A. Content of Namelist
The following are lists and explanations of the EnKF namelist
variables. Users can also check file params.f90 for the details.
Section nam_enkf
Variable Name Description |
Data Type |
Default |
datein Analysis date in YYYYMMDDHH |
integer |
0 |
datapath path to data directory (include trailing slash) |
Character (len=500) |
“” |
iassim_order = 0 for the order they are read in, =1 for random order = 2 for order of predicted posterior variance reduction (based on prior) |
integer |
0 |
covinflatemax maximum inflation |
real(r_single) |
1.e30 |
covinflatemin minimum inflation |
real(r_single) |
1.0 |
deterministic if true, use EnSRF w/o perturbed obs. if false, use perturbed obs EnKF. |
logical |
true |
sortinc if false, re-order obs to minimize regression errors as described in Anderson (2003). |
logical |
true |
corrlengthnh length for horizontal localization (in km) in north hemisphere |
real(r_single) |
2800 |
corrlengthtr length for horizontal localization (in km) in tropic |
real(r_single) |
2800 |
corrlengthsh length for horizontal localization (in km) in south hemisphere |
real(r_single) |
2800 |
Variable Name |
Description |
Data Type |
Default |
varqc |
Turn on varqc |
logical |
false |
huber |
use huber norm instead of “flat-tail” |
logical |
fales |
nlons |
number of lons |
integer |
0 |
nlats |
Number of lats |
integer |
0 |
smoothparm |
smoothing parameter for inflation (-1 for no smoothing) |
real(r_single) |
-1 |
readin_localization |
If true, read in localization length scales from an external file |
logical |
false |
zhuberleft |
Parameter for “huber norm” QC |
real(r_single) |
1.e30 |
zhuberright |
Parameter for “huber norm” QC |
real(r_single) |
1.e30 |
obtimelnh |
observation time localization in hours over north hemisphere |
real(r_single) |
25.925 |
obtimeltr |
observation time localization in hours over tropic |
real(r_single) |
25.925 |
obtimelsh |
observation time localization in hours over south hemisphere |
real(r_single) |
25.925 |
reducedgrid |
Smooth in a reduced grid with variable number of longitudes per latitude. Number of longitudes chosen so that zonal grid spacing is approx the same as at the equator logical |
false |
|
lnsigcutoffnh |
length for vertical localization in ln(p) over north hemisphere for conventional observation |
real(r_single) |
2.0 |
lnsigcutofftr |
length for vertical localization in ln(p) over tropic conventional observation |
real(r_single) |
2.0 |
lnsigcutoffsh |
length for vertical localization in ln(p) over south hemisphere for conventional observation |
real(r_single) |
2.0 |
lnsigcutoffsatnh |
length for vertical localization in ln(p) over north hemisphere for satellite radiance observation |
real(r_single) |
-999.0 |
lnsigcutoffsattr |
length for vertical localization in ln(p) over tropic satellite radiance observation |
real(r_single) |
-999.0 |
lnsigcutoffsatsh |
length for vertical localization in ln(p)over south hemisphere for satellite radiance observation |
real(r_single) |
-999.0 |
lnsigcutoffpsnh |
length for vertical localization in ln(p) over north hemisphere for surface pressure observation |
real(r_single) |
-999.0 |
lnsigcutoffpstr |
length for vertical localization in ln(p) over tropic surface pressure observation |
real(r_single) |
-999.0 |
lnsigcutoffpssh |
length for vertical localization in ln(p) over south hemisphere for surface pressure observation |
real(r_single) |
-999.0 |
Variable Name |
Description |
Data Type |
Default |
analpertwtnh |
adaptive posterior inflation parameter over north hemisphere: 1 means inflate all the way back to prior spread |
real(r_single) |
0.0 |
analpertwtsh |
adaptive posterior inflation parameter over tropic: 1 means inflate all the way back to prior spread |
