A daily water table depth computing model for poorly drained soils
The objective of this paper is to present a relatively simplified model to predict daily water table (WT) by solving ordinary differential equation dWT (t)/dt = F (α1, α2, α3, WT0(t), RF (t), PET (t)), with α1, α2, α3, WT0 as parameters, and RF (rainfall) and PET (potential evapotranspiration), respectively, as inputs. The model was calibrated and validated with WT on four poorly to moderately drained soils (Lenoir, Rains, Lynchburg, and Goldsboro) on a forested wetland. Calibration results were in good agreement with the measured WT for all soils, except the Goldsboro with deeper WT. r2 (coefficient of determination) and NSE (Nash-Sutcliffe Efficiency) statistics both ranged from 0.81 for the Lenoir to 0.89 and 0.87, respectively, for the Lynchburg. Average absolute daily deviation (AADD) varied from 10.8 cm for Lenoir to 16.7 cm for Rains. The performance was somewhat poorer, during relatively dry periods with deeper WT, yielding r2 and NSE as low as 0.55 and 0.29, respectively, for Lenoir, and large AADD for Lynchburg. Discrepancies were associated with WT overprediction for deeper depths. The new model is capable of describing the WT for poorly drained high water table soils, with a potential for assessing effects of land management, wetland hydrology, and climate changes.