This paper investigates the applicability of using artificial neural network (ann) and multilinear regression models to predict urban stormwater quality at unmonitored catchments. Models were constructed using logarithmically transformed environmental data. Violation of the assumption of data independence lead to the inclusion of insignificant variables when a straightforward stepwise regression was applied. To overcome this problem, cross validation was used to determine when to stop adding variables. Regression models calibrated using event mean concentration (emc) as the dependent variable were C297 more accurate than those using event load. Regression models developed on a regional subset of data were more accurate than the models developed on the entire data set. Even though regression and ann models yielded similar predictions, regression modelling was considered to be a more applicable approach. Compared to ann models, regression models were faster to construct and apply, more transparent and less likely to overfit the limited data.
Heavy metals in urban stormwater runoff can adversely impact aquatic ecosystems. Successful management of such systems requires the accurate prediction of contaminant concentrations. This has created the need for simplistic statistical models. In this study, models were constructed to predict three of the most prevalent heavy metal constituents in urban stormwater: copper (Cu), lead (Pb) and zinc (Zn). Data from the United States, obtained during the Nationwide Urban Runoff Program (NURP), were used to calibrate and verify the models. A comparison of the models revealed that regression models were more accurate than the landuse-based or metropolitan area averages of event mean concentration (EMC).
Urban stormwater quality is influenced by many interrelated processes. However, the site-specific nature of these complex processes makes stormwater quality difficult to predict using physically based process models. This has resulted in the need for more empirical techniques. In this study, artificial neural networks (ANN) were used to model urban stormwater quality. A total of 5 different constituents were analyzed-chemical oxygen demand, lead, suspended solids, total Kjeldahl nitrogen, and total phosphorus. Input variables were selected using stepwise linear regression models, calibrated on logarithmically transformed data. Artificial neural networks models were then developed and compared with the regression models. The results from the analyses indicate that multiple linear regression models were more applicable for predicting urban stormwater quality than ANN models. Water Environ. Res., 80, 4 (2008). IntroductionProgressive urbanization has led to the degradation of many waterways. To prevent this from further occurring, best management practices must be implemented. To do this effectively, techniques must be used to predict the extent of the water quality problem. This is often a difficult task when considering the wide range of interrelated processes that affect urban stormwater quality. As a result, empirical techniques have been developed to quantify the extent of the water pollution problem.Artificial neural network (ANN) models have been used in a number of previous studies to forecast environmental variables (Maier and Dandy, 2000). For example, Lek et al. (1996Lek et al. ( , 1999 used ANN to predict nutrient concentrations at approximately 1000 predominantly ''natural'' waterways located in the United States. It was observed that ANN models were more accurate than multiple linear regression models constructed on logarithmically transformed data. The main advantage of ANN models is their ability to model complex nonlinear phenomena, without specifying the exact functional forms associated with the system under study (Lek et al., 1999). This infers that ANN models may be applicable when modeling urban stormwater quality.
Excessive quantities of nutrients in urban storm-water runoff can lead to problems such as eutrophication in receiving water bodies. Accurate process based models are difficult to construct due to the vast array of complex phenomena affecting nutrient concentrations. Furthermore, it is often impossible to successfully apply process based models to catchments with limited or no sampling. This has created the need for simple models capable of predicting nutrient concentrations at unmonitored catchments. In this study, simple statistical models were constructed to predict six different types of nutrients present in urban storm-water runoff: ammonia ͑NH 3 ͒, nitrogen oxides ͑NO x ͒, total Kjeldahl nitrogen, total nitrogen, dissolved phosphorus, and total phosphorus. Models were constructed using data from the United States, collected as a part of the Nationwide Urban Stormwater Program more than two decades ago. Comparison between the models revealed that regression models were generally more applicable than the simple estimates of mean concentration from homogeneous subsets, separated based upon land use or the metropolitan area. Regression models were generally more accurate and provided valuable insight into the most important processes influencing nutrient concentrations in urban storm-water runoff.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.