| Home | Participants | Model table |
The attached file contains the 42 profiles defined on 43 pressure levels. These levels are those of the RTTOV-5 RTM. It was felt important that transmittances from that model be compared to those of other models because RTTOV-5 is widely used in NWP. The concentrations are provided for 8 gases: H2O, CO2, O3, N2O, CH4, CO, N2 and O2. The concentrations C are given in volume mixing ratio (v/v).
JACOBIANS ARE REQUESTED FOR 5 PROFILES: 6, 18, 19, 30, 31.
a) file format
The profiles can be read using the following FORTRAN code. Note that the skin temperature Ts is set equal to TLEV(43). Ps is also defined. All calculations should be done at nadir with a surface emissivity of 1.00 for HIRS channels and 0.60 for AMSU channels.
NGAS=8
NLEV=43
NPROF=42
DO 1 IPROF=1,NPROF
READ(*,*)
READ(*,*)
DO 2 ILEV=1,NLEV
READ(1,20)ILEV, ZLEV(ILEV), PLEV(ILEV), TLEV(ILEV),
x (CLEV(IGAS,ILEV),IGAS=1,NGAS)
20 FORMAT(I3,F7.2,F9.2, 0PF7.2, 10(:,1PE9.2))
2 CONTINUE
TS(IPROF) = TLEV(NLEV)
PS(IPROF) = PLEV(NLEV)
1 CONTINUE
b) conversion to specific humidity/concentration
To get specific humidity or concentration (kg/kg), one must do the conversion:
Specific concentration (Kg/kg) Qgas = Cgas (Mgas/Mwet) (1)
with
Mwet = (1 - Cw) Mdry + Cw Mw (2)
where Mdry is the molecular weight of dry air: Mdry = 28.964 g/mole. Mgas is the molecular weight of the gas (18 g/mole for H2O = Mw, 44g/mole for CO2, 48 g/mole for O3 etc). Cgas is the gas volume moxing ratio as provided and Cw is the water vapor volume mixing ratio.
APPENDIX 2. PLANCK SUBROUTINES
Here are routines to compute monochromatic equivalent brightness temperature, radiance, or derivative of the Planck function with temperature. For such a monochromatic equivalent treatment of a satellite channel, it is customary to use a "modified " Planck function in which the temperature T is replaced by T'= aT + b. The slope "a" is typically very close to 1.0 and the bias "b" close to zero. A routine provides a and b. By using all the same routine for this conversion, differences arising from slightly varying Planck functions are eliminated.
SUBROUTINE PLCKM(B,ZL,T,slope,xint)
C PLANCK FUNCTION B
C AT WAVELENGTH ZL IN METERS AND TEMPERATURE T IN KELVIN
C B IN WATTS/M2/SR/M
C OUTPUT: B, possibly DB if desired
C INPUTS: ZL, T,SLOPE, XINT
c Basic Planck has slope=1.00 and xint=0.
c Slope SLOPE and intercept XINT derived for specific
c channels (see routine PLKN14) with slope=tcof(ch,1)
c and xint=tcof(ch,2)
c Derivative with T: DB, also available
c (could be activated and added as output)
c
real*8 b,zl,t,slope,xint,c,h,gk,x,ex,db,tf
C
DATA C/2.99792458D8/
DATA H/6.6261762D-34 /
DATA GK/1.380662D-23/
TF=slope*T+xint
X=H*C/(GK*ZL*TF)
EX=EXP(X)
B=2.*H*C**2/(ZL**5*(EX-1.))
C DB gives derivative of B with temperature
C DB=B*H*C*SLOPE/(GK*TF**2*ZL*(EX-1.)) *EX
RETURN
END
SUBROUTINE BRITM(T,R,ZL,slope,xint)
C BRIGHTNESS TEMPERATURE T IN KELVIN
C CORRESPONDING TO RADIANCE R IN W/M2/SR/M
C AT WAVELENGTH ZL IN METERS
C OUTPUT: T (K)
C INPUTS: R, ZL, SLOPE, XINT (see PLKN14 routine)
real*8 T,R,ZL,SLOPE,XINT,C,H,GK,X,Z,ZP
DATA C/2.99792458E8/
DATA H/6.6261762E-34 /
DATA GK/1.380662E-23/
Z=ZL**5*R
ZP=(2.*H*C**2)/Z+1.
