You could probably activate SVD regularization by setting the appropriate values on the last tab of the tool. We do not use SENSAN as it does not seem to make life much easier than just writing a Tcl or R script to do sensitivity analysis for a particular model.
I'm currently trying to use the Predictive analysis section of PEST. I have read the PEST manual several times and have put in the the rquired information to Parameter estimation using the results from a previous parameter estimation and made some changes in the prediciton setting for the first three settings. When I try to run this I always get "No model value set for prediction" and nothing happens. I have tried numerous different options and always get the same answer. Am I missing something prediction value set?
What you have to do is click on the checkbox for predictive analysis, and you will see a message come up at the top of the Actions tab instructing you to click on the value to be predicted. Now click on the component in the model or tree view, and the caption of the value replaces (none) in the predictive analysis panel. You can now run the prediction.
You may well be the first Simile customer to use this feature, so I am keen to hear how it goes!
--Jasper
We are closer. In the parameter file I put in all the files from the prvious parameter estimation (calibration estimates). In the out puts I put in the salinity file (840 VALUES) used for parameter estimation. Settings are defalt. I select value salinity.ma which are the predictions for 840 out put files. When I run the comment "Some measured data not specified: /saliniy ma/salinity ma. After clicking OK I get an error message. The execution monitor says warning to many measurements for indices each of the 840 values of Salinity ma. I think this file is overloading the system and I need to cut it down some how. Any suggestions?
Comments
#1
You could probably activate SVD regularization by setting the appropriate values on the last tab of the tool. We do not use SENSAN as it does not seem to make life much easier than just writing a Tcl or R script to do sensitivity analysis for a particular model.
#2
I'm currently trying to use the Predictive analysis section of PEST. I have read the PEST manual several times and have put in the the rquired information to Parameter estimation using the results from a previous parameter estimation and made some changes in the prediciton setting for the first three settings. When I try to run this I always get "No model value set for prediction" and nothing happens. I have tried numerous different options and always get the same answer. Am I missing something prediction value set?
#3
What you have to do is click on the checkbox for predictive analysis, and you will see a message come up at the top of the Actions tab instructing you to click on the value to be predicted. Now click on the component in the model or tree view, and the caption of the value replaces (none) in the predictive analysis panel. You can now run the prediction.
You may well be the first Simile customer to use this feature, so I am keen to hear how it goes!
--Jasper
#4
We are closer. In the parameter file I put in all the files from the prvious parameter estimation (calibration estimates). In the out puts I put in the salinity file (840 VALUES) used for parameter estimation. Settings are defalt. I select value salinity.ma which are the predictions for 840 out put files. When I run the comment "Some measured data not specified: /saliniy ma/salinity ma. After clicking OK I get an error message. The execution monitor says warning to many measurements for indices each of the 840 values of Salinity ma. I think this file is overloading the system and I need to cut it down some how. Any suggestions?
#5
Disregard previous I had a screwed up salinity ma outputfile.
#6