release.txt MODFLOW-2000 Release Notes This file describes changes introduced into MODFLOW-2000 with each official release; these changes may substantially affect users. Changes to the post-processing programs RESAN-2000, YCINT-2000, and BEALE-2000 also are described. In this file, packages and processes are referred to by their 3- or 4-character abbreviation. For information and references related to these packages and processes, see mf2k.txt ("Summary of MODFLOW-2000"). NOTE: Any use of trade, product or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government. o Version 1.19.01 03/25/2010: This version fixes a minor bug in the GMG Package. Also, in the Unix distribution, the openspec.inc file was changed to correct a minor problem introduced at ver. 1.19.00. o Version 1.19.00 03/15/2010: This version includes the addition of a new package (MNW2), minor changes, and bug fixes. The revised Multi-Node Well (MNW2) Package was added. The MNW2 Package is documented in USGS Techniques and Methods 6-A30 by Konikow and others. The SFR2 Package was updated. This update is described in the readme_SFR2.pdf file, which is located in the doc\updates directory. The LAK3 Package was updated. This update is described in the readme_LAK3.pdf file, which is located in the doc\updates directory. The GAGE Package was updated. This update is described in the Gage_input_instructions.pdf file, which is located in the doc\updates directory. The ETS Package was updated to add support for NETSOP=3. When NETSOP is specified as 3, ET is simulated from the uppermost active cell in each stack of cells. The Link-MT3DMS (LMT6) Package was updated to version 6.4. The update is described in doc/updates/lmt6.txt. The directory and file organization of the MODFLOW-2000 distribution has changed slightly, relative to previous versions. Now the top-level directory contains the files cases.txt, Mf2k.txt, Readme.txt, and release.txt. The doc directory contains PDF files of the two original documentation reports for MODFLOW-2000 (USGS Open-File Reports 00-92 and 00-184, and corresponding errata files) and an "updates" directory. All documentation files that supplement the original two reports or other documentation are provided in the updates directory. The maximum number of zone arrays and multiplier arrays is changed from 500 to 1000. PCG2 now writes locations of isolated cells that are converted to no-flow. Output related to the table of STR-Package streamflow observations was modified to explain the situation that applies when leakage to aquifer is limited by flow into the reach. Bug fix, DRT: Program failed when LayR of DRT input was specified as 0, which is used to indicate a DRT cell with no associated return cell. Bug fix, DRT: Binary cell-by-cell flow file would be generated incorrectly when a recipient cell is inactive. Bug fix, LPF: Output identifying effect of a HANI parameter was incorrect. Bug fix, HUF: Interpolated head in hydrogeologic units were incorrectly calculated in some situations. o Version 1.18.01 06/20/2008: This version includes minor changes and bug fixes. An "Examples" section has been added to Time-varying-parameters.pdf in the "doc" directory. Two example input files illustrate the use of time-varying list-type and array-type parameters. Changes were made to HUF2 to check for input errors related to parameter type SYTP. HUF2, LPF, and BCF were revised to change handling of the situation where input specifies no nonzero conductances to a cell specified as active. In previous versions, head at these cells was assigned a value of 888.88. Now the head is assigned the value input for HNOFLO. In HUF2, a similar situation arises when input defines cell thickness less than or equal to zero; in this case, head now is assigned as HNOFLO (rather than 888.88) and a warning is written to the List or Global output file. Bug fix: HUF2 output related to errors encountered in calculation of vertical conductance incorrectly identified row and column indices. Bug fix: Multiple SY parameters and clusters were not handled correctly by HUF2. Previously, contributions to the coefficient matrix from multiple SY parameters/clusters were not summed; only the last cluster processed by the storage conversion routine made a contribution. This resulted in erroneous handling of unconfined storage when more than one SY parameter or cluster was applied at a given cell. Bug fix: A bug was fixed to allow proper calculation of SC1. When SC1 is computed for a cell, the contribution from a given hydrogeologic unit (HGU) depends on the thickness of the HGU within the cell. Previously, the method of computing SC1 allowed the portion an HGU that is above the topmost active cell or below the bottommost active cell to be included in the HGU thickness within those cells. Bug fix: A divide-by-zero error occurred when the SUB package was used in a model for which PERLEN was specified as 0.0 for a steady-state stress period. o Version 1.18.00 08/23/2007: This version introduces the Subsidence for Water-Table Aquifers (SWT) Package. Details on theory and use of the new package can be found in U.S. Geological Survey Techniques and Methods 6-A23, by Leake and Galloway. The SWT Package is an alternative to the existing SUB and IBS packages. This version fixes a bug that was introduced at version 1.17.04 that affected the use of SFR2 with BCF and HUF. Also fixed is a bug that affected compilation of the MNW1 Package. o Version 1.17.04 08/03/2007: This version includes minor changes and bug fixes. The maximum number of parameters that can be defined in MODFLOW-2000 was increased from 500 to 999. Partial support for use of SFR2 with the HUF Package has been added. Simulation of unsaturated flow below streams by SFR2 in combination with use of the HUF Package is not supported. Other capabilities of SFR2 may be used with HUF. In the GMG Package, the format used for printing of damping parameter to unit IUNITMHC was changed to accomodate a larger range of numeric values. Bug fixed: SFR2 did not work correctly when some segments were defined using parameters and others were defined without using parameters. Bug fixed: Program incorrectly stopped with the message "PARAMETER ESTIMATION DID NOT CONVERGE IN THE ALLOTTED NUMBER OF ITERATIONS" when MAX-ITER of the Parameter-Estimation Process input file was specified as 0 and parameter sensitivities were exceedingly small. Bug fixed: Comment lines in data input files, which are supposed to allow 200 characters, were truncated at 79 characters. In previous versions of the post-processing program RESAN-2000, the table of "INFLUENTIAL OBSERVATIONS WITH DFBETA > CRITICAL VALUE" in the .