Friday, February 29, 2008

PEST Seminar in May

I have scheduled my next Advanced Groundwater Vistas & Pest course. The seminar will be held May 28-30 at the Residence Inn, Tampa (near the airport).

Originally, my Advanced course had three topics of relatively equal weight: (1) model calibration, (2) uncertainty analysis using monte carlo techniques, and (3) optimization of remedial design and well field management. This evolved into about half the course devoted to model calibration, focusing on Pest, and half on the other 2 topics with a bit of MODFLOW-SURFACT thrown in. The next course will be primarily a PEST course, both from a calibration standpoint and for analysis of uncertainty. Topics will include the basics of calibration, use of Pest in zone-based models, Pilot Point calibration with Pest (including regularization and Singular Value Decomposition), and calibration-constrained monte carlo techniques. I will briefly discuss optimization as well, but probably just half a day.

I will take a maximum of 12 to 14 attendees. The seminar will cost $1,250 per person and you will need to bring a laptop computer.

Go to our web site at and click on seminars for more info.

The course is about half full now so don't wait too long if you plan on attending.

A Flurry of Activity

If you've been following this blog, you'll see we jumped from V5.12 Build 1 to V5.16 Build 1. Here is a summary of what has changed since V5.12 Build 1:

- There was a problem encountered when importing an existing MODFLOW Well package using the option to create Analytic Wells instead of BC Wells. If there were more than 9 wells in a single cell, then the ones beyond 9 were omitted.

- We added an error checking function for Fracture Wells to make sure they do not have a bottom layer that is in a no-flow cell. Turns out that MODFLOW-SURFACT cannot deal with this situation, giving bogus results when this occurs.

- We had to modify our DXF import routine for map files to account for some unexpected things in the Civil CAD program.

- When importing a transient calibration target or analytic well from a text file, you can often get trailing commas on the ends of lines when using Excel. In the past, GV would not import this type of file properly. We added some code to ignore these trailing commas so you no longer have to edit these files in a text editor after generating them from Excel.

- GV has long had a version of MODFLOW96 in double precision. We now added a version of MODFLOW2000 that is also in double-precision. Just go to Model/MODFLOW/Packages and check the double precision flag beneath the MODFLOW version selector. You will notice that the DLL is changed from MF2KWIN32.dll to MF2KDPWIN32.dll. The binary output files from these double-precision versions of MODFLOW are in single-precision so that there are no incompatibility issues with MODPATH and MT3DMS.

- When color flooding matrix properties you can now specify that only cells within the bounds of the matrix are color flooded. A matrix property is one that does not have an associated zone database. For example, bottom elevations are by default a matrix property. You turn properties from zone to matrix using Props/Options. By default, GV will assign blue to the lowest value in the matrix and red to the highest and color flood in between. You can manually change the bounds on a matrix to change what value represents blue and red using Props/Property Values/Manually Reset Matrix Bounds. Normally any value less than the lowest bound would still be blue and any value above the maximum value is red. If you select Props/Options and uncheck the option at the bottom of the dialog, however, then GV will not put a color in these cells that lie beyond the bounds you have established.

- Those of you with good powers of observation will see a new submenu under Model/Pest called Null Space Monte Carlo. This is a powerful feature in Pest where you can generate any number of calibrated models, all with different properties. These calibrated models can then be used to evaluate model uncertainty. This is analogous to the Stochastic part of Groundwater Vistas, which generates realizations that are not necessarily calibrated. In the past, you would need to use "Post Conditioning" to throw out realizations that do not meet some calibration criteria. Anyway, we are in the beginning stages of supporting Null Space Monte Carlo. For now, your menu will be grayed out so you cannot access it. We hope that by late April we will have this finished and those of you with the Advanced or Enterprise versions of Vistas will be able to start using it.

Wednesday, February 20, 2008

GV Version 5.12 Build 1

There are three new features added to Groundwater Vistas since Version 5.11 Build 3. Two of these features were requested by one of our users. I point this out because many of you may not realize that you have the power to help shape Groundwater Vistas. If you have an idea for a new option or feature (or modifications to existing tools), just email them to us, along with a clear explanation of what you need. If we agree that the new feature makes sense, then we will add it, often very quickly.

So, here are the new features:

- You can now modify analytic well flow rates for groups of wells automatically using AE/Modify/Well/Flow Rates

- You can have GV set the top of layer 1 in the MODFLOW2000/2005 Discretization Package (DIS) to be the starting heads in layer 1. This can be very handy if you have an unstable model where layer 1 is unconfined but is having dry cell problems. Using this option, you can make all layers confined, thereby making the model stable. Because the water table is the top of layer 1, the transmissivity is then approximately correct. Note that this only applies to MODFLOW2000 and MODFLOW2005. Note also that this option does not carry over to the ET surface or other packages which have the top of layer 1 defined in their own input file.

- The final option goes along with the previous one. There is a new option on the Model/MODFLOW/Package Options - Initial Heads tab, which sets the starting head in dry cells to be a specified height above the layer bottom in that cell. This means that the saturated thickness of all layers will be positive.

Monday, February 11, 2008

GV 5.11 Build 3 and other news

There are two modifications in GV 5.11 Build 3 released over the weekend:

- We added row and column locations to the analytic well text file export
- When importing a point shapefile for streams, the bottom elevation field was not being activated.

I also received an email last week from Arlen Harbaugh that the USGS intends to put the DDREFERENCE flag in the MODFLOW2005 output control file in the next release (1.5.00 in about a month). This is an obscure but useful feature in MODFLOW2000 that allows drawdown to be computed from the results of any particular time step. When the DDREFERENCE keyword is placed in the output control file (word version only), drawdowns from that point forward are computed based on the results of that time step. Unfortunately it was omitted from MODFLOW2005.

To use this feature in Groundwater Vistas, you select Model/MODFLOW/Package Options - Output Control tab and turn on the use of customized output control, also entering the number of time steps where you would like to save information. Then select Model/MODFLOW/Custom Output Control. There is a column in the spreadsheet for the DDREFERENCE flag (enter 1 to turn it on for a particular time step).

While on this topic, note that MODFLOW-SURFACT Version 3 now also has this flag. We requested that it be added to SURFACT and HGL was kind enough to make that change. At our request, they also added the ability to have steadystate and transient stress periods in the same simulation.