Robustness of the nitrate transfer model NIT-DRAIN in an artificially drained agricultural area

Nitrate leaching due to excessive agricultural fertilization affects the quality of both surface and ground water. The presence of subsurface drains in agricultural areas introduces significant modifications to the hydrological behavior and results in the increase of nutrients and fertilizers losses from farmland to surface water. The recent development of the NIT-DRAIN conceptual model allows the simulation of nitrate transfer at the agricultural drainage system outlet and the estimation of the initial prewinter nitrogen pool (PWNP), equivalent to the remaining nitrogen pool at the start of winter season. This model was applied to three representative drained agricultural areas in France (La Jaillière, 1 ha; Gobard, 36 ha and Rampillon, 355 ha). The hourly drainage discharge and nitrate concentration data are recorded over a period of several years. The objective of this study is to evaluate of the spatiotemporal robustness of the NIT-DRAIN model, by testing the functioning of the model regarding a single or a generic set of model parameters for the three study sites.

The results showed that the model estimation of the PWNP is more precise at the small scale (Jailliere site). At the large scale, the PWNP estimation is slightly different from the measurement (<10kgN/ha). The model calibration for each study site shows high model performance for nitrate fluxes and concentrations, with Nash-Sutcliffe criterion greater than 0.6. These performances are preserved when calibrating a generic set of parameters to all the three sites. These results validate the robustness of the NIT-Drain model. This study present a simplified and operational approach for the quantification of PWNP applied to the subsurface drained agricultural lands by measuring nitrate concentration at outlet instead of soil core sampling.