#rs output file was incorrectly generated when the number of parameters exceeded 20. Now, 999 parameters are supported. The current version of RESAN-2000 is '1.5 07/25/2007'. o Version 1.17.02 02/14/2007: This version includes fixes to bugs and minor changes in the following packages: LAK - Several arrays were converted from single precision to double precision. GAG - Two arrays that are common to GAG and LAK were converted from single precision to double precision. SFR - A bug that caused non-parameter data for stress period 2 and later stress periods not to be read correctly when unsaturated flow is not simulated was fixed. RCH - Interpretation of the IRCH array was modified to add support for 0 (zero) values in the array. A value of 0 is now interpreted to mean no recharge is to be simulated at this location. EVT - Interpretation of the IEVT array was modified to add support for 0 (zero) values in the array. A value of 0 is now interpreted to mean no evapotranspiration is to be simulated at this location. ETS - Interpretation of the IETS array was modified to add support for 0 (zero) values in the array. A value of 0 is now interpreted to mean no evapotranspiration is to be simulated at this location. HUF - Output identifying the zone array associated with each hydrogeologic unit now uses text "ZONE ARRAY" instead of "ZONE". o Version 1.17.01 09/22/2006: This version includes fixes to bugs in the following packages: SFR - Previous version (1.17.00) wrote incorrect flow values to the compact cell-by-cell budget file when a cell contains more than one SFR reach. - Non-parameter data for stress period 1 were not read correctly. FHB - A bug affected assignment of head for the special cases of having only one time at which heads are specified and having a steady-state stress period. - A bug caused an invalid entry to be written to a compact cell-by-cell budget file. LAK - Bug caused program to crash when THETA was specified as a negative, floating-point number. o Version 1.17.00 07/25/2006: Note to users of the Hydraulic-Conductivity Depth-Dependence (KDEP) capability of the HUF Package: A program error was found in the KDEP capability beginning with its first release with version 1.12 of MODFLOW-2000 (Mf2k) and continuing through version 1.15 of the code. The error was fixed beginning with version 1.16. The error has been found to cause problems in some (but not all) models that specified input variable IFKDEP equal to 0. Variable IFKDEP is a flag indicating the source of the array specifying reference-surface elevations used for the depth-dependent hydraulic- conductivity calculations. When IFKDEP = 0, the TOP array specified in the Discretization Package Input File is used; when IFKDEP = 1, the array is specified in the KDEP input file (variable RS). In some circumstances, specification of IFKDEP = 0 led to mis-allocation of elements in the Mf2k X array, which led to errors in some of the budget calculations. Models that used IFKDEP = 1 should not be affected by this error. If KDEP users of Mf2k versions 1.12 - 1.15 are uncertain as to whether or not their results are affected by this error, they are encouraged to run their data sets with Mf2k versions 1.16 and higher and compare the results calculated by version 1.16 with those of the earlier version. The LAK3, SFR2, and GAG5 Packages have been updated to support new capabilities. The new capabilities and other changes are documented in files "Read_me.LAK3.v1.17.txt" and "Read_me.SFR2.v1.17.txt" in the doc directory. Changes that should be made to USGS Techniques and Methods 6-A13 are documented in "tma6a13_SFR2revision_history.pdf", also in the doc directory. The GMG Package has been updated to support a new adaptive damping option, which may be useful in obtaining solutions to highly nonlinear problems. The new damping option is documented in "GMG_Damping.pdf" and "Banta_damping_paper.pdf" in the doc directory. The SUB Package was modified to allow comment lines to be inserted into the SUB input file anywhere except where arrays are to be read by one of the array-reader subroutines (U2DREL or U2DINT). A minor bug affecting output of the ADV Package has been fixed. Several bugs in the MNW Package have been fixed. A bug in the DRT Package that caused a budget error when a drain-return cell is inactive has been fixed. Now, if the return cell is inactive (IBOUND=0) or constant-head (IBOUND<0), no return flow takes place, and all drain flow exits the model. o Version 1.16.00 03/28/2006: This version includes updates to three packages, and several bug fixes. Version 2 of the SFR Package replaces SFR1 in this version. SFR2 adds support for simulation of unsaturated flow beneath streams. Files prepared according to SFR1 instructions will work as they did with SFR1. The GAGE and LAK3 Packages were revised to work with SFR2. SFR2 is documented in USGS report TM6-A13. The ADV2 Package has been modified to improve compatibility with MODPATH. The changes are documented in MODPATH-ADV_compatibility_3-2006.pdf, which is included in the doc directory of the MODFLOW-2000 distribution. Values in some tables produced by the STR Package have been modified to be written with additional decimal places. This version includes fixes to the following bugs: Calculation of sensitivities to EVT parameters failed when NEVTOP was specified as 3. "ALLOCATION OF ARRAYS RX, IR, RZ FAILED" when batch mode was used. Compact budget file was incorrectly written when drain-return cells specified in the DRT Package are dry. In LAK3, in cases in which a lake that had gone dry was underlain by a confined type of layer, an undefined variable caused a run-time error. Several bugs in HUF2. Array-bound access error was caused when there were LPF parameters of type HK or HANI not listed in the SEN file. o Version 1.15.01 04/05/2005: This version includes fixes to bugs in the following packages and post-processing program: SUB - Minor bug fix SFR - Fix to bug that prevents time-varying SFR parameters from working correctly - Fixed bugs that causes incorrect handling of no active cell at reach location - Fixes to other minor problems, which should improve precision and execute faster GMG - Fixes to problems in FORMAT statements LPF - Changed to skip calculation of sensitivities for a confined layer when PID equals 'SY'. - Fixed bug that caused incorrect deactivation of cells that have nonzero vertical K and all horizontal Ks zero BAS - Changed to flag parameters with zero sensitivities. MNW - Bug fix (divide by zero when a cell goes dry). - Converted MNW to use all double precision (except arrays from other packages). YCINT - Minor bug fix o Version 1.15.00 8/06/2004: