WRFDA for Himawari radiance

Post Reply
taopeng
Posts: 1
Joined: Wed Sep 02, 2020 2:40 am

WRFDA for Himawari radiance

Post by taopeng » Wed Sep 02, 2020 3:38 am

Hi,
we are just trying to assimilate radiance from Himawari_8 in WRFDA. Our scheme is briefed as follows:
1, the NWP data is derived from ECMWF;
2, the L1 radiance data in NetCDF format is derived from directory /jma/netcdf under ftp://ftp.ptree.jaxa.jp/ ;
3, no conventional observations created by OBSPROC is used;
3, WRFDA 3DVAR assimilation scheme is emplloyed.
4, assosciated settings are listed as below:
a)namelist:
&wrfvar1
var4d=false,
print_detail_grad=false,
/
&wrfvar2
/
&wrfvar3
ob_format=2,
/
&wrfvar4
use_ahiobs = .true.
/
&wrfvar5
/
&wrfvar6
max_ext_its=1,
ntmax=50,
orthonorm_gradient=true,
/
&wrfvar7
cv_options=3,
/
&wrfvar8
/
&wrfvar9
/
&wrfvar10
test_transforms=false,
test_gradient=false,
/
&wrfvar11
/
&wrfvar12
/
&wrfvar13
/
&wrfvar14
rtminit_nsensor = 1,
rtminit_platform = 31,
rtminit_satid = 8,
rtminit_sensor = 56,
rtm_option = 2,
/
&wrfvar15
/
&wrfvar16
/
&wrfvar17
/
&wrfvar18
analysis_date="2020-07-31_18:00:00.0000",
/
&wrfvar19
/
&wrfvar20
/
&wrfvar21
time_window_min="2020-07-31_17:00:00.0000",
/
&wrfvar22
time_window_max="2020-07-31_19:00:00.0000",
/

b)radiance_info/himawari-8-ahi.info:
sensor channel IR/MW use idum varch polarisation(0:vertical;1:horizontal)
478 1 1 -1 0 1.0520000000E+00 1.0000000000E+00 28.30175
478 2 1 -1 0 1.7000000000E+00 1.0000000000E+00 57.58830
478 3 1 -1 0 1.7000000000E+00 1.0000000000E+00 12.69287
478 4 1 -1 0 1.3500000000E+00 1.0000000000E+00 27.33099
478 5 1 -1 0 0.8140000000E+00 1.0000000000E+00 23.24269
478 6 1 -1 0 0.9310000000E+00 1.0000000000E+00 53.35099
478 7 1 -1 0 0.8710000000E+00 1.0000000000E+00 36.07700
478 8 1 1 0 0.9260000000E+00 1.0000000000E+00 33.61592
478 9 1 -1 0 0.9330000000E+00 1.0000000000E+00 33.61592
478 10 1 -1 0 0.7870000000E+00 1.0000000000E+00 33.61592


c)ahi_info:
data source:1.cma hdf5;2.geocat netcdf4;3.jaxa netcdf4;4.ncep bufr
3
nscan
3000
area information for geocat netcdf4 data: lonstart latstart nlongitude nlatitude
1,1,1500,1000
date infomation for cma hdf5 data
2016,07,18,19,00,00


As we renamed radiance data into L1AHITBR and placed it in working directory, the data can be read by da_wrfvar.exe, but warning message was genereted as follows:
--------------------------- WARNING ---------------------------
WARNING FROM FILE: da_read_obs_netcdf4ahi_jaxa.inc LINE: 165
Cannot open NETCDF4 file L1AHITBR
---------------------------------------------------------------
At line 8177 of file da_radiance.f
Fortran runtime error: attempt to DEALLOCATE unassigned‘vlatitude’

So we checked the context of file da_radiance.f near line 8177, as shown:
8172 !--------------------------------------------------------------
8173 ! 1.0 setup from namelist parameter
8174 !--------------------------------------------------------------
8175 nsensor = rtminit_nsensor
8176 allocate (nscan(nsensor))
8177 allocate (nchanl(nsensor))
8178
8179 !----------------------------------------------------------------
8180 ! 2.0 set up some common variables for innovation/observation structure
8181 !----------------------------------------------------------------
8182 iv % num_inst = nsensor
8183 ob % num_inst = nsensor
8184
8185 allocate (iv%instid(1:nsensor))
8186 allocate (ob%instid(1:nsensor))
8187 allocate (satinfo(1:nsensor))

Then we changed setting in namelist into 2 sensors:
&wrfvar14
rtminit_nsensor = 2,
rtminit_platform = 31,31,
rtminit_satid = 8,8,
rtminit_sensor = 56,56,
rtm_option = 2,
/
after running the following error message occurred:
---------------------------------------------------------------
At line 10214 of file da_radiance.f (unit = 26, file = 'radiance_info/himawari-8-ahi.info')
Fortran runtime error: End of file

actually line 10214 does not exist, only 9181 lines in file da_radiance.f.



In adition, could you give us answers of the following questions:
1, is it necessary to read ob.ascii observations from OBSPROC? If not, how to ignore ob.ascii by some setting?
2, does it need radiance data at multiple time points (someting like observations within time window processed by OBSPROC)? If so, how to combine data at different time points into a radiance file?

best regards
Tao Peng

Post Reply

Return to “Radiance DA”