SURF 附录

附录 A：透射率的谐波平均

Natural porous medium systems exhibit a wide range of heterogeneities which must be accounted for in groundwater flow simulations.  The heterogeneities of material properties are provided to a finite difference model as nodal values for the individual grid blocks (cells).  The original MODFLOW Block-Centered Flow Packages BCF1 and BCF2 use a harmonic mean of the effective transmissivities of adjacent blocks to calculate the interblock transmissivity.  For unconfined systems, the effective transmissivity of a block is computed as the effective hydraulic conductivity multiplied by the saturated block thickness.
Harmonic averaging of interblock transmissivities can cause misrepresentations in an aquifer system, as illustrated in one extreme example depicted in Figure A.1.  Consider A and B to be adjacent grid blocks in an unconfined aquifer being flooded from the left. The situation at a given instant is such that grid block A has an average water-table elevation of h a which is above the bottom of the layer, and block B has an average water-table elevation of h b, below the bottom of the layer.  The transmissivity of grid-block A is some positive value, since h a is above the bottom.  The transmissivity of grid-block B is zero, since its saturated thickness is zero.  The Harmonic mean transmissivity is therefore computed as zero, and grid-block B never saturates, contrary to what is physically expected.
A more appropriate and physically consistent procedure is to use the Harmonic mean of the saturated hydraulic conductivities (or fully saturated transmissivities), with a midpoint (or upstream) weighting on the relative permeabilities (which are related to the saturated grid-block thickness by the pseudo-soil functions for our variably saturated scheme implementation).  It is noted (Romeu and Noetinger, 1995) that the Harmonic mean of block conductivities leads to an underestimation of the equivalent conductivity, and refined grids may be used to overcome this problem (in addition to inaccuracies associated with grid-averaging of the results).  However, the harmonic averaging of head-dependent transmissivities leads to a further underestimation of aquifer transmissive property, as is evidenced by the larger drawdowns near wells predicted by the previous package of MODFLOW (see Problems 1 and 2 of the BCF4).  Further, this problem may not be overcome by grid refinement.

附录 B：无承压含水层稳态抽水模拟的输入数据

The input data sets used for the steady-state simulation of an unconfined aquifer system are listed in this Appendix.  Two simulations were performed using:  Case(a) the original MODFLOW option and Case(b) the variably-saturated flow option of the BCF4 Package.  The problem description is presented in Section 2.4.1 of the BCF4 Package. The input data files are listed in the following order:

附录 C：用于从无承压含水层抽水的全 3D 模拟的输入数据

The input data used for the fully 3-D simulation of withdrawal from an unconfined aquifer are presented in this Appendix.  The problem description is presented in Section 3.4.1 of the FWL4 Package.  The input data files are listed in the following order:

        1.    The Basic Flow (BAS) Package Input
        2.    The Block-Centered Flow (BCF4) Package Input
        3.    The Fractured-Well (FWL4) Package Input
        4.    The Preconditioned Conjugate Gradient (PCG4) Package Input
        5.    The Output Control (OC) Package Input

附录 D：方形路堤稳态渗流模拟的输入数据

The input data used for the simulation of steady seepage through a square embankment are listed in this Appendix.  The problem description is presented in Section 4.3.2 of the RSF4 Package.  The input data files are listed in the following order:

1.    The Basic Flow (BAS) Package Input
2.    The Block-Centered Flow (BCF4) Package Input
3.    The Recharge-Seepage Face Boundary (RSF4) Package Input
4.    The Strongly Implicit Procedure (SIP) Package Input

附录 E：使用 ATO4 程序包对无承压含水层中的抽水和采收进行瞬态模拟的输入数据

The input data used for the transient simulation of pumping from an unconfined aquifer system using the Adaptive Time-Stepping and Output Control (ATO4) Package are listed in this Appendix.  The problem description is presented in Section 5.3 of the ATO4 Package.  The input files are listed in the following order:

1.    The Basic Flow (BAS) Package Input
2.    The Block-Centered Flow (BCF4) Package Input
3.    The Well (WEL) Package Input
4.    The Recharge (RCH) Package Input
5.    The Strongly Implicit Procedure (SIP) Package Input
6.    The Adaptive Time-Stepping and Output Control (ATO4) Package Input.

