Can PX LSM be used with ERA5?

[Haorui Wang]

Hello everyone,

When I used the PX LSM with the ERA5, I had unusual temperatures around the shoreline and inland lakes. When I replace the ERA5 with the FNL, I don't have this problem. I also don't have a problem when I use the NOAH LSM with the ERA5. I am curious if the PX LSM can be used with the ERA5? I'd appreciate it if someone could give me some suggestions.

Related images and namelist have been uploaded.

 

[kwerner]

Hi,
To my knowledge, this issue has not been reported previously. Are you using the "invariant fields" file with the ERA5 data? See How to Process ERA5 Data. If you're doing all of that correctly, can you let me know if it happens with any ERA5 (i.e. other dates) data, and different domains? Can you also let me know which version of WRF you're using? Thanks!

 

[Haorui Wang]

Hi,
To my knowledge, this issue has not been reported previously. Are you using the "invariant fields" file with the ERA5 data? See How to Process ERA5 Data. If you're doing all of that correctly, can you let me know if it happens with any ERA5 (i.e. other dates) data, and different domains? Can you also let me know which version of WRF you're using? Thanks!

Dear kwerner,

Thank you so much. I used WPS-4.6.0 and WRF-4.6.0. The resolution is 27km. I simulated according to the steps you provided, but the result is still the same as before. According to your suggestion, I simulated a total of 10 days from 2020-05-21 to 2020-05-31 and tried different domains. The results show that no matter which day or domain it is, the ERA5+PX LSM always shows a strange 2m temperature, while the ERA5+NOAH LSM and FNL+PX LSM are normal. Therefore, I think the ERA5 processing is correct (after all, ERA5+NOAH shows reasonable results), and the problem should lie in the coupling of ERA5 and PX LSM. Further I found that in the ERA5+PX LSM case, only the soil temperature, ground temperature and air temperature near the ground are abnormal, and the remaining variables are reasonable. I suspect that this may be related to the soil model of the PX LSM. I would appreciate any suggestions you may have!

Thanks again!

​

 

[Haorui Wang]

UPDATE:To rule out the computer environment, I used 2 different computers with the same results. I double checked every relevant variable in the input and output files of WPS and WRF.I found that the temperature anomalies near the shoreline and inland lakes are entirely due to anomalies in the deep soil temperatures. The ERA5 data always initialises the soil temperatures near the shoreline and inland lakes to 0 K (the FNL data does not lead to such temperature anomalies) and this is what I see from the wrfinput file, whether I am using the PX LSM or the NOAH LSM. The difference is that when using the NOAH LSM, the soil temperature quickly returns to normal as the simulation runs. When using PX LSM, on the other hand, as the simulation runs, the deep soil temperatures near the shoreline and inland lakes stay close to 0 K, resulting in anomalies in the shallow soil temperatures and air temperatures. Even when I manually initialise the deep soil temperature to a reasonable value, it rapidly becomes unreasonable as the model runs (temperature will be significantly higher or lower compared to the surrounding area). Overall, when ERA5 input data is used with PX LSM, it always produces unexpected soil temperatures near coastlines and inland lakes, no matter how I test. I suspect the problem is with the landsea-mask or with the PX LSM soil model. If someone can give me some advice or do some tests, I'd appreciate it.

If the soil temperature has not been initialized manually:



If the soil temperature has been initialized manually (Initialize according to the official tutorial at https://www2.mmm.ucar.edu/wrf/users/docs/PX-ACM.pdf):