A new solver is introduced with this release. The Geometric Multi-Grid (GMG) Solver Package is documented in USGS Open-File Report 2004-1261 by Wilson and Naff. Minor bugs in the RES and HUF Packages were fixed. This release also includes the following modifications to the SFR1 Package: The solution scheme in GWF1SFR1FM has been changed to a mixed bisection-newton method. Convergence using this method to estimate stream depth at the midpoint is faster then using the bisection method alone. For the solution scheme to work correctly, the DLEAK term used to solve for stream depth at the midpoint should be equal to or less than the closure criteria used in MODFLOW, although some problems may solve when DLEAK is greater than the closure criteria. Some slight modifications were also made to GWF1SFR1BD to properly account for overland runoff, precipitation, and evapotranspiration for all ICALC values. Also, when there is no flow into the stream reach, as surface inflow (including runoff); as ground-water discharge to the stream; and evapotranspiration exceeds precipitation, then evapotranspiration is set to zero. One additional change was made by Suzanne Tillery of New Mexico State University. She changed the program in GWF1SFR1FM and GWF1SFR1BD to allow for multiple diversions at the end of a stream segment. For multiple diversions from a single segment, each diversion must be a new segment and its flow is diverted from the end of the same segment in order of the segment numbering. This means that a diversion placed first in the data input has its flow subtracted first. o Version 1.14.00 7/01/2004: The new Stream-Flow Routing (SFR1) Package is introduced with this release. The new package encompasses the capabilities of the STR Package that are part of the Ground Water Flow Process and adds enhancements. The SFR1 Package is documented in USGS Open-File Report 2004-1042. The SFR1 Package is integrated with the MODFLOW Ground Water Transport Process to allow simulation of solute transport in streams. The report also describes enhancements to the GAGE capability to work with the SFR Package. The LAK3 Package is updated to work with the new SFR1 Package. The maximum number of layers has been increased from 200 to 999. Simulated equivalents are now calculated using double-precision variables; differences are expected to be small. If available system memory is insufficient to allocate the main arrays, a message explaining the problem is printed and the program stops. Two minor bugs were fixed in the MNW Package. Errata sheets for USGS Open-File Reports 00-92 and 00-184 have been added to the doc directory. The names of the files are ofr-00-92-errata.pdf and ofr-00-184-errata.pdf. o Version 1.13.00 1/22/2004: This release includes revisions to the LAK3 Package, including changes that affect: (1) Simulation of mixed steady-state and transient stress periods; (2) Simulation of the top layer as confined; (3) Lake stage saved or printed in the head array, rather than HNOFLO; (4) Checking of consistency between lake cells and IBOUND; (5) Simulation of lake cells in the bottom layer; and (6) other minor output changes. Please see LakePackage.read_me.txt in the doc directory of the Modflow-2000 distribution for more details. Additional explanatory text is written to the Global output file when the simulation time is truncated by the Sensitivity Process, to the time step required for the last observation, to reduce execution time. The SUB Package was modified to work correctly when batch mode is used. A minor change was made to the PCG2 Package to correct a problem with damping when the polynomial preconditioner is used (NPCOND=2 in PCG2 input). Documentation of a previously undocumented feature that allows control of printing of a multiplier array that is a function of other multiplier arrays: An IPRN value can be specified after the last multipler-array name in item 4 of the MULT file. Valid IPRN values are listed on p. 87 of Open-File Report 00-92. Assign IPRN<0 to suppress printing. A bug that caused a problem reading the DRT Package input file in fixed format was corrected. Other bugs related to the use of DRT-Package parameters were fixed. A revised version of the HUF2 Package, which fixes bugs that affected calculation of flow in hydrogeologic units, replaces the previous version. o Version 1.12.01 10/03/2003: The version number (1.12.01) for this release uses a modified identification scheme. The addition of a subrelease number (".01" in this case) is used to identify releases that are made only to fix bugs. This release fixes the following bugs: In previous versions, incorrect final simulated-equivalent sensitivities were printed when LASTX > 0 in the PES input file and parameter estimation converged. Note that these simulated-equivalent sensitivities are never used in the regression. Previous versions had an error in the calculation of the weight matrix for correlated prior information. This bug would have affected regression results only when correlated prior information is used by entering a WTP array in item 9 of the PES input file. In previous versions, incorrect weighted residuals were printed for correlated flow observations when an observation is correlated with another observation that is omitted due to drying of cells included in the observation. The bug affected correlated flow observations for the DRN, RIV, GHB, STR, and DRT Packages. In version 1.12 simulated-equivalent sensitivities of constant-head flow observations to LPF-Package parameter types HK, VK, VANI, or VKCB were incorrectly calculated when a constant-head cell used as a flow observation location had an overlying or underlying active cell. In previous versions of BEALE2K, incorrect results were produced when correlated prior information was used. The new version of BEALE2K has the version identifier "1.2.01 10/03/2003". Numbers up to 30 characters in length can now be read from input files when the FREE option of the Basic Package is specified. The previous limit was 20 characters. o Version 1.12 09/08/2003: The new Subsidence and Aquifer-System Compaction (SUB) Package is introduced with this release. The new package encompasses all the capabilities of the Interbed Storage (IBS) Package and adds enhancements. This version of Modflow-2000 also includes the IBS Package, but the IBS Package is considered obsolescent, and future versions of Modflow-2000 may not support it. Concurrent use of the SUB and IBS Packages will result in an error message. The SUB Package is documented in USGS Open-File Report 03-233. Version 2 of the HUF Package replaces version 1. The new version of HUF includes the Layer