Introduction to PRZM3
PRZM3 links two models, PRZM and VADOFT to predict pesticide transport and transformation down through the crop root and vadose (unsaturated) zone to the water table. PRZM3 incorporates soil temperature simulation, volatilization and vapor phase transport in soils, irrigation simulation, and microbial transformation. PRZM is a onedimensional finitedifference model which uses a method of characteristics (MOC) algorithm to eliminate numerical dispersion. VADOFT is a onedimensional finiteelement code that solves Richards' equation for flow in the unsaturated zone. The user may make use of constituting relationships between pressure, water content, and hydraulic conductivity to solve the flow equation. PRZM3 is capable of simulating multiple pesticides or parentdaughter relationships. PRZM3 is also capable of estimating probabilities of concentrations or fluxes in or from various media for the purpose of performing exposure assessments. PRZM and VADOFT are linked together with the aid of a flexible execution supervisor that allows the user to build models that are tailored to sitespecific situations. Monte Carlo pre and postprocessors are provided in order to perform probabilitybased exposure assessments.
PRZM3 Applications
PRZM3 is a U.S. EPA Model for Predicting Pesticide Fate in the Crop Root and Unsaturated Soil Zones. PRZM3 simulates the transport of fieldapplied pesticides in the crop root zone and the vadose zone taking into account the effects of agricultural management practices. The model provides estimates of probable exposure concentrations by taking into account the variability in the natural systems and the uncertainties in system properties and processes. The program utilizes extended memory.
PRZM3 links two models PRZM and VADOFT in order to predict pesticide transport and transformation down through the crop root and unsaturated zone. PRZM is a one dimensional, finitedifference model that accounts for pesticide fate in the crop root zone. PRZM3 incorporates several new features, specifically:
 Soil temperature simulation
 Volatilization and vapor phase transport in soils
 Irrigation simulation
 Microbial transformation
 A method of characteristics (MOC) algorithm to eliminate numerical dispersion
PRZM is capable of simulating transport and transformation of the parent compound and as many as twodaughter species. VADOFT is a onedimensional finiteelement code that solves Richards' equation for flow in the unsaturated zone. The user may make use of constitutive relationships between pressure, water content, and hydraulic conductivity to solve the flow equations. VADOFT may also simulate the fate of two parent and two daughter products. The PRZM and VADOFT codes are linked together with the aid of a flexible execution supervisor that allows the user to build loading models that are tailored to sitespecific situations. In order to perform probabilitybased exposure assessments, the code is also equipped with a Monte Carlo preprocessor and postprocessor.
PRZM3 Capability
PRZM has the capability to simulate multiple zones. This allows PRZM and VADOFT to combine different root zone and vadose zone characteristics into a single simulation. Zones can be visualized as multiple land segments joined together in a horizontal manner. There are three reasons a user may choose for implementing multiple zones:
 to simulate heterogeneous PRZM root zones with a homogeneous vadose zone
 to simulate a homogeneous root zone with heterogeneous vadose zones
 to simulate multiple homogeneous root zones with multiple homogeneous vadose zones
Another added feature is the ability to simulate as many as three chemicals simultaneously as separate compounds or as a parentdaughter relationship. This gives the user the option to observe the effects of multiple chemicals without making additional runs or the ability to enter a mass transformation factor from a parent chemical to one or two daughter products.
Predictions are made on a daily basis. Output can be summarized for a daily, monthly, or annual period. Daily time series values of various fluxes or storages can be written to sequential files during program execution for subsequent analysis.
PRZM Limitations
Hydrologic and hydraulic computations are still performed in PRZM on a daily time step even though, for some of the processes involved (evaporation, runoff, erosion), finer time steps might be used to ensure greater accuracy and realism. For instance, simulation of erosion by runoff depends upon the peak runoff rate which is in turn dependent upon the time base of the runoff hydrograph. This depends to some extent upon the duration of the precipitation event. PRZM retains its daily time step primarily due to the relative availability of daily versus shorter time step meteorological data. This limitation has been mitigated, in part, by enhanced parameter guidance.
In the previous version of PRZM, the soil hydraulics were simple  all drainage to field capacity water content was assumed to occur within one day. (An option to make drainage time dependent also was included, but there is not much evidence to suggest that it was utilized by model users to any great extent.) This had oneday drainage assumption the effect, especially in deeper soils, of inducing a greaterthananticipated movement of chemical through the profile. While this representation of soil hydraulics has been retained in PRZM, the user has the option of coupling PRZM to VADOFT. PRZM is then used to represent the root zone, while VADOFT, with a more rigorous representation of unsaturated flow, is used to simulate the thicker vadose zone. For short distances from the soil surface to the water table, PRZM can be used to represent the entire vadose zone without invoking the use of VADOFT as long as no layers that would restrict drainage are present.
