By Scott T. Prinos 

CORRELATION ANALYSIS OF A GROUNDWATER LEVEL MONITORING NETWORKAnalytical ConsiderationsThe results of a correlation analysis between two continuous variables, although technically correct, could be misleading under several circumstances. First, data that correlate may not be identical. For example, strongly associated water levels at two different stations may not have the same mean value. Waterlevel monitoring networks often are used to assess the slope of the potentiometeric surface as well as the variability of water levels. Therefore, even wells with strongly associated water levels, may provide additional waterlevel information that is needed. If the data are strongly associated, assessing the relation in greater detail will help to determine whether the data are redundant. A second consideration is that if the two continuous variables consist of timeseries data, then temporal variation in either the short or long term could result in misinterpretation of the results. For example, the seasonal variation in rainfall and hydrology could result in seasonal changes in correlation that may not be understood unless specifically considered in the analysis. Additionally, periodic or gradual declines in the degree of correlation caused by droughts, climatological cycles, or changes to the hydrologic system could both result in data that in the short term are substantially less correlated than indicated by a purely longterm analysis of correlation. With this in mind, a correlation analysis of data from a groundwater level monitoring network was considered for both the short and long term. Another previously discussed consideration is that the correlation analysis of two continuous variables does not establish causality. To understand the association, the data must be examined in greater detail. The groundwater level monitoring network in MiamiDade County is influenced by recharge from canals, operation of canal flow regulation structures, evapotranspiration, precipitation, and withdrawals for municipal water supply. Each of these factors can be represented spatially. Therefore, spatial examination of the correlation analysis results was used to help determine which correlations could have been influenced by the same causal factors. Next: Spatial Relations 
 