Variable-Direction Horizontal Anisotropy (LVDA) capability documented in USGS Open-File Report 02-409, as well as three additional capabilities documented in USGS Open-File Report 03-347. The ADV Package has been updated to version 2.4. The new version adds support for porosity parameters, documented in ADV-porosity-parameters.pdf in the doc directory of the Modflow-2000 distribution. The use of the ADV Package is now restricted to simulations in which the first stress period is steady-state, and which include no other steady-state stress periods. Unfortunately, due to licensing restrictions, the USGS is no longer able to publicly distribute the Algebraic Multi-Grid (AMG) solver, on which the Link-AMG (LMG) Package relies. USGS users can download a version of Modflow-2000 that includes AMG from a restricted web site (please contact Ned Banta or Arlen Harbaugh for the URL). Users outside the USGS can request the code directly from Dr. Klaus Stuben of Fraunhofer-Institute for Algorithms and Scientific Computing (SCAI), Sankt Augustin, Germany, at amg1r5@scai.fraunhofer.de. A new optional setting in the OBS Process input file is available. If "ALLFILES" is specified following the ISCALS value, and if the value specified for OUTNAM is not "NONE", up to three additional output files are generated with the filename base defined by OUTNAM. The "underscore.txt" file in the doc directory describes the new files. A fifth column has been added to the _os output file, which was originally described in table 5 on page 28 of USGS Open-File Report 00-184. The fifth column lists the observation time, in model time units, from the beginning of the simulation. For prior information, -1.0 is written. The code modification for ADV assumes that all ADV observations are in the first stress period. Efficiency of the check for duplicate observation names was improved. The DATA AT HEAD LOCATIONS table was edited to improve clarity. The LMT6 Package (which provides linkage between Modflow-2000 and the MT3DMS multi-species mass transport model, see USGS Open-File Report 01-82) has been updated to support the HUF Package, version 2, and the MNW, ETS, and DRT Packages. Please see lmt6.txt in the doc directory for more information. Bug fixes: A bug that caused problems when the character is used as a delimiter in input files has been fixed. A bug that could incorrectly cause printing of an error message referring to incorrect specification of interblock transmissivity averaging method when calculating sensitivity for "HK" or "HANI" type parameters, even if averaging method is correctly specified as HARMONIC MEAN, has been fixed. A bug that caused inaccurate sensitivity results or program failure when calculating sensitivities in a situation where one or both cells adjacent to a horizontal flow barrier (of the HFB Package) are inactive (IBOUND=0) has been fixed. Fixed bug that caused calculated heads to be incorrect when OBS Process is active, SEN Process is inactive, LPF Package is active, and first parameter defined in LPF file is either of type VKCB or HK. Several bugs in the MNW Package were fixed. A minor bug in the DAF Package was fixed. In Resan2k, a bug in the calculation of Cook's D was fixed, and the table showing Cook's D was modified to eliminate colums VAR(E) and VAR(Y), and add column VAR(Y)/VAR(E). o Version 1.11 04/10/2003: This version includes two new packages. The DAF Package supports coupling of Modflow-2000 with the Diffusion Analogy Surface-Water Flow Model (DAFLOW), documented in Jobson, H.E. and Harbaugh, A.W. (1999, Modifications to the Diffusion Analogy Surface-Water Flow Model (DAFLOW) for coupling to the modular finite-difference ground-water flow model (MODFLOW), USGS Open-File Report 99-217, 107 p.). The MNW Package allows simulation of wells completed in intervals represented by multiple cells. The MNW Package is documented in Halford, K.J. and Hanson, R.T. (2002, User guide for the Drawdown-Limited, Multi-Node Well (MNW) Package for the U.S. Geological Survey's modular three-dimensional finite-difference ground-water flow model, versions MODFLOW-96 and MODFLOW-2000, USGS Open-File Report 02-293, 33 p.). A modification to the PCG2 Package changes the way the DAMP variable works. In previous versions of PCG2, DAMP was used to control oscillations that could occur within a single call to the solver routine, among inner iterations. In the current version of PCG2, DAMP does not affect inner iterations; instead, it affects outer iterations. This change likely will allow the solver to behave more predictably in solving highly nonlinear problems. Note that for a given model, DAMP may need to be set to a different value than for previous versions to achieve optimum results. In addition, this version of PCG2 includes changes that should make the solver less prone to errors that cause the program to abort, possibly generating diagnostic messages referring to a floating overflow exception in PCG2AP. A new Output Control option allows users to specify that drawdowns, which can be printed to the List file or saved in a separate output file, are to be calculated relative to model-calculated heads at specified time steps. In previous versions of Modflow and by default in the current version, all drawdowns are calculated relative to heads entered by the user in the STRT array read by the BAS Package. The new option allows the user, when using the word method of Output Control, to indicate one or more time steps, relative to which subsequent drawdowns are to be calculated. To use this option, the keyword "DDREFERENCE" is entered following the value entered for ITSOC of item 2 of the Output Control file (USGS Open-File Report 00-92, pp. 52-54) for a time step relative to which drawdown in later time steps is to be calculated. "DDREFERENCE" can be entered for as many time steps as desired. In each case, heads calculated for the time step for which "DDREFERENCE" is entered will serve as the basis relative to which drawdowns for subsequent time steps are to be calculated, until the next occurrence of "DDREFERENCE" is found. The following example of an Output Control file illustrates the use of the DDREFERENCE option: DRAWDOWN PRINT FORMAT 11 PERIOD 1 STEP 1 DDREFERENCE PRINT DRAWDOWN PERIOD 2 STEP 3 PRINT DRAWDOWN PERIOD 2 STEP 5 DDREFERENCE PRINT DRAWDOWN PERIOD 3 STEP 5 PRINT DRAWDOWN In this example, drawdowns printed (in the List file in format 20F5.4) for stress period 1 time step 1 are the differences between the heads entered by the user in the STRT array and heads calculated for stress period 1 time step 1. No output control is entered for time steps 1 and 2 of stress period 2, so no