PRZM simulates only advective, downward movement of water and does not account for diffusive movement due to soil water gradients. This means that PRZM is unable to simulate the upward movement of water in response to gradients induced by evapotranspiration. This process has been identified as an important one for simulating the effects of volitization. However, the process would seem less likely to impact the movement of chemicals with high vapor pressures. For these chemicals, vapor diffusion would be a major process for renewing the chemical concentration in the surface soil.
The final limitation is the use of fieldaveraged water and chemical transport parameters to represent spatiallyheterogeneous soils. Several researchers have shown that this approach produces slower breakthrough times than are observed using stochastic approaches. This concern has been addressed by adding the capability to run PRZM3 in a Monte Carlo framework. Thus, distributional, rather than fieldaveraged, values can be utilized as inputs that will produce distributional outputs of the relevant variables (e.g., flux to the water table).
VADOFT Overview
VADOFT is a finiteelement code for simulating moisture movement and solute transport in the vadose zone. It is the second part of the twocomponent PRZM3 model for predicting the movement of pesticides within and below the plant root zone and assessing subsequent groundwater contamination. The VADOFT code simulates onedimensional, singlephase moisture and solute transport in unconfined, variablysaturated porous media. Transport processes include hydrodynamic dispersion, advection, linear equilibrium sorption, and firstorder decay. The code predicts the infiltration or recharge rate and solute mass flux entering the saturated zone.
The code, which employs the Galerkin finiteelement technique to approximate the governing equations for flow and transport, allows for a wide range of nonlinear flow conditions. Boundary conditions of the variablysaturated flow problems may be specified in terms of prescribed pressure head or prescribed volumetric water flux per unit area. Boundary conditions of the solute transport problem may be specified in terms of prescribed concentration or prescribed solute mass flux per unit area. All boundary conditions may be time dependent. An important feature of the algorithm is the use of constitutive relationships for soil water characteristic curves based on soil texture.
VADOFT Limitations
Major assumptions of the flow model are that the flow of the fluid phase is onedimensional, isothermal and governed by Darcy's law, and that the fluid is slightly compressible and homogeneous. Hysteresis effects in the constitutive relationships of relative permeability versus water saturation, and water saturation versus capillary pressure head, are assumed to be negligible.
Major assumptions of the solute transport model are that advection and dispersion are one dimensional and that fluid properties are independent of contaminant concentrations. Diffusive/dispersive transport in the porousmedium system is governed by Fick's law. The hydrodynamic dispersion coefficient is defined as the sum of the coefficients of mechanical dispersion and molecular diffusion. Adsorption and decay of the solute is described by a linear equilibrium isotherm and a lumped firstorder decay constant. Parent/daughter chemical relationships may be simulated.
The code handles only singlephase flow (i.e., water) and ignores the presence of a second phase, i.e., air. The code does not take into account sorption nonlinearity or kinetic sorption effects that, in some instances, can be important. The code considers only singleporosity (granular) soil media. It does not simulate flow or transport in fractured porous media or structured soils.
Monte Carlo Overview
Monte Carlo performs all the functions necessary to execute a Monte Carlo simulation. It reads special data for parameters to be varied (e.g., distribution types and moments) and output variables to be observed; generates random numbers, correlates them and performs transformations; exchanges these generated values for PRZM3 parameters; performs statistical analysis on the output variables; and writes out statistical summaries for the output variables.
The MCARLO module makes use of an input and output file. Many of the parameters entered in the MCARLO input file once designated as constants will be used in lieu of that same parameter value entered in the standard input file.
A small number of input variables may be changed at random by invoking the Monte Carlo routines. It is not difficult to add additional variables, however.
PRZM3 can be run in a Monte Carlo mode so that probabilistic estimates of pesticide loadings to the saturated zone from the source area can be made. The input preprocessor allows the user to select distributions for key parameters from a variety of distributions:
 The Johnson family (which includes the normal and lognormal)
 Uniform
 Exponential
 Empirical
If the user selects distributions from the Johnson family, he or she may also specify correlations between the input parameters. The Monte Carlo processor reads the standard deterministic input data sets for each model and reads a Monte Carlo input file that specifies which parameters are to be allowed to vary, their distributions, the distribution parameters, and correlation matrix. The model then executes a prespecified number of runs.
The output processor is capable of preparing statistics of the specified output variables including mean, maximum values and quantiles of the output distribution. The output processor also can tabulate cumulative frequency histograms of the output variables and send them to a line printer for plotting.
PRZM3 includes the source code, executable version, user's manual, and technical support.
PRZM3 Requirements: PC 486 with 2 MB RAM and math coprocessor
