Hi there,
when using grib2arl on ECMWF 0.125 degree data with either -g3 or -g4 options, there appears to be a scaling problem: SHGT T02M WWND features are displaced with respect to the proper terrain location. My guess is that grib2arl is having trouble interpretting the last decimal place in the resolution and interpretting it as 0.13 degrees. 0.25 degree ECMWF data works fine.
I tried to get around this bug by using the -g0 option to reproject the ec0.125 degree data onto a conformal grid. This correctly placed the SHGT and TMP fields, BUT the UWND fields failed to reproject properly (some levels had UWND = 0). I would guess this is a southern hemisphere issue.
If anyone is interested in looking at these issues I would be happy to make the sample data and control files available for reproducing the bug.
Cheers,
Cory.
scaling issue with ECMWF 0.125 degree data
-
- Posts: 451
- Joined: November 9th, 2012, 4:23 pm
- Registered HYSPLIT User: Yes
Re: scaling issue with ECMWF 0.125 degree data
It shouldn't be a southern hemisphere nor grid size >1000 issue. Yes, I'd like to get a sample of the ECMWF data and your output. Please email me at barbara.stunder@noaa.gov
-
- Posts: 1
- Joined: May 1st, 2014, 10:02 pm
- Registered HYSPLIT User: Yes
Re: scaling issue with ECMWF 0.125 degree data
The following patch needs to be applied to overcome this issue with ECMWF 0.125:cory wrote:Hi there,
when using grib2arl on ECMWF 0.125 degree data with either -g3 or -g4 options, there appears to be a scaling problem: SHGT T02M WWND features are displaced with respect to the proper terrain location. My guess is that grib2arl is having trouble interpretting the last decimal place in the resolution and interpretting it as 0.13 degrees. 0.25 degree ECMWF data works fine.
I tried to get around this bug by using the -g0 option to reproject the ec0.125 degree data onto a conformal grid. This correctly placed the SHGT and TMP fields, BUT the UWND fields failed to reproject properly (some levels had UWND = 0). I would guess this is a southern hemisphere issue.
If anyone is interested in looking at these issues I would be happy to make the sample data and control files available for reproducing the bug.
Cheers,
Cory.
--- grib2arl-org.f 2014-05-02 14:11:44.001413536 +1200
+++ grib2arl-fixed.f 2014-05-02 14:13:47.773283611 +1200
@@ -939,7 +939,7 @@
! grid geolocation parameters and projection
DO I=1,12
- WRITE(30,'(A20,F10.2)')LABEL(I+3),GRIDS(I)
+ WRITE(30,'(A20,F10.3)')LABEL(I+3),GRIDS(I)
!# UKMET WRITE(30,'(A20,F10.2)')LABEL(I+3),GRIDS(I)
END DO
@@ -2024,7 +2024,7 @@
! shift to 0 to 360 system
IF(CLON1.LT.0.0)CLON1=360.0+CLON1
- WRITE(50,'(1X,A,4F10.2)') &
+ WRITE(50,'(1X,A,4F10.3)') &
'Input data corner and spacing: ',CLAT1,CLON1,DLAT,DLON
ALLOCATE (XVAR(NLON,NLAT),RVAR(NLON*NLAT))
-
- Posts: 451
- Joined: November 9th, 2012, 4:23 pm
- Registered HYSPLIT User: Yes
Re: scaling issue with ECMWF 0.125 degree data
Thank you for the patch. I've made the change in our grib2arl.f