drawdown arrays are printed for these time steps. Drawdowns printed for stress period 2 time step 3 are the differences between heads calculated for the end of stress period 1 time step 1 and heads calculated for the end of stress period 2 time step 3. Drawdowns printed for stress period 2 time step 5 are the differences between heads calculated for the end of stress period 1 time step 1 and heads calculated for the end of stress period 2 time step 5. Drawdowns printed for stress period 3 time step 5 are the differences between heads calculated for the end of stress period 2 time step 5 and heads calculated for the end of stress period 3 time step 5. A new BASIC Package option, which allows progress to be monitored as the program solves the ground-water flow and sensitivity equations, has been added. To invoke this option, list the keyword "SHOWPROGRESS" as an option in item 1 of the BAS input file. When invoked, this option displays on the monitor the stress period, time step, and the equation currently being solved. Unfortunately, support for this option is processor-dependent; as a result, when this option is selected on some systems, lines showing progress will not be overwritten by subsequent lines. In these cases, use of this option will cause the display to scroll as the progress line is updated. Prior to the current version, duplicate observation names were accepted without notifying the user of the duplication. However, support in Modflow-2000 for duplicate observation names is obsolescent -- future versions of Modflow-2000 may require unique names. The current version accepts duplicate observation names, but if duplicate names are found, a warning concerning the duplication is written to the Global file. The maximum allowed numbers of zone and multiplier arrays were increased from 200 to 500. The effect of the IFHBSS variable in input to the FHB Package was modified slightly. Please see Readme_FHB.txt in the doc directory for details. The format of the "DATA AT HEAD LOCATIONS" table was modified slightly to match that of the "DATA FOR FLOWS" tables. Observations specified at the exact end of the simulation are no longer flagged as being after the end of the simulation. The table of residuals for parameters with correlated prior information no longer refers to "parameter sums". A rarely-encountered bug, identified by an error message referring to a floating division exception in SSEN1LPF1NL, was fixed. For the LMG Package, the AMG1R5 solver was updated to AMG1R6, which fixes a bug related to memory usage. For the HFB Package, minor output changes were made, and bugs related to calculation of sensitivities were fixed. Two bugs related to correlated flow observations were fixed. A minor bug in the NOPRINT capability was fixed. A minor bug related to the printing of DRN and DRT Package observations was fixed. If MAX-ITER of the PES-Process input is less than zero, the program now stops with an appropriate error message. Now, when a parameter value is changed because it is unreasonable (for example, a negative hydraulic conductivity) in the final parameter- estimation iteration, the reasonable value that takes its place is written to the Global and _b files and to the screen. In previous versions, the unreasonable value was written. A bug related to correlated prior information for log-transformed parameters was fixed. The program now stops with an error message if sensitivities are to be calculated for HK or HANI type parameters of the LPF Package if the transmissivity averaging method for a layer affected by the parameter is other than HARMONIC MEAN. For developers, all STOP statements have been replaced by calls to a new subroutine, USTOP. Richard Yager of USGS in Ithaca, NY, kindly supplied the following information related to the _rc and _rb files produced by RESAN-2000: The "_rc" file lists Cook's D, a measure of the influence of the observation on the set of estimated parameter values, along with the observation name and plot symbol. The "_rb" file lists the name and plot symbol of each observation along with the set of DFBETAS, a measure of the influence of the observation on each parameter value. The derivation of Cook's D is given in Cook and Weisburg (1982) and the derivation of DFBETAS is given in Belsley and others (1980). The application of both influence statistics to nonlinear regression of ground-water flow modeling is discussed by Yager (1998). Belsley, D.A., Kuh, E., and Welsch, R.E., 1980, Regression diagnostics: identifying influential data and sources of colinearity, John Wiley, New York. Cook, R.D., and Weisburg, S., 1982, Residuals and influence in regression, Monographs on Statistics and Applied Probability, v. 128, Chapman and Hall, New York. Yager, R.M., 1998, Detecting influential observations in nonlinear regression modeling of groundwater flow: Water Resources Research, v. 34, no. 7, p. 1623-1633. o Version 1.10 07/26/2002: This version incorporates fixes to bugs that showed up when compiling the source code distributed with version 1.9. It also fixes a bug related to a seldom-used part of the LMG Package involving a random-number generating algorithm. o Version 1.9 07/15/2002: In this version, the LAK package was modified to add compatibility with the LPF and HUF packages, and MODPATH. Output of lake information to the listing file is clarified, and cumulative lake budget reports are added. The default value of MXLKND (the maximum allowable number of lake/aquifer cell interfaces) is reduced. In the GAGE package, options are added for additional output for lake gages. The changes to the input are described in the file \doc\gage_input.pdf. The options are invoked in the gage input by specifying a negative value of UNIT. If a negative UNIT is specified, an additional parameter is read from Data Set 2 (after UNIT) called OUTTYPE. OUTTYPE is a code for the format of expanded listing desired in output file: 0 Use standard default listing of time, stage, volume, and concentration. 1 Default values plus time-step fluxes for lake and total lake conductance. 2 Default values plus changes in stage, volume, and concentration for lake. 3 All of the above. In the HUF Package, a problem in the calculation of horizontal- conductance sensitivities was fixed. The current version of HUF is "1.04, ERA 5/14/2002." Some problems related to omission of observations (from the sum of squared, weighted residuals objective function and other statistics and the underscore ("_") output files) were corrected. Now, these rules apply: 1. For a given set of parameter values, the statistics have the same values and "_" files include the same observations, regardless of whether the Sensitivity Process is active or inactive, whether a particular parameter is one for which sensitivities are being calculated, whether the Parameter-Estimation Process is active or inactive, the value of MAX-ITER, which parameter-estimation iteration is being performed, or whether or not parameter estimation converged. 2. When a simulated-equivalent head cannot be calculated, for example, when the observation cell is dry or one or more cells needed for a multilayer head observation is dry, that observation is flagged as omitted in the DATA AT HEAD LOCATIONS table, is not listed in the "_" files, and does not contribute to the statistics. 3. Regarding flow observations where the heads in all the cells included in an observation are below the bottom of the river or stream bed or below the drain elevation: These cases now conform to item 1 above. Flow observations of this type are never be omitted from either the statistics or the "_" files. 4. Regarding flow observations where all the cells included in an observation are inactive: Flow observations like this are flagged as omitted in the "DATA FOR FLOWS..." tables and omitted from statistics and the "_" files. The following bug was fixed: When the DE45 was used and IFREQ was specified as 3, external iterations were not being performed in solving the sensitivity equation. In the current version, external iterations are performed in solving both the ground-water flow equation and the sensitivity equation when ITMX > 1 and IFREQ = 3. o Version 1.8 05/01/2002: This version adds support for an option that can be used to suppress printing of cell lists in any or all of the RIV, GHB, DRN, DRT, HFB, WEL, and CHD packages. In input to the RIV, GHB, DRN, WEL, and CHD packages, the NOPRINT option is invoked by entering the keyword "NOPRINT" as an option on Item 2. In input to the DRT package, the NOPRINT option is invoked by entering "NOPRINT" as an option on Item 1. In input to the HFB package, the NOPRINT option is invoked by entering "NOPRINT" in Item 1 after the value entered for NHFBNP. In the LPF Package, the sensitivity of vertical conductance to head in the calculation of sensitivities for unconfined cells is now correctly calculated. In the HUF Package: a number of small issues were uncovered by John Doherty, and many thanks go to him for his careful scrutiny of the HUF code. Among these changes, geohydrologic units are now truncated if the bottom of the unit is in an inactive cell. A problem with the rewetting capability was fixed. The sensitivity of vertical conductance to head in the calculation of sensitivities for unconfined cells is now correctly calculated. The current version of HUF is 1.03. In the ADV Package: the current version (2.3) fixes a problem with threshold for stagnant velocities. Also fixed: a problem with scaled sensitivities in a confining unit. In the LMT Package: This version corrects an inconsistency between the Link-MT3DMS (LMT6) package and MT3DMS 4.0. This inconsistency causes an I/O error when the flow-transport link file saved by LMT6 is read by MT3DMS. The problem only occurs when the STR package is used in the flow simulation, and when the Extended header option is specified for saving the flow-transport link file. The current version of LMT is 6.1. The RESAN2K, BEALE2K, and YCINT2K post-processing programs were converted to use dynamic memory allocation. This means that the programs will not need to be recompiled to accommodate data sets that include large numbers of observations. This version also fixes the following bugs: In previous versions of MODFLOW-2000, when the OBS Process was active and the PES Process was inactive, one element (for the cell in layer 1, column 1, row 1) of one array (the affected array depended on which flow package was active) was overwritten with an invalid value. The bug did not affect parameter estimation, and it did not affect simulations in which that cell was specified as inactive. Thanks to Evan Anderman for identifying this bug. In previous versions of RESAN2K, observation names were not being ordered in the ORDERED, INDEPENDENT NORMAL DEVIATES table. The ordering problem also affected the _rd file, where the plot symbols also were not being ordered. Thanks to Claire Tiedeman for identifying this bug. This version fixes compiler- and operating-system-dependent problems related to writing of the _sc, _sd, _s1, and _rs files. Thanks to Terrence Conlon for identifying these problems. Also included in the 1.8 release is a revised str6.pdf file, which describes the implementation of the STR Package in MODFLOW-2000. The revisions describe limitations related to use of parameters in the STR Package. A distribution file prepared for the Unix operating system is available for MODFLOW-2000 for the first time. For the current version, the Unix distribution file is mf2k1_8.tar.Z. o Version 1.7 12/04/2001: Time-varying parameters are introduced with this version. Most packages that read input each stress period support time-varying parameters. These include both list-type packages such as RIV and STR and array-type packages such as RCH and EVT. Time-varying parameters are useful when the user desires to estimate one parameter but vary properties included in the parameter definition from one stress period to another. In the RIV Package for example, a parameter may be defined to control riverbed conductance at a group of model cells throughout a simulation, and multiple user-defined "instances" associated with that parameter may be used to specify a series of values for river stage for different stress periods in the simulation. See files Time-varying-parameters.pdf and str6.pdf in the doc directory for details. The start time, end time, and elapsed time required to execute MODFLOW-2000 are now written to the screen. These times also can be written to the global output file by specifying "PRINTTIME" as an option in Item 1 of the BAS Package input file. When the program is used in batch mode, these times are determined and written for each simulation, that is, for each name file listed in modflow.bf. This capability is expected to be particularly useful in evaluating performance of the program in different serial- and parallel-processing environments. Fixed problem where, when using parallel processing, files of type DATA, DATA(BINARY), DATAGLO, and DATAGLO(BINARY) may not be opened correctly for MPI processes other than the master process. This version uses the name file options described for version 1.6 below to determine the correct method for opening files of these types. To correctly generate output files of these types, use the "REPLACE" option. The "OLD" option may be used to designate input files. Output related to observations of flow at constant-head boundaries no longer refers (mistakenly) to the "Constant-Head Boundary Package." o Version 1.6 10/19/2001: New options for the name file allow users to designate files of type DATA, DATA(BINARY), DATAGLO, or DATAGLO(BINARY) as either input or output files. One of the options (either "OLD" for an input file or "REPLACE" for an output file) may be placed after the file name on the line listing the file type, unit number, and file name. If "OLD" is specified, the file must exist when the program is started. If "REPLACE" is specified and the file exists when the program is started, the existing file is deleted, then opened. The status of each file ("OLD", "REPLACE", or "UNKNOWN") is now shown in the global output file. Note that the "OLD" or "REPLACE" option is not required. If neither is listed, the file status is shown as "UNKNOWN" and program execution continues normally. When output to a BINARY file from an earlier model run exceeds the amount of output generated by the current model run, specifying "REPLACE" may be required to ensure the file does not include output from the previous run after the output generated by the current run. The options may be entered in any combination of uppercase and lowercase letters. This version fixes a bug that caused simulation time to be truncated, for runs that involve calculation of sensitivities, to a simulation period one time step too short when the last flow observation in a transient simulation falls in a later time step than the final head observation. Corrected problem that caused observation sensitivities for constant-head flow observations to be incorrect for log-transformed parameters. STR6 documentation (str6.pdf) was modified to clarify the format for STR input. ADV2 was updated to fix a problem in the calculation of particles released to weak sinks. In addition, a change was made in the reading of the WTQ array. See adv.txt for the modified format required to input the WTQ array. o Version 1.5 08/16/2001: The LMT Package was added -- it adds support for generating a link file to be read by the contaminant-transport program MT3DMS (see ). Fixed bug that caused interpolation to be incorrect for a head observation that occurs before the end of the first time step of an initial transient stress period. LMG was revised to work without causing program to bomb when using parallel processing. Fixed bug where divide-by-zero error would occur if using parallel processing with more than one MPI process and one parameter is being estimated. Hydfmt and hydpost were revised to enable them to read the binary files generated by MODFLOW-2000 beginning with version 1.2. o Version 1.4 07/10/2001: A new solver package, the LMG (Link-AMG) Package, was added. LMG provides a link to the Algebraic Multi-Grid (AMG) solver. A problem related to RMAR of the Parameter-Estimation (PES) Process input file was fixed. Prior to the fix, RMAR was set to 0.001 regardless of the value entered in the PES input file. Output describing RMAR, which is written to the global file, was corrected. Support for the use of the tab character as a separator in free-format input was added. YCINT2K was modified to correct an error in calculated confidence intervals. Previously, the variance of the measurement error expected for prediction was incorrectly included in the calculation of confidence intervals. o Version 1.3 06/11/2001: Fixed problem in LPF where if a lower cell goes dry while the cell above is still wet, there is a non-zero vertical conductance to a dry cell, causing calculated heads and flows to be in error. Fixed problem in PCG2 that caused tables of maximum head and residual changes each iteration to be printed when MUTPCG is set to 1. Fixed bug in modified Gauss-Newton routine PES1GAU1AP that caused divide-by-zero error when a sensitivity equals zero. Support for printing of auxiliary variables was increased from 2 to 5 auxiliary variables. Added support for calculation of approximate sensitivities for STR parameters when ICALC of the STR Package is > 0. The formulation relies on the assumption that the dependence of the stream stage on SFR parameter values is negligible. In most cases, the error introduced by this assumption is small and the sensitivities can be used for parameter estimation. If the user does not feel that this assumption is justifiable, the use of UCODE (Poeter, E.P., and Hill, M.C., 1998, Documentation of UCODE, a computer code for universal inverse modeling: U.S. Geological Survey Water-Resources Investigations Report 98-4080, 116 p.) is recommended. Fixed bug that caused sensitivities of STR-Package flow observations to STR parameters to be incorrectly calculated in some cases. The ADV Package was modified to fix a problem with tracking through confining layers. The problem occured when observation times or particle-tracking time step (ADVSTP) times were reached when the particle was within a confining unit, the subsequent tracking was incorrect. The HUF Package was modified to fix a number of bugs. Updated the way flow between constant-head cells are calculated to correspond to the equivalent calculation in the LPF Package. Messages printed when STAT-FLAG for a prior-information equation is invalid were reworded to improve clarity. Several bug(s) related to parallel processing were fixed. Corrected output of calculated error variance (variable VAR) to _b1 files. BEALE2K was modified to read the revised format of _b1 files. MFPTO2K was corrected to correctly account for common error variance (variable EV of MODFLOWP input data set 5) values other than 1.0, and to account for numbers of time steps and stress periods correctly. o Version 1.2 04/12/2001: The program was modified to support the use of non-standard options in OPEN statements for unformatted files. This is useful on personal computers because it makes it possible to use different compilers and still have MODFLOW produce unformatted files with the same structure. When standard Fortran is used, different compilers result in unformatted files that have different structures. In MODFLOW-2000, the files affected are those listed in the name file with file type DATA(BINARY) or DATAGLO(BINARY) and those referenced in array-control records for the array-reading utility modules U2DINT and U2DREL where FMTIN is specified as "(BINARY)". The include file openspec.inc (in the \src\mf2k\ directory in the MODFLOW-2000 distribution) describes the specifiers for the OPEN statement that can be modified. The runfile distributed by the USGS for use on personal computers uses options