A stochastic-empirical approach to modelling nitrate leaching

Abstract. Techniques for determining the probability density function (pdf) of travel times of solute molecules through a defined volume of soil, following a pulse or step-change input to the soil surface, are described. A stochastic transfer function model (TFM) based on the pdf of nitrate travel times works satisfactorily when the nitrate originates from a pulse input of soluble fertilizer to the soil surface. However, a TFM based on the pdf of a surface-applied tracer, such as chloride or tritiated water, is less satisfactory for simulating the leaching of indigenous soil nitrate. The main problems seem to be the difficulty of estimating mean nitrate concentrations because of the spatial variability of nitrate in field soils, accounting for denitrification during leaching, and the uncertain reproducibility of the soil's transport characteristics, as embodied in its operationally defined fractional transport volume, θ _st , Nevertheless, for many practical applications, a simplified empirical model which treats the soil's transport volume as a well mixed reactor of average initial concentration C, can provide satisfactory predictions of the quantity of nitrogen leached over extended periods. Irrespective of which model is used, a comprehensive treatment of nitrate leaching, particularly for soil generated nitrate, requires a detailed knowledge of transfers of labile nitrogen within the transport volume, and across its boundaries other than those monitored at the input and output surfaces.