Ensemble collapse and NaN values during Croton test case DA by HydroDART
Fatemeh Shirkhanloo
Dear WRF-Hydro Community,
I am running the standalone version of WRF-Hydro coupled with DART (HydroDART) for the Croton test case to perform Data Assimilation from 2011-08-26 to 2011-09-01.
As you can see in the attached graph, my data assimilation cycle is failing after 2011-08-29, exactly after the hurricane event. As a result, in my output file (preassim_sd_d01.2011082705.nc and analysis_sd_d01.2011083000.nc), I am seeing NaN values for qlink1 variable. This indicates that something is corrupting the model, which causes the ensemble spread to collapse to zero in subsequent cycles.
Any help or assistance on how to resolve this would be greatly appreciated.
Thank you,
Fatemeh
Arezoo RafieeiNasab
--
You received this message because you are subscribed to the Google Groups "wrf-hydro_users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wrf-hydro_use...@ucar.edu.
To view this discussion visit https://groups.google.com/a/ucar.edu/d/msgid/wrf-hydro_users/e8eb506d-dbe3-4285-915c-132bc816545bn%40ucar.edu.
Fatemeh Shirkhanloo
Dear Arezoo,
Yes, I investigated this issue when I was working with the Croton test case.
I found that qlink1 became negative after the hurricane event. Since streamflow should not be negative, these values appeared to lead to NaN values in HydroDART, which then affected the following assimilation cycles.
I also noticed a large overestimation in the prior ensemble. Before assimilation, the simulated discharge exceeded 80 m³/s, while the observed discharge peaked at around 25 m³/s. After examining the perturbed forcing files, I found that the forcing perturbations were introducing too much water into the system.
In the forcing-perturbation setup, the imposed minimum standard deviation was too large for the Croton test case. In my own case study, I reduced this lower limit from 5 to 0.25. This decreased the magnitude of the perturbations, produced a more realistic prior ensemble, and prevented the unrealistically high streamflow values that were leading to negative qlink1 values and subsequent NaNs.
This adjustment resolved the issue in my case study, although the appropriate lower limit may vary depending on the basin characteristics and the forcing variables being perturbed.
Best regards,
Fatemeh
Arezoo RafieeiNasab
To view this discussion visit https://groups.google.com/a/ucar.edu/d/msgid/wrf-hydro_users/CAKnuMgqYj1nntNTb44U0nVPQ57z6T5Ui0ium5i-_Syh_n1CXfw%40mail.gmail.com.
Fatemeh Shirkhanloo
Dear Arezoo,
Thank you for your kind words. Please don't worry about the delay.
I'm happy to hear that my notes on the Croton test case will be helpful to others running into similar data assimilation issues.
Have a great week!
Best regards,
Fatemeh