About EPIC1102 Code

22 views
Skip to first unread message

TX H.

unread,
Jun 9, 2021, 4:34:28 PMJun 9
to EPIC / APEX Modeling Forum
Hi,
I have several questions related to the EPIC 1102 code when running the model. Could you please help me with them?
Q1: Random numbers: IX
IX was initialized with 21 very large numbers in Line 604 in the main program, but the actual values when run the code do not match the initialized values (under Intel Fortran compiler). The possible reason is that IX was declared as INTEGER (4) which can not store the very large numbers provide in the initialization program (AINIX.f90). How do the random numbers matter to some results we are interested (e.g., hydrology, soil,  or crop related variables)?

Q2: In Line 462 of BSIM.f90 
What does the variable GWPS mean? It was declared and allocated but not initialized. 
Thus, the GWPS has different values when running the EPIC compiled by Intel Fortran and gfortran compiler. Because of different values of some variables such as GWPS, the same model produces different results using the same inputs under different compilers.

Q3: Is the EPIC 1102 still a main component of APEX? 

Thank you!

Best,
Tony



Luca

unread,
Jun 14, 2021, 11:14:15 AMJun 14
to EPIC / APEX Modeling Forum
Dear Tony,
Starting from your third question, as you know APEX is based on EPIC, and the modifications introduced in EPIC are then ported to APEX. Are you interested in any specific feature of EPIC that you need in APEX?

Looking at the code, I think GWPS is the pesticide in the groundwater. We are aware that some variables are not initialized and this can lead to different results even when running executables created with different versions of Intel Fortran Compiler. Thanks for reporting about this variable not being initialized, we will take action to fix this issue.

About your first question, I will need some more time to look at the code but, I have the feeling that the random numbers are used only to generate weather data when the daily weather data are not available. If my assumption is true, this should not be a problem if a complete set of weather data is provided for the simulation.

Best,
Luca

TX H.

unread,
Jun 14, 2021, 12:11:29 PMJun 14
to EPIC / APEX Modeling Forum
Thank you so much, Luca.
I am actually modifying the EPIC code so that the EPIC model would be much easier to read and extend (e.g., adding functionalities by modules). Basically, I am making it follow the standards and style of modern Fortran. I asked the third question because I hope to make sure that my current work might be useful in the future, for example, being ported to the APEX model. Discrepancies appear when testing and comparing my modified version of EPIC with the online version you are handling. Sometimes the discrepancy is because of variables and sometimes because I do not understand the mechanism behind the code, so I posted them here to ask for your help, explanation, or instructions.  

Best,
Tony
Reply all
Reply to author
Forward
0 new messages