Forested and agricultural land use impacts on subsurface floodplain storage capacity using coupled vadose zone-saturated zone modeling

A floodplain study was implemented in central Missouri, USA, to better understand land use effects on floodplain storage capacity. Study sites included a heavily instrumented historic agricultural field (Ag) and a remnant bottomland hardwood forest (BHF). The groundwater flow model MODFLOW was used to simulate hydrology for the study. The model was calibrated using observed 2011 and 2012 water year data including precipitation, groundwater level, volumetric water content (VWC), and evapotranspiration (ET) estimated using Penman–Monteith. The model performed similarly well for both sites and is considered robust given optimization constraints, as supported by model fit for head (BHF: Root Mean Square Error or RMSE = 0.40 m; and Ag: RMSE = 0.31 m), VWC (BHF: RMSE = 2.5 %; and Ag: RMSE = 2.3 %), and ET (BHF: RMSE = 0.74 mm; and Ag: RMSE = 0.84 mm). Model predictions for the full vertical profile at the BHF site (median = 520 mm) indicated 28 % more vadose zone storage than the Ag site (median = 409 mm) on a median percent difference basis. On a cumulative basis, the BHF site accepted 117 mm (75 %) more water into storage than the Ag site. Subsurface storage differences between the two sites were likely, at least in part, the result of subsurface flow paths and evapotranspiration rates, which differed by more than 16 % (BHF > Ag). Results highlight several important differences in hydrology between the two land use types and support the re-establishment of riparian forests in floodplains of the Midwest as a strategy to address increasing flood concerns.