X=LOG(ZP)
T=(H*C/(GK*ZL*X)-xint)/slope
RETURN
END
SUBROUTINE PLKN14(tcof,maxchan)
C
*******************************************************************
*
C SLOPE and intercept to get effective temperature for modified Planck
c function for NOAA 14
c First 19 HIRS channels; ignore channels 20-22.
c set MAXCHAN IN MAIN TO 22, AND TCOF(22,2)
c OUTPUT: TCOF :
c NOTE that slope is FIRST parameter, i.e. TCOF(channel, 1) = slope
c NOTE that intercept is SECOND parameter, i.e. TCOF(channel,2) = intercept
c INPUTS: MAXCHAN, NSAT
real*8 tcof(maxchan,2),a(22),b(22)
C** NOAA 14-J, HIRS/2 (JSAT=8), NESS107-B Amend 12- 2-1994
DATA (A(J),J=1,22) /
1 0.002, 0.000, 0.011, 0.001,-0.014, 0.026, 0.019, 0.067,
2 0.050, 0.021, 0.073, 0.284, 0.021, 0.020, 0.024, 0.018,
3 0.026, 0.042, 0.313,3*0./
DATA (B(J),J=1,22) /
1 0.99998,0.99997,0.99994,0.99994,0.99997,0.99989,0.99991,
2 0.99977,0.99980,0.99990,0.99971,0.99931,0.99996,0.99997,
3 0.99998,0.99996,0.99992,0.99993,0.99946,3*1./
do 1 k=1,maxchan
tcof(k,1)=b(k)
tcof(k,2)=a(k)
1 continue
RETURN
END
APPENDIX 3. TECHNICAL DETAILS FOR HIRS
The spectral response functions for NOAA-14 can be found on the web at: //poes.nesdis.noaa.gov You than select NOAA filter functions which leads you to sub-directory "filter_functions.html". You then click on "Noaa-14 filter functions". A window opens, you click for invoice in the form of a zip file : noaa14flt.zip. This file will end up in your home directory. Then you do:unzip noaa14flt.zip and you obtain the final file. This file has 5 columns. The first column is the wavenumber. The last column contains the normalized response functions.You can also ask for the file of centroid wavenumbers. The relevant ones are listed below.
Centroid wavelength (m) and wavenumber(cm-1) of NOAA-14 selected channels.
Gases with non-negligible influence are listed (in principle LBL models should
consider all these gases).
ch meters cm-1 non-negligible gases
2 0.147197E-04 679.36 h2o,co2,o3
5 0.139958E-04 714.50 h2o,co2,o3
9 0.972479E-05 1028.3 h2o,co2,o3
10 0.125622E-04 796.04 h2o,co2,o3
11 0.734754E-05 1361.00 h2o,co2,ch4
12 0.675219E-05 1481.00 h2o,ch4,o2
15 0.447147E-05 2236.40 h2o,co2,n2o,n2
APPENDIX 4. TECHNICAL DETAILS FOR AMSU
For the intercomparison of the NOAA-15 AMSU-A (ch 6, 10, 14) and AMSU-B (ch 18), one shoulduse the channel characteristics as listed in the table below. It is assumed that each channel has a box-car shape. Radiative effects above 0.1 mb should in principle be ignored.
References:
For AMSU-A:
Mo, T., 1996: Prelaunch calibration of the advanced microwave sounding Unit-A for NOAA-K, IEEE Trans. on microwave theory and techniques, 44, 1460-1469.
For AMSU-B:
Saunders, R.W., T.J. Hewison, S.J. Stringer, N.C. Atkinson, 1995: The radiometric characterization of AMSU-B, IEEE Trans. on microwave theory and techniques, 43, 760-771.
Table of channel characteristics for AMSU-A and AMSU-B
Channel # fcent side1 side2 hband
=======================================================
6 54.40 0.0 0.0 190.27
10 57.290344 0.217 0.0 38.29
14 57.290344 0.3222 0.0045 1.465
18 183.31 1.0 0.0 250.
