Vical changes

Study of temporal and scan-angle dependencies of vicarious calibration coefficients

GLI calibration 5th group Murakami, 22 March 2004

We are now investigating temporal and scan-angle dependencies of vicarious calibration coefficients (see a presentation in ADEOS-2 workshop).

Today's recommendation for dark targets is "CH13-19 B-side base coefs considered ver.1 de-striping coefficients",
scantable_gmCb_C1_13sp.txt (reading and applying ways are shown the last part of this page).

Figures 1, 2 and 3 show channel and temporal changes of B-side and A-side vicarious coefficients, and their A/B ratios)

VCOEF_GMCB_C1_13_1.GIF - 27,183BYTES
Figure 1 Mirror-B-side vicarious calibration coefficients basing CH13-19 B-side.

Figure 2 Mirror-A-side vicarious calibration coefficients basing CH13-19B.

Figure 3 A-side by B-side coefficient ratio (upper) and corresponding radiance difference (lower).

DSP_VCOEF_ABDIF_13C1R_4.GIF - 12,812BYTES Figure 4 Average input radiance used for deriving Figures 1 and 2

Difference of A/B radiance is calculated by average input radiance shown in Figure 4. The tendency is just the opposit of results by on-board calibrations (see http://suzaku.eorc.jaxa.jp/GLI/cal/presen/2_tanaka.pdf).

Figure 5 Scan-angle dependencies for each sample date (B-side based on CH13-19B).

Figure 6 Scan-angle dependencies for each sample date (A-side based on CH13-19B).

Figures 5 and 6 show the scan-angle dependencies for each sample date for mirror sides B and A derived basing CH13-19B.
We set a model equation and regression tables for each channel. The equations, tables and application are not decided yet, however, we show tentative ones below for level-2 and other investigations.

L_corr[W/m^2/str/um] = L_org[W/m^2/str/um] * coef
coef = tbl(1,nb) +tbl(2,nb)*phi(p) +tbl(3,nb)*phi(p)^2 +tbl(4,nb)*phi(p)^3
+tbl(5,nb)*jdy +tbl(6,nb)*jdy^2 +tbl(7,nb)*jdy^3
+tbl(8,nb)*phi(p)*jdy

L_corr: corrected radiance for each GLI channel
L_org: original (de-striped) radiance for each GLI channel
coef: vicarious calibration coefficients
jdy: day from ADEOS-2 launch derived by 2) below
phi: scan mirror incident angle derived by 3) below
tbl: Table below read by 1) below
    CH13-19 B-side base coefs: scantable_gmCb_C1_13C1.txt
    CH13-19 A-side base coefs: scantable_gmCb_C1_13C0.txt
    CH13-19 B-side base coefs correctecd by de-striping coefficients:
                                             scantable_gmCb_C1_13sp.txt (<=today's recommendation)
nb: GLI channels (1-25 for CH01-19, 24-29)
p: sample number (cross-track) in level-1A [1-1276] or level-1B [21-1256]

The A-side ones should be used basically considering current definition of de-striping processing.
However, difference between B and A-side coefficients in Figures 5 and 6 show that stray light may influence to the A-side coefficients from green to SWIR, especially from February to April 2003.
The A-side coefficients can be used for the ocean area (or for dark targets, e.g., aerosol over the ocean), however, they cannot used for the bright target, because the stray light may influence to Lt_org as "offset" not as "ratio", and the vicarious coefficients are described as "ratio" around input level over the ocean.

Practically, we resommend one solution for the A-side problem,
using B-side coefficients corrected by de-striping coefficients: scantable_gmCb_C1_13sp.txt (see Figure 7).

Figure 7 Scan-angle dependencies for each sample date (B-side based on CH13-19B correctecd by de-striping coefficients).

For example, nLw estimations at MOBY site are changed by the above coefficient as below.
Original B-side coefficients A-side coefficients B-side coefficients correctecd by de-striping

--------------------------------------
1) Read table
      real*4      vctbl(8,25)
       open(44,file='scantable_gmCb_C1_13C0.txt',err=444,
     & access='sequential',form='formatted',status='old')
       do nb=1,25
        read(44,'(2x,8e16.7)') (vctbl(k,nb),k=1,8)
       enddo
       close(44)
       goto 445
444 do nb=1,25
        vctbl(1,nb)=1.
        do k=2,8
         vctbl(k,nb)=0.
        enddo
       enddo
445 continue

B side: scantable_gmCb_C1_13C1.txt
A side: scantable_gmCb_C1_13C0.txt
B side corr: scantable_gmCb_C1_13sp.txt <- recommended

2) Calculate day from ADEOS-2 launch
e.g.,   Product_Name='A2GL10307110612OD1_PV1B0000000.00'

      data mnl/0,31,59,90,120,151,181,212,243,273,304,334/
       read(Product_Name,'(5x,3i2)') iy,im,id
       jdy=17+(iy-3)*365+mnl(im)+id
    (17: count from 2002/12/14 as 0)

3) Calculate mirror incident angle (phi)
      npl=1276*ndets/12
      do np=1,npl
       scag=17.1534+0.035810*12./ndets*(np-1)
       phi(np)=acos(sin(10.*d2r)*cos(tilt0*d2r)*cos(scag*d2r)
    &          +cos(10.*d2r)*sin(scag*d2r))/d2r
      enddo

ndets=12 for 1km, 48 for 250m
d2r=3.14159/180
tilt0=-15.8, 0, +15.8 degree (=tilt_angle; mechanical angle)

4) Apply
      vnirb/1,2,3,4,6,8,9,11,13,14,15,16,17,18,19,20,21,22,23/
      swirb/1,2,3,4,5,6/
     nbr=vnirb(nb) or swir(nb-23)

       do m=1,nline
        do n=1,1236
         slope=gsys(nbr)   ! or gsys_2km(nbr-4)
         rad0=(idata-iflag)*slope
         np=n+20
         if(nb.gt.23) nbd1=nb-4
         vct1=vctbl(1,nbd1)+vctbl(2,nbd1)*phi(np)
     &                         +vctbl(3,nbd1)*phi(np)**2
     &                         +vctbl(4,nbd1)*phi(np)**3
     &                         +vctbl(5,nbd1)*jdy
     &                         +vctbl(6,nbd1)*jdy**2
     &                         +vctbl(7,nbd1)*jdy**3
     &                         +vctbl(8,nbd1)*phi(np)*jdy
          rad1=vct1*rad0
         endif
        enddo
       enddo

nb:band
idata : 2-byte uint from l1b_ch#_data
iflag: flag field from l1b_ch#_data
----------------------------------------