Browsing by Author "Danda, Tererai"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    Mean annual runoff maping of the Rufiji basin
    (University of Dar es Salaam, 1999) Danda, Tererai
    In this report, the mean annual flow map for the Rufiji basin was created. In order to do this, maps for runoff coefficient and rainfall were first created for the basin. Runoff coefficient and mean rainfall were interpolated and extrapolated, using Krigging in SURFER and 0.5°x 0.5° grid based maps created for them. The two maps were then multiplied, to get a grid based map of mean annual runoff. In an attempt to use an alternative procedure to estimate flow for ungauged catchments, three conceptual models were tested on nine gauged catchments. The Soil Moisture Accounting and Routing Model (SMAR) gave poor results of R2 for both calibration and verification periods. The Xinanjiong model and NAM models, both gave reasonable values of R2, as high as 67, 57% for subcatchment 1ka39a with the Xinanjiong model and 63.43% for sub-catchment lkb8 with the NAM model. However, mapping as many as 15 parameters for the Xinanjiong model and combining the maps to estimate flow, proved to be time consuming, given the time for the research. Nevertheless, recommendations were made about these models. A further attempt to develop a procedure to estimate flow for ungauged catchments was done by applying the Linear Perturbation model and the Gamma Function model. These models were applied on 20 catchments with both rainfall and discharge series smoothed via FOUrfer series using three harmonics and the ratios for each Fourier coefficient (flow/rainfall) were calculated for each catchment. These ratios for a0, al, a2, 0, bl. b2 and b3, were then interpolated and extrapolated and their grid based maps created for the whole basin. Application of the Gamma function model involved optimisation of the gamma function parameter k, with n fixed to 2. Likewise, k was mapped on the basin surface. Thus estimation of flow on a seasonal basis for a pixel would require extraction of a0, al, a2, a3, bl, b2, b3 from the maps and applying them to the rainfall input and the pulse response ordinates would be calculated by using the mapped k values. The concept was tested on 20 catchments by running the inverted Linear Perturbation Model and acceptable values of R2 were obtained.