========================================================
fcent = central frequency (GHz)
side1 = first sideband or local oscillator frequency (GHz)
side2 = second sideband or local oscillator frequency (GHz)
hband = half-bandwidth (MHz)
Examples of results for HIRS-9 and HIRS-10 are presented in graphic form using a GIF format. These plots are for the first profile of the list (mean tropical profile). For these plots, 40 levels were used so that the actual absolute values should be similar to the ones based on 43 levels. On channel 9 plot, the T (T label), Q(H2O) (Q label) and O3 (O label) are plotted. The (..S) symbol indicates that the Ts Jacobian was > 0.3. On the channel 10 plot, the T and Q(H2O) Jacobian are shown along with the Ts Jacobian (S symbol). Again results provide the brightness temperature change associated with a 1 K T or Ts increase or with a 10% Q(H2O) or Q(O3) decrease.
APPENDIX 6. SENDING YOUR RESULTS
A) Formal results
For each channel, results should be sent in the format shown below. The participant may choose to send a subset of the channels, but ALL PROFILES for a given channel should be processed (one channel times 42 profiles is the minimum participation). Non-calculated transmittances or Jacobians should be set to 999. Do not send results with default (missing) values for channels not processed at all. Avoid if possible to send multiple realizations of the same model modified in a minor way. Again Jacobians are requested for profiles #6,19,30,31 and 42.
PARAMETER (MAXCH=11, NPROF=42, NLEV=43)
c NCHAN is actual number of channels processed and ILIST gives its index in
c the CHAN array
PARAMETER (NCHAN=11)
CHARACTER*6 CHAN(MAXCH)
DATA CHAN/'HIRS02','HIRS05','HIRS09','HIRS10','HIRS11',
x 'HIRS12', 'HIRS15', 'AMSU06', 'AMSU10', 'AMSU14', 'AMSU18'/
INTEGER ILIST(NCHAN)
c definition of subset of channels processed. Example: if only AMSU, then
c NCHAN=4 and the ILIST is defined by DATA ILIST/8,9,10,11/
DATA ILIST/1,2,3,4,5,6,7,8,9,10,11/
WRITE(*,4)NCHAN,ILIST
4 FORMAT(12I5)
DO 1 ICHAN=1,NCHAN
DO 2 IPROF=1,NPROF
WRITE (*,10) IPROF, CHAN(ILIST(ICHAN))
WRITE (*,11) BT, JTS,JPS
Do 100 I=1,NLEV
c 999. Represent missing values
WRITE(*,12) I, TRALL(I), TRH2O(I), TRO3(I), JT(I), JH2O(I),JO3(I)
100 continue
2 continue
1 continue
10 format('PROFILE NUMBER:',I4,2x,'CHANNEL:',A6)
11 format('BT JTS JPS:',3E15.7)
12 format(I4,2x, 6E15.7)
IPROF is the profile number. BT is the outgoing brightness temperature (K). JTS is the Ts Jacobian (K/K) and JPS is the Ps Jacobian (K/mb). TRALL, TRH2O and TRO3 are respectively the total, water vapor and ozone transmittance from each level to the top level. JT is the temperature Jacobian (K/K), JH2O is the H2O Jacobian (K per 10% decrease in QH2O and JO3 is the ozone Jacobian (K per 10% decrease in QO3) as defined in the main text. The O3 inferences are only requested for HIRS channels 2, 5 and 9.
Results should be sent via email to louis.garand@ec.gc.ca
B) Ancillary information on your RTM
Participants should briefly describe their model and provide if possible a published reference. They should indicate which channels were processed (minimum 1, maximum 11).They should indicate the level of complexity of their model (LBL, narrow band or single band; how many bands were used for each channel?). They should indicate which gases were treated in each band. They should mention if Jacobians were derived by brute force or by an actual gradient or Jacobian routine.
C) First step of the participation process
Eventual participants should indicate by email as early as possible their intention to participate. An email list will be built and you will be informed on issues raised by this intercomparison and on its evolution.
MAKE THIS INTERCOMPARISON A SUCCESS!