Considerations and Pitfalls in the Spatial Analysis of Water Quality Data and Its Association With Hydraulic Fracturing
Chapter
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
Linking areas of high unconventional oil and gas development (UD) activity to groundwater quality is statistically challenging. As contaminant pathways reflect spatial processes, their elucidation requires spatially explicit analyses. Here, we consider complications in the statistical evaluation of suites of chemical constituents, review basics of spatial analysis, and illustrate geographically weighted regression (GWR) and Hot Spot Analysis (spatial clustering) using eight indicator variables from groundwater samples collected from the Trinity and Woodbine aquifers overlying the Barnett Shale in northern Texas, a region of high UD activity. GWR indicated that moderate variation in some variables (e.g., total dissolved solids) but zero variance in others (e.g., methanol) is explained by kernel density of UD wells. Hot Spot Analysis complemented GWR analyses and indicated several subregions of elevated concentrations for most variables. With the exception of a single area of extreme contamination straddling the ParkerHood County line, hot spots showed little to moderate spatial congruence across variables. Collectively, our results suggest that while some groundwater contamination has resulted from UD activity, overall groundwater contamination is multifactorial, and contamination related to UD activity is likely stochastic rather than systematic.
