Assimilation of stratospheric chemistry observations into the operational NWP model GEM 1
2
2
2
2
S. Chabrillat , C. Charette , M. Charron , J. Degrandpré , P. Gauthier , 2 2 2 2 A. Robichaud , Y. Rochon , Y. Yang and R. Ménard 2
Belgian Institute for Space Aeronomy ,
[email protected]
Fig. 1. Ozone column (D.U.) above South Pole on 2003/10/03
Abstract
GEM with free running chem
TOMS observations
(A07S0302/km2003100312_08p)
(version 8)
O3 analysis increment at level 29
GEM with offline assim of MIPAS O3
(A08S0302/rebm2003100312, p = 52 hPa)
(A08S0302/anlm2003100318))
The operational Numerical Weather Prediction (NWP) model GEM now includes the stratosphere (lid at 0.1 hPa, 80 s-p levels) and the stratospheric chemistry module developed for BASCOE. This module can impact the dynamics through ozone and water vapor radiative effects. The 3D-VAR module has been extended to allow incremental assimilation of chemistry observations. This is (one of?) the first operational system to assimilate chemistry and dynamics in a fully coupled way. Here we assimilate MIPAS O3 and CH4 observations during Fall 2003 with dynamical fields overwritten every 6h from an offline assimilation of meteorology. The results are compared with previous experiments where no assimilation is performed, or where dynamics only is assimilated (free-running chemistry). We validate with observations from TOMS (fig.1), MIPAS and two other assimilation systems (BASCOE and ECMWF, fig. 2). The final experiment, with simultaneous assimilation of dynamics and all MIPAS chemistry observations, remains to be done.
Validation against MIPAS (2003/09/01-04 is before ozone hole)
Ozone Results very similar to BASCOE analyses (cyan): model has less ozone than MIPAS ESA (patly due to obs, cf ASSET), assim corrects this below 10hPa but not above. