A statistical power analysis approach to estimate groundwater-monitoring network size in Victoria County Groundwater Conservation District, Texas

A properly designed groundwater-monitoring network is critical to evaluate groundwater management policies and regulations and establish a scientifically credible, risk-informed approach towards management of aquifer resources. To achieve proper design, the network must comprise a sufficient number of wells that comprehensively covers the area of interest. In this study, statistical power analysis is utilized to define the size of a proposed monitoring network within the Victoria County Groundwater Conservation District (VCGCD). In particular, the network seeks to provide empirical information to evaluate whether the VCGCD is meeting the criteria of the desired future conditions (DFC) established by the Groundwater Management Area 15 (GMA 15) through the joint planning process, as required by the statutes of Texas. The evaluation of DFCs is expressed as a hypothesis-testing problem with the null hypothesis stating that the VCGCD complies with the DFC, and the alternative hypothesis stating that the VCGCD does not comply with the DFC. The power analysis quantifies the ability of the statistical test to correctly reject the null hypothesis. The power of the statistical test is a function of the sample size and this relationship can be exploited to determine the size of the required groundwater-monitoring network if the effect size can be specified. However, prior to establishing a monitoring network, the groundwater level monitoring tends to be ad hoc; therefore, the statistical moments required for characterizing the effect size are not likely to be known with a high degree of certainty. As such, an innovative framework that integrates power analysis with bootstrap resampling protocols was developed to estimate monitoring network size under uncertainty. The results of the study indicated that a minimum set of approximately 70 wells is required within the VCGCD to statistically evaluate the compliance with DFCs with 90 % reliability and at a significance level of 10 % and 90 % power. The number of wells increased to about 85 when the significance level was reduced to 5 %. Geostatistical analysis indicated that these monitoring wells must be at least 3 miles apart to ensure statistically independent information. The developed methodology provides a practical framework to size a groundwater-monitoring network under data sparse situations.