Calibrating a distributed model to estimate runoff for ungauged catchments in Zimbabwe

Available modeling approaches requiring a high spatial intensity of ground-based measurements, a fine time-scale for monitoring and continuity of observed time-series have limited applicability for water resources planning and management in Zimbabwe. A review of existing runoff gauges on this research showed that data from more than 30% of them was unreliable, the number of rainfall, runoff and evaporation observation stations with usable data was decreasing and so were the available funds to support monitoring and maintenance of the stations. On the other hand freely available hydrological tools reviewed require large data sets from such measurement systems. This research investigated the feasibility of estimating runoff using a distributed parameter value estimation approach for various catchment sizes. Storm durations of the order of magnitude of hours and minutes were simulated to accurately model runoff generation processes experienced. Runoff of different segments is calculated simultaneously but independently. Each segment represents an area of different rainfall, soil condition, vegetation cover and land use, influencing runoff. The model has been calibrated and validated on cumulative and incremental catchments. Basins have been defined to generate stream flow for larger watersheds. Modules (576) were fitted into 12 basins each module having a period of runoff estimation of up to 90 years. This methodology can be applied to estimate runoff for the whole country more frequently. Both estimates used empirical and manual techniques to determine input parameter values. The model enables more frequent runoff updates using techniques, which better represent the hydrological characteristics of a catchment. Hydrologic problems can be resolved by using runoff time-series thus avoiding many simplifying assumptions made in alternative methods. However, digital data may be expensive and may take considerable time to procure and configure.