- 5 views
- 9 downloads
Response of Urban Stormwater Quality to Rainfall Characteristics and Land-Use/Land-Cover
-
- Author / Creator
- Yan, Haibin
-
Urban stormwater quality is a critical concern in urban planning and stormwater management due to its significant influence on receiving water bodies and ecosystems. The dynamics of stormwater quality is influenced by complex factors, with rainfall characteristics and land-use/land cover (LULC) playing primary roles. The joint effects of rainfall characteristics and LULC remain inadequately understood. This research aims to enhance the understanding of how urban stormwater quality responds to these variables and utilize the created knowledge to improve the performance of urban stormwater quality models.
First, a data mining framework was designed to study the impacts of rainfall characteristics on urban stormwater quality. Specifically, the relationship between rainfall characteristics and stormwater quality was studied. Rainfall events were classified using a K-means clustering method. A rainfall event type-based (RTB) calibration approach was used to improve water quality model performance. Antecedent dry days, average rainfall intensity, and rainfall duration were the most critical rainfall characteristics affecting the event mean concentrations (EMCs) of total suspended solids (TSS), total nitrogen (TN), and total phosphorus (TP). The RTB calibration approach can improve water quality model accuracy. The calibrated stormwater quality parameters could be transferred to adjacent catchments with similar characteristics.
Secondly, to study the impacts of land cover on the simulation of stormwater runoff and pollutant loading, a land-cover based (LCB) PCSWMM model was built. The LCB approach performed better than the watershed delineation tool (WDT) approach in hydrological simulation. The two models showed comparable performances in simulation of TSS, TN, and TP. The LCB approach parameters could be regionalized based on land cover types. The hydrologic-hydraulic parameters could potentially be transferred to similar catchments. The transferring of water quality parameters did not perform as satisfactory. The LCB approach could quantitively evaluate the contribution to runoff and pollutant loads of different land covers. Roads and roofs were found to be the major contributors to urban runoff and pollutants in the two urban catchments.
Thirdly, to study the effects of mixed land use on urban stormwater quality under different rainfall event types, the dataset from the two-year field monitoring program in four urban catchments in Calgary was utilized. EMC and event pollutant load (EPL) were employed to evaluate TSS and nutrients. EMC of TSS was positively correlated with most nutrient components. EPLs exhibited higher correlation compared to EMCs. Mixed land use could influence the generation of stormwater pollutants. Intense rainfall and long antecedent dry days could yield higher EMC and EPL and long rainfall duration could generate elevated EPL. Seasonal variations were found in EMC and EPL, with higher values in the spring and summer than the fall.
Fourthly, to investigate the effects of land use and rainfall conditions on the particle size distribution (PSD), the dataset from a six-year water sampling program across 15 study sites in Calgary was used to characterization of PSDs. The median particle size decreased in the order: paved residential, commercial, gravel lane residential, mixed land use, industrial and roads. Fine particles are the dominant particles of suspended sediments in runoff in Calgary. The impact of rainfall event types could vary depending on land use types. Long antecedent dry period tends to result in the accumulation of fine particles. High rainfall intensity and long duration could wash off more coarse particles. The season is a factor in the PSD of suspended sediment in runoff. The PSD in spring exhibits the finest particles. Particles in snowmelt are finer for the same land use than that during rainfall events.
Finally, to predict stormwater quality in data-deficient areas using data-driven models, a semi-supervised machine learning framework was proposed. This approach was applied in four catchments in Calgary to predict EMCs of TSS, TN and TP. This study demonstrated that machine learning is an effective tool for predicting stormwater quality. Relying on limited data and features may lead to overfitting. By integrating data from different catchments, the model performance could be enhanced. Consideration of catchment characteristics could improve the model capacity. The pseudo-labeling learning could enrich the training dataset, and this semi-supervised machine learning approach could elevate the model predictive ability. Rainfall characteristics are important variables for predicting all three pollutants, with varying focus on duration, amount, intensity and antecedent dry days. -
- Graduation date
- Spring 2024
-
- Type of Item
- Thesis
-
- Degree
- Doctor of Philosophy
-
- License
- This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.