附录 F：MODFLOW-SURFACT/MODHMS 的 MODPATH 文档

   F.1.0 INTRODUCTION     

MODPATH is a particle tracking post-processing program designed to work with MODFLOW. MODPATH is documented in User's Guide for MODPATH/MODPATH-PLOT Version 3 (Pollock, 1994), which provides the particle tracking methodology, details of input and output files, time concepts, and examples. The document is available online at:
http://water.usgs.gov/software/code/ground_water/modpath/doc/ofr94464.pdf 
The version of MODPATH documented here (henceforth referred to as MODPATH-SURFACT)  is modified from MODPATH Version 3 (Pollock, 1994) to make it compatible with MODFLOW-SURFACT.  This documentation for MODPATH-SURFACT, is consistent with the original MODPATH documentation.  Additional input variables required by this version of MODPATH-SURFACT are shown in a shaded box.
 MODPATH-SURFACT requires these additional variables because of the following enhancements made to MODFLOW-SURFACT. 
1. MODFLOW-SURFACT uses the BCF4 package that provides a rigorous treatment of unconfined flow using a variably saturated formulation with pseudo-soil water retention functions to enhance convergence for desaturation/resaturation conditions.  The original MODFLOW code converts a desaturated variable-head cell into a no-flow cell.
2. The BCF4 package also includes real soil options to solve variably saturated flow situations in a more rigorous fashion by providing moisture retention and permeability characteristics of the unsaturated layers. 
3. The adaptive time-stepping and output control (ATO) package used by MODFLOW-SURFACT uses a time-stepping scheme with automatic generation and control of time steps that are not known aprori.  In the MODFLOW code, the time domain (for transient flow simulations) is discretized, with predetermined time-step values.  
4. MODFLOW-SURFACT uses a fracture-well (FWL4 and FWL5) packages that simulate a wellbore as a high conductivity fracture tube.  The FWL4 and FWL5 packages truly emulate a multilayer well and automatically apportion the net well withdrawal to well nodes.  Also, with FWL4 and FWL5 packages, the total well withdrawal is automatically adjusted when the water level in the well reaches the well bottom. 
Before MODPATH-SURFACT can be run, MODFLOW-SURFACT must be run with the following settings.
1. Flag MPHFLAG in Basic Flow (BAS) package should be equal to 1. If MPHFLAG =1, then a binary file *.mph is generated by MODFLOW-SURFACT. This MPH file is used by MODPATH-SURFACT to determine time step information. As described above, MODFLOW-SURFACT uses adaptive time-stepping (ATO) package which dynamically determines time stepping depending on the anticipated non-linearities of the system for a given calculation. 
2. All cell-by-cell flow components should be recorded in a single, common file (CBB file). Use the same unit number for cell-by-cell outputs from all packages(IBCFCB in BCF file, IFWLCB in FWL file, IDRNCB in DRN file, IRIVCB in RIV file, IGHBCB in GHB file, IRCHCB in RCH file and so on). 

3. The cell-by-cell flow information should be written to the CBB file at every time step during the simulation. This is achieved by setting appropriate flags in the Adaptive Time-stepping and Output Control (ATO) package (Flags ICBCFL=1, NPRTS=0 and NPSTP=1), and by providing a unit number in Block Centered Flow (BCF) package (IBCFCB) as well as all other boundary condition packages (FWL, GHB, RIV, RCH, EVT etc.) used in the simulation.  

F.2.0 MODPATH FILES

MODPATH obtains data from a combination of file and interactive keyboard input. Files that contain basic information about the flow system, such as geometry, boundary conditions, and cell-by-cell budget terms, are referred to as Flow System Files. Interactive keyboard input is used primarily to supply information about modeling options that change frequently from run to run. Keyboard input is recorded in a special data file called a Response File. Output from MODPATH consists of a series of files containing information about particle coordinates and time of travel. Detailed descriptions about particle tracking methodology, time concepts, input and output files, different options for running MODPATH are presented in MODPATH Version 3 (Pollock, 1994).
MODPATH always requires the following set of data files:

2.1. Flow System Files - MODFLOW-SURFACT and MODPATH data files that contain information about the physical dimensions of the ground-water system, boundary conditions, hydrologic properties, cell-by-cell budget output, and head.  MODPATH-SURFACT requires the following flow system files:

F.3.0 EXAMPLES

        Three sample problems are presented to illustrate application of MODPATH-SURFACT. All the problems are based on the same hypothetical flow system shown in Figure 1. There are 9 fracture wells performing extraction pumping at a rate of 50,000 ft3/day/well. These wells extract water from all 5 layers. All the layers use LAYCON = 43 option of the BCF4 package used by MODFLOW-SURFACT. Recharge occurs at the western portion of the model at a rate of 0.1 ft/day as shown in Figure 1. 
F.3.1 Example 1
In this example the system is at steady-state. MODFLOW-SURFACT was used to simulate flow with pseudo-soil water retention functions. This example demonstrates the use of LAYCON option 43 in BCF4 package, and the use of fracture wells via the FWL4 package when using MODPATH-SURFACT.
The resulting head distribution in Layer 3 is shown in Figure 2. Particle tracking was performed using this version of MODPATH. Particles were introduced in layer 1 of the model. Figure 2 also shows particle pathlines in plan view as well as the cross-sectional view. 
F.3.2 Example 2
Example 2 is a transient flow simulation performed on the problem described in Example 1. The simulation was performed for a 100 day period. This example demonstrates how to use MODPATH when a MODFLOW-SURFACT simulation uses the ATO package. 
Results for this simulation are presented in Figure 3. This figure presents head distribution in layer 3 as well as particle pathlines. These particles were introduced in layer 1 of the model. 
F.3.3 Example 3
This example is an extension of Example 2. Example 3 uses real-soil moisture retention functions for a transient MODFLOW-SURFACT simulation. This illustrates the use of MODPATH with MODFLOW-SURFACT when REALSL option is used in the BCF4 package. 

