Background

As reported by Hutson and Wagenet (1992) that “mathematical modelling is an accepted scientific practice, providing a mechanism for comprehensively integrating basic processes and describing a system beyond what can be accomplished using subjective human judgements. As our understanding of basic processes deepens, it is possible to construct models that better represent the natural system and to use these models in an objective manner to guide both our future research efforts and current management practices”.

Different approaches have been used to describe the movement of water and solute movement in field soils. Few models have been developed to study the solute leaching patterns. The models that have been developed so far vary widely in their conceptual approach and degree of complexity, and are strongly influenced by the environment, training, and biases of their developers. Many of the models have been produced as the result of research into the basic physics and chemistry of salt, N or pesticide transport and transformation in agricultural soils (Hutson and Wagenet, 1992; Vanclooster et al., 1995; Kroses and Roelsma, 1998; Abrahasen and Hansen, 2000; Leaonard et al., 1987; Eckersten et al., 1994; Close et al., 2003, 2005; Nolan et al., 2005).

Over the past four decade, the development and use of simulation models for predicting nutrient and pesticide behaviour in the root zone of agricultural land systems, and in the underlying unsaturated zone, has received considerable attention. The models currently available for predicting the fate and transpiration of pesticides and nutrients in soils and groundwater have been critically reviewed by, among others, (Addiscott and Wagenet, 1985; Donigian and Rao, 1990; Wagenet and Rao, 1990; Cichota and Snow, 2009). There has been a considerable debate over the appropriate conceptual representation of environmental processes used in nutrients and/or pesticides simulation models and as well as the formulation of key criteria for evaluation and comparing the ability of

models to predict observed pesticide or nutrients behaviour. Although considerable efforts have put in to develop nutrient simulation models (Wagenet and Rao, 1990), comparatively little work has been done to validate these models using independent data sets. This is primarily due to the fact that considerable physical and financial resources are required to conduct field studies. In addition, many of the field studies conducted to date lack adequate measurement of input parameters required to run the model, and field data required evaluating the models. As highlighted earlier, the need for efficient use of agricultural chemicals and their potential adverse impact on critical water resources have increased the use of simulation models of the soil and plant system. Nevertheless, there is currently little or no agreement concerning model validity and applicability in varied soils and environments (Sogbedji et al., 2001).

Hutson and Wagenet (1992) reported that “models have not been used for management

purposes in the past. A major reason is that there is apparently little recognition in most model development efforts that different types of models are developed for different purposes. The quantity of required input data, depth of consideration of basic processes, and sensitivity and accuracy of simulations all depend upon whether the modeller intends to approach the simulation from a research or management perspective. A short discussion or review on the range of modelling experiences on water and solute movement in soil may establish the logic behind the development of a model. A review of modelling approaches (Addiscott and Wagenet, 1985; Cichota and Snow, 2009) identified a number of research and management models that have been reported in the scientific literature. The research models were identified as generally intended to provide quantitative estimates of water and solute movement, but with comprehensive data demands regarding the system to be simulated. Additionally, few of either type of model have been tested against field data, and little attention has been paid to the use of the so-called management models for the actual purposes of managing applications to soil of salty irrigation water, effluents, fertiliser, amendments, pesticides or other solute. Appreciation of the strengths and weaknesses of these models is a necessary preliminary step before proceeding to use a model.

Deterministic, mechanistic models of solute movement based upon miscible displacement theory (Nielsen and Biggar, 1962) have been the most widely used modeling approach in soil science over the past 50 years. Hutson and Wagenet (1992) mentioned that “these models presume that soil-water and solute displacement

processes operate so that the occurrence of a given set of physical and chemical events leads to a uniquely definable water or solute distribution in the soil profile. Water flow is assumed to be describable as the product of hydraulic gradient and water content dependent hydraulic conductivity. It is assumed that physical convection (mass flow) and chemical diffusion combine to displace a solute in porous media”. This type of solute model is summarised in the Convection-Dispersion Equation (CDE), which has been derived in detail in Kirkham and Powers (1972), and solved analytically for a variety of particular initial and boundary conditions (reviewed by van Genuchten and Alves, 1982). These analytical solutions, representing particular research models of water and solute movement have been successfully applied to the results of laboratory scale studies in which water flow is a constant (i.e. steady state rate). Based on these studies, the CDE is a well-established solute modeling approach for such cases (Hutson and Wagenet, 1992).

The movement of both water and solute under field conditions varies with depth and time; require numerical (rather than analytical) solution of the CDE to properly represent the influence of changing water contents, fluxes and solute concentrations. Further, the Richard’s equation (i.e. the transient water flow equation) must be numerically solved to describe the water flow regime. Research models that have been constructed based on Richard’s equation are numerous, including those that simulate plant growth in response to transient water regimes (Nimah and Hanks, 1973), nitrogen movement and transformation (Tillotson and Wagenet, 1982), management of irrigation water and salts of wastewater (Iskander and Selim, 1981). The LEACHM model is similar in structure to these models, and has evolved from modeling experiences of the last two decades. It has already been successfully used to describe pesticide movement in field soils (Wagenet et al., 1989; Close et al., 1999) and is used by many research groups in New Zealand, United States and other countries (Hutson and Wagenet, 1992;

Pearson et al., 1996; Mahmood et al., 2002a, 2002b; Mahmood, 2005; Close et al., 2003, 2005).

As we know that it is impractical to directly measure the nutrient losses to the environment and therefore simulation models could be the best alternative to asses the potential loss of nutrients to the local environment. However, there are a number of models that have been used in New Zealand to simulate nutrient loss from different sources to the natural environment. Cichota and Snow (2009) presented an overview of some models that are relevant to New Zealand pastoral farms. They reviewed the models for nutrient loss estimation that are used in New Zealand at farm scale (including OVERSEER®, NPLAS - Nitrogen and Phosphorus Load Assessment System, SPASMO - Soil-Plant-Atmosphere System Model, EcoMod, LUCI - Land Use Change and Intensification, and APSIM - Agricultural Production Systems Simulator) and at large scales i.e. catchment level (including NLE - Nitrogen Leaching Estimation, SPARROW - Spatial Referenced Regression On Watershed attributes), ROTAN - ROtorua and TAupo Nitrogen, CLUES - Catchment Land Use and Environmental Sustainability, and AquiferSim).

Cichota and Snow (2009) also reviewed some soil process models (e.g. GLEAMS (Groundwater Loading Effects of Agriculture Systems), LEACHM, and HYDRUS). In their review, they mentioned that nutrient management models are being extensively used in New Zealand. They said that most of the models have shown their usefulness to estimate nutrient losses in order to prevent environmental impacts at a small or larger scale levels. However, there is a lack of information about how these models work and what is their main focus, and therefore, it is important to know the main purpose, strengths, and weaknesses of a model so that the most appropriate model could be selected.