real(r_single) |
0.0 |
analpertwttr |
adaptive posterior inflation parameter over south hemisphere: 1 means inflate all the way back to prior spread |
real(r_single) |
0.0 |
sprd_tol |
tolerance for background check: observations are not used if they are more than sqrt(S+R) from mean, S is ensemble variance, R is observation error variance. |
real(r_single) |
9.9e31 |
nlevs |
total number of levels |
integer |
0 |
nanals |
number of ensemble members |
integer |
0 |
nvars |
number of 3d variables to update. For hydrostatic models, typically 5 (u,v,T,q,ozone). |
integer |
5 |
saterrfact |
factor to multiply sat radiance errors |
real(r_single) |
1.0 |
univaroz |
If true, ozone observations only affect ozone |
logical |
true |
regional |
If true, analysis is for regional |
logical |
false |
use_gfs_nemsio |
If true, GFS background is in NEMS format |
logical |
false |
paoverpb_thresh |
if obs space posterior variance divided by prior variance less than this value, observation is skipped during serial processing. 1.0 = don’t skip any obs |
(r_single) |
1.0 |
latbound |
definition of tropics and mid-latitudes (for inflation). |
real(r_single) |
25.0 |
delat |
width of transition zone |
real(r_single) |
10.0 |
pseudo_rh |
use ’pseudo-rh’ analysis variable, as in GSI |
logical |
false |
numiter |
number of times to iterate state/bias correction update. (only relevant when satellite radiances assimilated, i.e. nobs_sat>0) |
integer |
1.0 |
Variable Name |
Description |
Data Type |
Default |
biasvar |
background error variance for rad bias coeffs (used in radbias.f90). Default is (old) GSI value. if negative, bias coeff error |
|
|
|
variace is set to - biasvar/N, where N is number of obs per instrument/channel. if newpc4pred is .true., biasvar is not used - |
|
|
|
the estimated analysis error variance from the previous cycle is used instead (same as in the GSI). |
real(r_single) |
0.1 |
lupd_satbiasc |
if performing satellite bias correction update |
logical |
true |
cliptracers |
if true, tracers are clipped to zero when read in, and just before they are written out. |
logical |
true |
simple_partition |
partition obs for enkf using Graham’s rule |
logical |
true |
adp_anglebc |
turn off or on the variational radiance angle bias correction |
logical |
false |
angord |
order of polynomial for angle bias correction |
Integer |
0 |
newpc4pred |
controls preconditioning due to sat-bias correction term |
logical |
|
nmmb |
If true, ensemble forecast is NMMB |
logical |
false |
iau |
|
logical |
false |
nhr_anal |
background forecast time for analysis |
integer |
6 |
letkf_flag |
If true, do LETKF |
logical |
false |
boxsize |
Observation box size for LETKF (deg) |
real(r_single) |
90.0 |
massbal_adjust |
mass balance adjustment for GFS |
logical |
false |
use_edges |
logical to use data on scan edges (.true.=to use) |
logical |
true |
emiss_bc |
If true, turn on emissivity bias correction |
logical |
false |
Section nam_wrf
Variable Name |
Description |
Data Type |
Default |
arw |
regional dynamical core ARW |
logical |
false |
nmm |
regional dynamical core NMM |
logical |
true |
doubly_periodic |
|
logical |
true |
Section satobs_enkf
Variable Name |
Description |
Data Type |
Default |
sattypes_rad |
strings describing the satellite data type (which form part of the diag* filename). |
character(len=20) array (nsatmax_rad) |
’”“ |
dsis |
strings corresponding to sattypes_rad which correspond to the names in the NCEP global_satinfo file. |
character(len=20) array (nsatmax_rad) |
“” |
Section ozobs_enkf
Variable Name |
Description |
Data Type |
Default |
sattypes_oz |
strings describing the ozone satellite data type (which form part of the diag* filename) |
character(len=20) array (nsatmax_oz) |
“” |