Jonas Posted September 23, 2016 Report Share Posted September 23, 2016 Hi again! I attempt to use Null Space Monte Carlo in a similar manner as presented in PEST - Null Space Monte Carlo I tutorial (http://gmstutorials-10.1.aquaveo.com/MODFLOW-StochasticModeling-NullSpaceMonteCarloI.pdf) As presented in the attached text file I have some model that converge although most models does not converge for some reason (Ideas on how to adress this are also appreciated). As seen in the text file, there are different calculated values for the pilot points between the four converged solutions. The bpa files for the different solutions are also different. All good so far. My issue is that there is no difference between the grid files of the HK parameter for the four different solutions. Any ideas on how to solve this issue? Could it have something to do with the Kriging process? Best regards, Jonas T40_nsmc.txt Quote Link to comment Share on other sites More sharing options...
Jonas Posted September 26, 2016 Author Report Share Posted September 26, 2016 I use the NWT solver with UPW in my model. When I changed to the same solver in the NSMC I tutorial, the same thing happens. Different values in the bpa files but non-unique grid files with HK values between the different solutions. Ideas on this issue are appreciated! Quote Link to comment Share on other sites More sharing options...
Jonas Posted September 28, 2016 Author Report Share Posted September 28, 2016 Now I have simplified the model a lot to only one layer and use the LPF package. It seems that the only solver that creates unique grid files is 2000. For 2005 and NWT it is the same issue as previous specified. Others with similar experience? Kind regards, Jonas Quote Link to comment Share on other sites More sharing options...
woodward Posted September 28, 2016 Report Share Posted September 28, 2016 I have used PEST with Groundwater Vistas. I assume it's similar in GMS. If you want the realisations to converge, you can change the convergence criterion/increase the number of iterations. The number of iterations is usually set to only 2 for NSMC to make it fast. This might not be enough for your model. Have you checked the model batch file that PEST calls? Can you see where the HK file is created? Maybe the batch file is failing to create the HK file for some reason. Quote Link to comment Share on other sites More sharing options...
Alan Lemon Posted September 30, 2016 Report Share Posted September 30, 2016 Jonas Could you contact support@aquaveo.com to look into the HK from each realization? The simpler version of your model will be easiest for the support team to look at. Quote Link to comment Share on other sites More sharing options...
Jonas Posted October 5, 2016 Author Report Share Posted October 5, 2016 Thank you Alan, I have done so. Looking forward for their response. Woodward, the more simplified converges in almost all cases. The problem is that non-unique grids are created. Quote Link to comment Share on other sites More sharing options...
Jonas Posted January 27, 2017 Author Report Share Posted January 27, 2017 For others interested in this issue, the support team has now fixed the bug that caused the issue of PEST and null space MC with the NWT solver. Download the latest nightly build (Jan 25th) from: http://www.aquaveo.com/downloads-gms Woodward, it was a pleasure to read your article, "Uncertainty in the modelling of spatial and temporal patterns of shallow groundwater flow paths: The role of geological and hydrological site information" in Journal of Hydrology where you have applied this method. Quote Link to comment Share on other sites More sharing options...
woodward Posted January 30, 2017 Report Share Posted January 30, 2017 Why thank you sir! Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.