in the OPEN statement that cause the structure of unformatted files to be different than what was used in previous versions of the runfile. The new files are referred to as "unstructured" because they eliminate the vendor-specific structure normally included in such files. Programs that will read unformatted files produced by the USGS distributed MODFLOW-2000 runfile will need to be modified to read unstructured, unformatted files (using OPEN-statement specifiers and options appropriate for the compiler being used) and recompiled. Updated versions of ZONEBUDGET and MODPATH that read the new unstructured unformatted files are being released concurrently with this new release of MODFLOW. The calculation of dimensionless scaled sensitivities (DSS) for log-transformed parameters was modified. Previously, DSS was calculated as X*B*(WT**0.5)*abs(ln(B)), where X is the sensitivity of the observation to the native parameter value, B is the parameter value, WT is the weight associated with an observation, abs() is the absolute-value function, and ln() is the natural logarithm function. Now, DSS is calculated as X*B*(WT**0.5). The revised formula allows DSS and composite scaled sensitivity (CSS) values to be the same whether a parameter is log-transformed or not. This change also eliminates the problem that for log-transformed parameters with a value of 1.0, DSS and CSS would be listed as 0.0. The format of _rs files generated when OUTNAM is not 'NONE' was changed to provide more significant figures for values passed to RESAN2K. RESAN2K was similarly changed to read the new format of _rs files. Fixed bug in RESAN2K that caused incorrect results when prior-information equations were used. Fixed bug in LPF Package that caused incorrect results when VKCB parameters are used and a layer without a confining bed overlies a layer with a confining bed. Fixed two bugs in parallel-processing capability. The first bug caused the program to fail occasionally and inconsistently. The second caused incorrect values of solver discrepancy to be output to the global file in some circumstances. Fixed bug that caused program to fail or produce incorrect results when ISENS > 0 for a single, non-log-transformed parameter. Fixed bug in HFB Package that caused program to generate zero sensitivities for HFB parameters in some circumstances. This bug would have had unpredictable, incorrect results for sensitivities when nonparameterized flow barriers were specified. Revised format of table of smallest and largest weighted residuals to list percent contribution to the sum of squared, weighted residuals objective function. This table now may include residuals related to prior information. Clarified messages written to LIST file when head observations are omitted because IBOUND=0 at cell(s) required for interpolation. Corrected problem related to cases where a prior-information equation is for a log-transformed parameter and STAT-FLAG indicates that STAT is a coefficient of variation. Previously, the weight was not being calculated correctly in this situation. Output related to prior information was clarified. Fixed bug that caused printed unweighted measured, calculated, and residual values for correlated prior information to be incorrect. Made change to allow user, when prompted to enter a name file interactively, to enter either the full name file name or just the base name, to which ".nam" will be added when the program tries to find the name file. ADV Package was changed to fix a problem related to tracking a particle through a confining bed. Fixed bug that caused error when LASTX (in PES file) was specified > 0. Fixed bug that caused RMAR (in PES file) to be ignored. Fixed bug that caused _rs file to contain incorrect parameter number for parameters with correlated prior information in certain circumstances. If a parameter definition for a list-type parameter does not include at least one cell, the program now stops with an error message. Moved printing of ordered residuals from GLOBAL file to LIST file. Modified program to allow it to run in batch mode successfully even if parameter estimation fails to converge for one of the data sets. Array dimensions in RESAN-2000 were increased to allow up to 500 parameters, 10,000 observations (total), 500 flow observations, 100 prior-information equations, and 100 parameters with correlated prior information. BEALE-2000 was changed to correctly write names of observations other than head observations. Array dimensions were increased to allow up to 10,000 observations (total), 10,000 head observations, 500 parameters. o Version 1.1 01/17/2001: Added IBS, HUF, LAK, ETS, and DRT Packages. Added support for the HYDMOD option -- Please see Hanson and Leake (1999, USGS Open-File Report 98-564) for documentation. The EVT Package was modified as described in the LAK documentation to include the option (NEVTOP=3) to apply ET to the uppermost wet cell in a vertical column of cells as done in the RCH Package. Added support for generation of an output file with the extension _pc when OUTNAM is not 'NONE', the Parameter-Estimation Process is active, and parameter estimation converges. This file contains a record for each estimated parameter listing: parameter name, LN, optimized parameter value, BL, BU, lower 95-percent confidence limit on the parameter value, and upper 95-percent confidence limit on the parameter value. Fixed bug that incorrectly allowed substantial parameter changes in some circumstances when MAX-ITER was specified as 0 in the PES input file. An effect of this fix is that if MAX-CHANGE is specified as a value smaller than 0.4, the MAX-CHANGE (used) value never differs from MAX-CHANGE. Fixed bugs related to writing of _ww and _ws files. Previously, incorrect values were written for weighted observed and weighted simulated values for flow observations. Fixed bug that caused MODFLOW-2000 to incorrectly stop with an error message when constant-head flow observations were specified and the GHB Package was inactive. Fixed problem where incorrect values of BL and BU were written to the GLOBAL file under some circumstances. Increased the maximum number of allowed parameters from 40 to 500. Eliminated 1000-observation limit. Now the number of observations and prior-information equations are not limited by the program. Changed the file-type identifier required in the name file to activate the ADV Package from ADV to ADV2. The directory structure was changed to put para-non.f into the \src\mf2k\serial\ directory and para-mpi.f into the \src\mf2k\parallel\ directory. o Version 1.0 07/20/2000: Initial release of MODFLOW-2000, RESAN-2000, YCINT-2000, and BEALE-2000.