealing the importance of seepage from the channels through the vegetated banks. This process leads to progressive loss of water from channels. A portion of the water seeping to floodplains evaporates, another portion contributes to the spread of inundated area, while some may be intercepted by downstream systems of channels (Porter and Muzila, 1988). A channel aggradation cycle was described by McCarthy et al (1992). In this cycle, the within-channel accumulation of bedload Botswana Notes & Records, Volume 37 180 Fig. 1. Location of the Okavango Delta and its principal features. leads to a reduction of channel flow velocities and an increase of channel losses to surrounding swamps. Subsequently, vegetation blockages develop and finally the channel is abandoned. Groundwater behaviour under islands and its role in entrapment of dissolved salts was described by McCarthy et al (1993) and McCarthy and Ellery (1995). Modelling work of the floodplain-island groundwater flows was done by Gieske (1996), and recently by Wolski and Savenije (2004). Water balances at the scale of a single floodplain were studied by Dincer et al (1976) at Beacon Island, and by Ramberg et al (2005) at Phelo’s floodplain on the SW side of Chief’s Island. The conceptual understanding of the Delta hydrology obtained from these studies was used to develop and subsequently improve several mathematical models of the system. The most notable efforts are those by SMEC (1990), described also by Dincer et al (1987), Scudder et al (1993), WTC (1997) and Gieske (1997). Recently, a distributed MODFLOW-based model was developed by Bauer (2004) and a hybrid reservoir-GIS modelling approach was applied by Wolski et al (2005b). These latter two models are able to simulate the observed outflow and inundation area relatively well, but are rather complex and computationally intensive. Apart from the complex models, simple statistical procedures have also been applied in the past to simulate and analyse the observed response of the Delta to the hydrological inputs. A regression model of the Delta outflows at Maun was described in various versions by Dincer et al (1987), SMEC (1990), Scudder et al (1993) and McCarthy et al (1998). In that model the variation in total annual rainfall, total annual inflow, evaporation and antecedent wetness (previous year outflow) were shown to explain 93% of the observed variability of total annual outflow at Maun. A regression model of maximum annual inundation area was developed by Gumbricht et al (2004). In this model, the maximum annual inundation area derived from satellite images, was related to inflow, outflow, rainfall, potential evapotranspiration and antecedent wetness conditions. The last factor was expressed by maximum area of flooding in the preceding year. The multiple regression model had a coefficient of determination (r2) of 0.87 with a standard error of 695 km2 and the potential evapotranspiration variable was shown to be not significant. Additionally, the model was able to predict the maximum area of flooding three months in advance. The predictions are published on the Harry Oppenheimer Okavango Research Centre web site (www.orc.ub.bw). The Okavango Delta The Okavango Delta in northern Botswana (Figure 1) is an alluvial fan covering about 22 000 km2. The Delta is fed from the central Angolan highlands through the Okavango River. Inflow into the Delta at Mohembo peaks usually in April, whereas the outflow in the Thamalakane River as well as inundation extent peak usually in August-September. The average annual discharge of the Okavango River at Mohembo is 1.01 x 1010 m3 but is quite variable, ranging from a low 6.0 x 109 m3 to a high of 1.64 x 1010 m3 over the last 60 years. Base flow in the Okavango River sustains 4,000-6,000 km2 of permanent swamp around the apex of the alluvial fan, but at the peak flood, an additional 2,000-6,000 km2 of the so-called seasonal and occasional floodplains may be inundated. Temperatures are generally high during summer months (October-April) with average temperatures of 27°C and 25°C for Maun and Shakawe, respectively. The winter months (MaySeptember) are dry with temperatures averaging at 17°C in Maun and 15°C in Shakawe. The mean annual pan A evaporation (uncorrected) is 2730 mm for Maun and 2460 mm for Shakawe. Botswana Notes & Records, Volume 37