Inconsistency between GFS analysis RH and met_em RH

[chunchih]

Hi,

Out of curiosity, I looked into the RH (relative humidity) field in my GFS analysis data (the 0.5 deg GFS analysis) and compared that with the RH field in my met_em file processed by WPS, at the same time. I found that at higher pressure levels (i.e., closer to the ground), the RH in my met_em is fairly consistent with the GFS analysis RH (a figure showing the 850 hPa RH from the GFS analysis and met_em is attached. The 'RH_P0_L100_GLL0' variable from my GFS analysis file, after converting from grb2 to netCDF, is plotted). However, as I go higher in the atmosphere, the met_em RH becomes significantly smaller than the GFS analysis RH at the same pressure level (a figure showing the RH at 300 hPa from the GFS analysis and met_em is also attached). What is the reason behind this change?

Another strange thing I found is that in my GFS analysis data, there is no point with a negative (<0%) RH nor a point with RH > 100%. However, in my met_em, I found a few points with slightly negative RH (as low as -3%) and a few with RH slightly greater than 100% (as large as 101%). Is this normal? The slightly negative RH worries me a bit as that doesn't really make sense.

Thanks,
David

 

[haiqingsong]

Hi,

Out of curiosity, I looked into the RH (relative humidity) field in my GFS analysis data (the 0.5 deg GFS analysis) and compared that with the RH field in my met_em file processed by WPS, at the same time. I found that at higher pressure levels (i.e., closer to the ground), the RH in my met_em is fairly consistent with the GFS analysis RH (a figure showing the 850 hPa RH from the GFS analysis and met_em is attached. The 'RH_P0_L100_GLL0' variable from my GFS analysis file, after converting from grb2 to netCDF, is plotted). However, as I go higher in the atmosphere, the met_em RH becomes significantly smaller than the GFS analysis RH at the same pressure level (a figure showing the RH at 300 hPa from the GFS analysis and met_em is also attached). What is the reason behind this change?

Another strange thing I found is that in my GFS analysis data, there is no point with a negative (<0%) RH nor a point with RH > 100%. However, in my met_em, I found a few points with slightly negative RH (as low as -3%) and a few with RH slightly greater than 100% (as large as 101%). Is this normal? The slightly negative RH worries me a bit as that doesn't really make sense.

Thanks,
David

I had the same problem last year. Do you know how to solve it

 

[chunchih]

I am not sure if both are normal behaviors.

My guess is that for the first issue, the RH with respect to ice is computed in the GFS analysis but in WPS and WRF, RH is only calculated with respect to liquid. I did some tests and found that the difference between the GFS analysis RH and met_em RH increases as the temperature decreases to well below 0C.

--Update: my guess is confirmed, I think. In ungrib's (/ungrib/src/) rrpr.F, the GFS RH (with respect to ice) is replaced with RH with respect to liquid in the subroutine fix_gfs_rh:

subroutine fix_gfs_rh (ix, jx, plvl)
! This routine replaces GFS RH (wrt ice) with RH wrt liquid (which is what is assumed in real.exe).
use storage_module
implicit none
integer :: ix, jx, i, j
real :: plvl, eis, ews, r
real, allocatable, dimension, :: rh, tt

 

[Ming Chen]

You are right. RH in GFS is calculated with respect to ice when T is below zero degree. In WRF, however, RH is always calculated with respect to liquid.
This is why you see the difference in RH between GFS and met_em files.

 

[chunchih]

Hi Ming,

Thanks!

However, I am still unsure about the second issue with the slightly negative RH in met_em.

 

[Ming Chen]

Hi David,

I guess there are a few reasons for the slight negative RH:

(1) Can you take a look at your METGRID.TBL, and find the section for RH, e.g., below is what I can see in my METGRID.TBL

name=RH
interp_option=sixteen_pt+four_pt+average_4pt
fill_missing=0.
fill_lev=200100:const(-1.E30)

What is the value of fill_lev in your file?

if it is fill_lev=200100:const(-1.), then it may explain why slight negative RH appears in met-em file. Otherwise, I don't think interpolation is responsible for negative RH.

(2) in rrpr.F, RH is recalculated from specific humidity or dew point and other variables of temperature, pressure, etc. I am suspicious this is when negative RH is created. You may examine your intermediate file to see whether there is negative RH.

Hope this is helpful. Sorry that I don't have a definite answer to your question at present.

 

[chunchih]

Hi Ming,

In my METGRID.TBL, the fill_lev line for RH is indeed fill_lev=200100:const(-1.E30).

I haven't checked my intermediate file yet. rd_intermediate.exe only printed out limited information about the fields in my intermediate file. Is there another utility program that I should use?

Anyway, I am assuming the slightly negative RH in my met_em file isn't too big of an issue. There is no negative water vapor mixing ratio in my first timeframe WRF output. Please correct me if I am wrong.

 

[Ming Chen]

I agree that it is not a big deal, although it remains mysterious why negative RH could appear in met_em file. We will pay attention to this issue ....