Dynamic factor analysis of surface water management impacts on soil and bedrock water contents in Southern Florida Lowlands

Abstract

As part of the C111 spreader canal project, structural and operational modifications involving incremental raises in canal stage are planned along one of the major canals (i.e., C111) separating Everglades National Park and agricultural production areas to the east of the park. This study used Dynamic Factor Analysis (DFA) as an alternative tool to physically based models to explore the relationship between different hydrologic variables and the effect of proposed changes in surface water management on soil and bedrock water contents in south Florida. To achieve the goal, objectives were to: (1) use DFA to identify the most important factors affecting temporal variation in soil and bedrock water contents, (2) develop a simplified DFA based regression model for predicting soil and bedrock water contents as a function of canal stage and (3) assess the effect of the proposed incremental raises in canal stage on soil and bedrock water contents. DFA revealed that 5 common trends were the minimum required to describe unexplained variation in the 11 time series studied. Introducing canal stage, water table evaporation and net recharge resulted in lower Akaike information criterion (AIC) and higher Nash-Sutcliffe (C[subscript eff]) values. Results indicated that canal stage significantly (t > 2) drives temporal variation in soil and bedrock water contents, which was represented as scaled frequency while net surface recharge was significant in 7 out of the 11 time series analyzed. The effect of water table evaporation was not significant at all sites. Results also indicated that the most important factor influencing temporal variation in soil and bedrock water contents in terms of regression coefficient magnitude was canal stage. Based on DFA results, a simple regression model was developed to predict soil and bedrock water contents at various elevations as a function of canal stage and net recharge. The performance of the simple model ranged from good (C[subscript eff] ranging from 0.56 to 0.74) to poor (C[subscript eff] ranging from 0.10 to 0.15), performance was better at sites with smaller depths to water table (< 1 m) highlighting the effect of micro-topography on soil and bedrock water content dynamics. Assessment of the effect of 6, 9 and 12 cm increases in canal stage using the simple regression model indicated that changes in temporal variation in soil and bedrock water contents were negligible (average<1.0% average change) at 500 to 2000 m from C111 (or low elevations) which may be attributed to the near saturation conditions already occurring at these sites. This study used DFA to explore the relationship between soil and bedrock water dynamics and surface water stage in shallow water table environments. This approach can be applied to any system in which detailed physical modeling would be limited by inadequate information on parameters or processes governing the physical system.