Please use this identifier to cite or link to this item:
|Title:||Characterization and modeling of storm runoff quality from a tropical catchment with diverse land use||Authors:||Le, Song Ha||Keywords:||DRNTU::Engineering::Environmental engineering::Water supply||Issue Date:||2014||Source:||Le, S. H. (2014). Characterization and modeling of storm runoff quality from a tropical catchment with diverse land use. Doctoral thesis, Nanyang Technological University, Singapore.||Abstract:||The increasing deterioration of runoff water quality and its effect on receiving waterbodies has raised interest in characterizing catchment water quality and developing predictive models for stormwater loads and concentrations. In contrast to the vast literature on stormwater characterization and modeling in temperate catchments, limited studies have been conducted in tropical catchments. Tropical catchments however, may exhibit significantly different stormwater quality characteristics due to the much higher rainfall intensities, flashier response and higher runoff volume which are typical in these catchments. This thesis therefore, aimed to improve the characterization and modeling of stormwater quality for a tropical catchment with diverse land use, with the Kranji Reservoir catchment in Singapore adopted as the study catchment. In this study, 10 sub-catchments comprising urban, agricultural and forest land use were investigated. A total of 1424 samples collected from 113 storm events were analyzed for Total Nitrogen (TN), Nitrate (NO3), Amonium (NH4), Total Phosphorus (TP), Total Dissolved Phosphorus (TDP), Ortho-phosphorus (OP) and Total Suspended Solids (TSS). The results from the characterization study showed that the event mean concentrations (EMCs) of TSS and nutrients from agricultural catchments, especially from land-based vegetable farming activities were significantly higher than from urban catchments. An analysis of the results showed rainfall depth, duration, and runoff volume to be negatively correlated with the EMCs for small and highly impervious catchments, which may be attributed to the flashy response and the large runoff volumes typical of highly urbanized catchments in the tropics. In addition, correlations between EMCs with Antecedent Dry Period (ADP) were found to be insignificant, which may be attributed to the short inter-storm periods. First flush was quantified using four methods: (1) Fitting the function L = Vb and comparing the values of the exponent, b (Bertrand-Krajewski et al., 1998), (2) Determining the maximum distance between cumulative load and cumulative runoff volume curves (Geiger 1987), (3) Cumulative mass delivered at 20% and 30% of cumulative runoff volume or FF20 and FF30, respectively (Bertrand-Krajewski et al., 1998, Deletic 1998), and (4) Characteristic pollutographs (Bach et al., 2010). The first flush of TSS was found to be stronger in agricultural than in urban catchments, which may be due to the erodibility of the surficial soil layer, as a result of agricultural activities. Among the four methods, the FF30 is the most stringent method, followed by the FF20 method. The first flush for TN, NO3, TDP and OP was only detected using the Bach et al., (2010) method. However, the detection of first flush by the Bach et al., (2010) method for these parameters has to be considered with care as this method is sensitive to the relative values of the initial and background concentrations. The time series of TSS washoff rate was simulated using the empirical washoff model in which the washoff rate W (mg/hr) is proportional to the runoff rate q (mm/hr) and the initial mass on surface Bini (kg/ha). A sensitivity analysis was carried out using the Monte-Carlo technique to obtain the optimal values of the washoff model parameters: washoff coefficient c3, washoff exponent c4 and initial mass on surface Bini for the sampled events and test the dependence of these parameters on external factors. Correlation analyses were carried out for these parameters with rainfall and flow characteristics in addition to other factors such as land use and catchment size. These analyses showed c3 and Bini to be functions of rainfall depth, land use and catchment size. No correlations of Bini with rainfall depth of the previous events and ADP were found. In addition, the slopes of the regression between c3 and Bini with the rainfall depth increase with catchment area for the catchments under study. This result implies that washoff behavior in smaller catchments is less sensitive to changes in rainfall depth compared to larger catchments. Based on the results from the sensitivity study, calibration and uncertainty analysis of the washoff model in which Bini and c3 are functions of rainfall depth was conducted. This allows the value of c3 to be varied across events rather than assuming c3 to be constant. In addition, relating Bini to rainfall depth instead of the antecedent dry period may be more suitable for tropical catchments. The calibration of the washoff model which takes into account the dependence of Bini and c3 on rainfall depth provides a physical basis for the choice of c3 and Bini. The results obtained from the calibration and uncertainty analysis in this study are comparable or better than the results obtained from a traditional application of the buildup and washoff model where the buildup was a function of antecedent dry period, calibrated using the Generalized Likelihood Uncertainty Estimation (GLUE) method.||URI:||http://hdl.handle.net/10356/55444||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||CEE Theses|
Page view(s) 50320
checked on Oct 21, 2020
checked on Oct 21, 2020
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.