Results of this simulation are presented in Figure 4. 

附录 G：将 ET 时间序列与蒸散量 (EVT) 包或分段蒸散量 (ET2) 包一起使用

G.1   GENERAL

ETmax, the maximum rate of evapotranspiration, is taken to be constant within a stress period of MODFLOW.  For rapid changes in ETmax, it would then become tedious to break up a short time period into several stress periods during a simulation.  An option has therefore been provided, to vary ETmax within a simulation, independent of the stress periods.  The transient ETmax  is applied zonally, and an index, MXZETS, is read by the EVT or ET2 packages indicating the maximum number of zones on which transient ETmax is applied.  If MXZETS is zero, the transient ETmax package is not invoked by the simulation.  However, if transient ETmax exists, an additional integer array is read by the EVT or ET2 packages, identifying the areal location of each zone.  A zonal value of zero
indicates that no transient ETmax zone occurs over that location, and the default value of ETmax is applied to that location, as read in for the stress period in the EVTR or ETSR arrays.  At locations where the value is not zero, the EVTR or ETSR array is updated with
the appropriate value for that zone, from the ETmax time series data which is supplied in a separate file.  This file, with default extension .ETS, contains the ETmax time series tabulated for all zones.  Each row of the file contains the starting time (TSTART), the ending time (TEND), a multiplying factor (for unit conversions), followed by one column of ETmax rates for each of the zones.  If a subsequent TSTART value does not match with a TEND value from the previous row, zero ETmax is applied to each of the zones in the interim period.  Thus, long periods without ET need not be included as a separate row in the .ETS file.
The time periods between events, or within an event need not be fixed, as this module is designed for use with adaptive time stepping schemes.  The adaptive time marching scheme ensures that each TSTART and TEND is hit exactly by the simulation, to ensure that the temporal changes are accurately represented.  Within a period or between periods, the adaptive time stepping schemes try to maximize computational efficiency. 
Details on the EVT and ET2 packages are available in the respective USGS reports documenting these packages.  Note that the ET2 package is denoted as ETS1 in the original MODFLOW documentation (Banta, 2000).  The current documentation describes extension of these packages to include time-series input of ETmax.

G.3.   EVAPOTRANSPIRATION SEGMENTS (ET2) PACKAGE

G.4 瞬态势蒸发时间序列文件

附录 H：带有阶跃函数选项的时变流量和头部边界包

H.1 GENERAL

The Time-Variant Flow and Head Boundary Package was developed at the USGS (Leake and Lilly, 1997) to allow prescribed flux and prescribed head cells to take on different values for each time step.  It provides a means of simulating flux variations and head variations with time that can be provided in a tabular fashion.  This package is adapted here to extend the time-varying prescribed flux and prescribed heads to include step-functions between tabulated time values as an additional option to interpolation.  
Details of the FHB package are provided in the USGS documentation of the FHB package.  The step function implementation of this package is done via an input flag ITHTFHB.  When this flag is set to zero, the original MODFLOW implementation fo this package is used with interpolation of head or flux values from the tabular input.  Note that adaptive time-stepping may be used with this package and the computational time values need not honor the tabulated time values of the FHB package.  This is also the case for use of MODFLOW’s time stepping scheme.  However, if adaptive time stepping is used and if the flag ITHTFHB is set to 1 then adaptive time-stepping will honor the time values provided in the tabular time array for FHB input.  Head and flux values for intermediate times are still interpolated from the tabular input values.  If, however, the flag ITHTFHB is set to 2, the head and flux values are implement as step functions instead of interpolation.  Note that this package has also been extended for the surface-water domains of MODHMS.  

附录 I：PCG5 求解器的附录

I.1 GENERAL

The PCG5 package is a proprietary solver that may be included with MODFLOW-SURFACT.  This package implements the MATIS matrix solver developed by Dr. Peter Forsyth at the University of Waterloo.  The PCG schemes are the same as those discussed in Chapter 6 of Volume I of this document.  However, the PCG5 solver provides for additional levels of fill during ILU decomposition (the PCG4 solver only accommodates a fill level of zero).  It further includes red-black ordering schemes and a variety of acceleration techniques (only the Orthomin acceleration scheme is currently released with MODFLOW-SURFACT) to make this package more robust and efficient than the PCG4 package.  Experience has shown this package to be up to 20 times faster than the PCG4 package in solving typical flow and transport problems.   

I.2 ITERATIVE SPARSE MATRIX SOLVER (PCG5) PACKAGE INPUT