By Scott T. Prinos |
|
|
CORRELATION ANALYSIS OF A GROUND-WATER LEVEL MONITORING NETWORKA correlation analysis can be used to assess the strength of association between two continuous variables, but cannot be used to prove a casual relation between the two variables (Helsel and Hirsch, 1992). Three main methods of determining correlation are Pearson's r, Spearman's p, and Kendall's T. Pearson's r is used to evaluate linear correlation, Spearman's p is used to evaluate nonlinear correlation, and Kendall's tau can be used to evaluate either linear or nonlinear correlation. Spearman's p and Kendall's T are both rank-based methods. Correlation analysis can be applied to water-level data from two or more monitoring wells to determine whether or not the data are strongly associated. A correlation analysis alone cannot indicate which wells should be removed or added. In fact, any purely statistical analysis probably could not provide this information without considering additional factors. The statistical analysis generally must be used in conjunction with a thorough understanding by water managers of the hydrologic system and the optimal resolution of data required from each location in the network to meet the necessary monitoring needs. The required resolution may vary spatially. For example, small head differentials in nested wells may be critical in some parts of the network, but unimportant in other parts. If numerous wells are providing essentially the same water-level data from the same part of the aquifer, then this implies redundancy and some unnecessary monitoring wells could possibly be eliminated. If the water-level data from each monitoring well are unique in a specific part of the aquifer, then those data would likely be necessary to fully characterize the hydrologic conditions of the aquifer, and the number of monitoring wells in that part of the aquifer may need to be increased, depending on the level of precision needed. |
| ||||||||||||||||||||||||||||||||