Quantifying the contributions of structural factors on runoff water quality from green roofs and optimizing assembled combinations using Taguchi method

2020 
Abstract Runoff water quality of green roofs often comes under debate, and the mechanism of runoff pollution retention is still unclear. How to quantify the influencing contributions of structural factors to runoff pollution of green roofs and optimize the assembled combinations for improving water quality are ambiguous. In this study, runoff plots of extensive green roofs with Taguchi designed structural factors and levels were constructed and simulated rainfall experiments were conducted. Influences of structural factors on outflow water quality of green roofs were statistically assessed and quantified. Runoff water quality of green roofs with assembled combinations at specific levels were optimized and predicted by using the Taguchi method. Results showed that except for the pH and NO- 3 concentrations, the extensive green roofs acted as a source of the tested pollutants in stormwater runoff. Contributions of substrate materials on pH, EC, ESP (exchangeable sodium percentage), F-, NO- 3 and NO2-N concentrations were the highest among the structural factors, and contribution percentages ranged from 33.38% to 64.47%. The vegetation types had important contributions on Cl-, SO2- 4 and TP concentrations, and the contribution percentage was 74.72%, 71.23% and 45.16% respectively. Influences of substrate depths and slope gradients on outflow water quality were small. Most of the determination coefficients (R2) of regression analysis between the measured and predicted water quality parameters under Taguchi design were ranged from 0.843 to 0.997. Except the pH and NO- 3 parameters, the Taguchi predicted water quality under the optimum conditions were improved by 10.2-77.6%. These results will aid understanding of the retention mechanism of green roofs on runoff pollutants and improving runoff water quality through the optimization of green roof design.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    73
    References
    4
    Citations
    NaN
    KQI
    []