[1] Understanding the hydrological functioning of tidally influenced floodplain forests is essential for advancing ecosystem protection and restoration goals in impacted systems. However, finding direct relationships between basic hydrological inputs and floodplain hydrology is hindered by complex interactions between surface water, groundwater, and atmospheric fluxes in a variably saturated matrix with heterogeneous soils, vegetation, and topography. Thus, an explanatory method for identifying common trends and causal factors is required. Dynamic factor analysis (DFA), a time series dimension reduction technique, models temporal variation in observed data as linear combinations of common trends, which represent unidentified common factors, and explanatory variables. In this work, DFA was applied to model water table elevation (WTE) in the floodplain of the Loxahatchee River (Florida, USA), where altered watershed hydrology has led to changing hydroperiod and salinity regimes and undesired vegetative changes in the floodplain forest. The technique proved to be a powerful tool for the study of interactions among 29 long-term, nonstationary hydrological time series (12 WTE series and 17 candidate explanatory variables). Regional groundwater circulation, surface water elevations, and spatially variable net local recharge (cumulative rainfall -cumulative evapotranspiration) were found to be the main factors explaining groundwater profiles. The relative importance of these factors was spatially related to floodplain elevation, distance from the river channel, and distance upstream from the river mouth. The resulting dynamic factor model (DFM) simulated the WTE time series well (overall coefficient of efficiency, C eff = 0.91) and is useful for assessing management scenarios for ecosystem restoration and predicted sea level rise.