H13L-1741
Analyzing the Relative Linkages of Land Use and Hydrologic Variables with Urban Surface Water Quality using Multivariate Techniques
Monday, 14 December 2015
Poster Hall (Moscone South)
Shakil Ahmed and Omar I. Abdul-Aziz, West Virginia University, Civil and Environmental Engineering, Morgantown, WV, United States
Abstract:
We used a systematic data-analytics approach to analyze and quantify relative linkages of four stream water quality indicators (total nitrogen, TN; total phosphorus, TP; chlorophyll-a, Chla; and dissolved oxygen, DO) with six land use and four hydrologic variables, along with the potential external (upstream in-land and downstream coastal) controls in highly complex coastal urban watersheds of southeast Florida, U.S.A. Multivariate pattern recognition techniques of principle component and factor analyses, in concert with Pearson correlation analysis, were applied to map interrelations and identify latent patterns of the participatory variables. Relative linkages of the in-stream water quality variables with their associated drivers were then quantified by developing dimensionless partial least squares (PLS) regression model based on standardized data. Model fitting efficiency (R2=0.71-0.87) and accuracy (ratio of root-mean-square error to the standard deviation of the observations, RSR=0.35-0.53) suggested good predictions of the water quality variables in both wet and dry seasons. Agricultural land and groundwater exhibited substantial controls on surface water quality. In-stream TN concentration appeared to be mostly contributed by the upstream water entering from Everglades in both wet and dry seasons. In contrast, watershed land uses had stronger linkages with TP and Chla than that of the watershed hydrologic and upstream (Everglades) components for both seasons. Both land use and hydrologic components showed strong linkages with DO in wet season; however, the land use linkage appeared to be less in dry season. The data-analytics method provided a comprehensive empirical framework to achieve crucial mechanistic insights into the urban stream water quality processes. Our study quantitatively identified dominant drivers of water quality, indicating key management targets to maintain healthy stream ecosystems in complex urban-natural environments near the coast.