Simulation of nitrogen and hydrology loading of forested fields in eastern North Carolina using DRAINMOD-N II.

Abstract

A new version of DRAINMOD-N (DRAINMOD-N II) was used to evaluate the combined effects of soil variability, vegetation, drainage intensity, climate, and management practices on the hydrology and nitrogen (N) transport in forests. A better understanding of these processes will be useful in the development of management practices for reducing N loads from forests and in future large-scale modeling studies.

The objective of this study was to accurately model nitrogen loading at the field scale for three Coastal Plain forests in North Carolina using DRAINMOD-N II. Supporting objectives were to accurately model the hydrology of three forested fields using DRAINMOD, to determine litterfall and N uptake at the study sites, and to evaluate DRAINMOD and DRAINMOD-N II model accuracy by comparing predictions with measured values.

Water table elevations, drainage losses, and water quality were continuously measured at the study sites from 1995-2001. Soils on two of the fields were organic; the third field had a highly organic mineral soil. DRAINMOD was used to simulate the hydrology of the forested sites from 1995-2001. Very porous, highly organic soils made it difficult to determine hydrology input parameters using standard field methods. Several input parameters were calculated from measurements of water table elevation and drainage outflow or from model calibration.

DRAINMOD-N II was used to predict cumulative N process rates and N losses in drainage. Litterfall production and N uptake inputs were determined using the forest productivity model PnET-CN. Michaelis-Menten input parameters for nitrification and denitrification in DRAINMOD-N II were determined by calibration. N mineralization was modeled as a function of organic matter (OM) content in the soil (initial OM and OM added by litterfall) and organic carbon decomposition rates.

DRAINMOD predicted water table elevations and drainage losses reasonably well when compared to observed data. Despite difficulties encountered in accurately determining the soil properties of the forest surface layers, and the hydrologic effect of maturing trees on evapotranspiration, the average absolute daily difference (AADD) of the water table depth predictions from 1995-2001 ranged from 13.0-28.4 cm, and the model efficiency, E, of the water table depth predictions ranged from 0.59-0.83. R2 values for the daily drainage rate predictions from 1996-2001 ranged from 0.69-0.85. Model efficiency (E), values for the daily drainage rate predictions from 1996-2001 ranged from 0.68-0.70. The normalized errors in cumulative drainage predictions from 1996-2001 ranged from -17.1 to 2.7 %.

DRAINMOD-N II performed reasonably well in predicting N concentrations and cumulative N loads. Using a three-year calibration, the model generally overpredicted N losses during the validation period. Normalized errors in predicting cumulative NO3-N loads from 1996-2001 ranged from -2.5 to 28.9 % for the three fields. Normalized errors in predicting cumulative NH4-N loads for the 1996-2001 period ranged from 48.2 to 54.6 %. Normalized errors in predicting cumulative dissolved inorganic nitrogen (DIN) loads ranged from -6.4 to 23.9 %. The model was also calibrated using a six-year calibration for a better understanding of N transport processes and for use in future modeling studies.

The results of this study documented the reliability of DRAINMOD for predicting water table depth and outflow volume from forested fields on highly organic soils. The study also showed the potential of DRAINMOD-N II for simulating N fate and transport in forested systems. DRAINMOD-N II predictions of N loads depended on the DRAINMOD hydrology predictions. Most of the input parameters for DRAINMOD-N II were determined from the literature or by calibration. A more accurate evaluation of the applicability of DRAINMOD-N II for modeling forested systems will require more field and/or laboratory measurements to determine model inputs.