Comparative modelling of phosphorous production in rural catchments

[摘要] ENGLISH ABSTRACT: The objective of this research has been to compare nonpoint sources assessment techniques forsimulating phosphorous production in rural catchments which have a variety ofland use types. Fournonpoint source assessment techniques capable of simulating phosphorous production, operating atdifferent spatial and temporal resolutions, were selected after an intensive literature review. Themodel selection criteria included the capability to simulate phosphorous production, the need for thestudy to cover a range of spatial and temporal resolutions, model data requirements, modelaffordability and availability in South Africa. The models selected using these criteria are thePhosphorous Export Model (PEM) (Weddepohl & Meyer, 1992), Impoundment and RiverManagement and Planning Assessment Tool for Water Quality Simulation Model (IMPAQ)(DWAF,1995), the Hydrological Simulation Program Fortran (HSPF) (Bricknell,1993) and theAgricultural Catchments Research Unit Model (ACRU) (Smithers and Caldecott, 1994).Four ofthe study catchments were selected within the Berg River basin in the Western Cape and theremaining four were selected within the Amatole catchments in the Eastern Cape. The four subcatchmentsin the Berg River basin are the Twenty-Four Rivers, Leeu River, Kompanjies River andDoring River catchments and the four in the Amatole catchments are the Upper Buffalo, Cwencwe,Yellowwoods and Gqunube River catchments. The range of land use/cover types comprises:Western Cape catchments : wheat, grapes, natural vegetation and forestryEastern Cape catchments : natural vegetation and forestryThe PEM and IMPAQ models were applied reasonably successfully to all the catchments to simulatephosphorous production, with the observed flow as the input. The HSPF model could notsuccessfully be applied to the catchments to simulate both the catchment hydrology and phosphorousproduction. Hence, the investigation into HSPF was abandoned, and in its place, the ACRU dailyphosphorous yield model was incorporated at a fairly late stage in the research. ACRU was appliedto only the Western Cape catchments.The estimated parameters for different land use types were compared to investigate the potential forparameter transfer in space and time. Both the PEM and IMP AQ models showed promise that landuse parameters could be transferred in time for catchments located in the Western Cape catchments,but did not show promise for catchments located in the Eastern Cape. The IMPAQ model showedpromise that land use parameters could be transferred in space for catchments located in the EasternCape, but did not perform as well in the Western Cape catchments. The PEM model showed promisethat land use parameters could be transferred in space for catchments located in the Western Cape,but did not perform as well in the Eastern Cape. Since the ACRU phosphorous yield model wasincluded at a late stage of the research, the potential for land use parameter transfer in space and timecould not investigated. The model results were verified at the relevant flow and water qualitygauging stations. The ACRU phosphorous model verification results showed promise for catchments located in humidparts of the Berg River basin, but did not perform as well in the catchment located in the semi-aridpart.RECOMMENDATIONS FOR FURTHER RESEARCH:I. Intensive research should be undertaken to develop a database ofland use parameters/ exportcoefficients related to phosphorous production (and other non-conservative constituents) inSouth African catchments. Availability of these parameters would make phosphorousmodelling much easier.HSPF should be configured and calibrated, more especially its water quality component, forcatchments with hourly rainfall and rainfall stations located within/on the catchmentboundaries, to investigate its performance under South African conditions. Given thecomplexity of the HSPF algorithms and the time required to familiarise oneself with themodel, it is recommended that such an investigation be undertaken which is not inclusive ofany other models.The spatial resolution ofPEM is extremely coarse, and should be improved to allow the userto partition the total flow in the catchment according to contributions from the variety oflanduse types and to estimate soluble and particulate phosphorous parameters for each land usetype.A study should be undertaken to investigate the potential for the ACRU phosphorous yieldmodel parameter transfer in time and space.Sampling frequency of water quality data in South Africa should be improved, because it isdifficult to assess the performance of the calibrated water quality models, more especiallyphosphorous export models, due to a lack of continuous data sets.Rainfall data collection in gauged catchments, more especially Western Cape catchments(e.g. Twenty-Four Rivers, Leeu, Kompanjies and the Doring River catchments), should beimproved. There should be at least one rainfall gauging station located within the catchmentboundaries. This would contribute towards achieving reasonable hydrological calibration orverification. Since runoff is the driving factor for water quality components, improvedhydrological calibration/verification would result in reasonable water qualitycalibration/verification.

[发布机构] Stellenbosch University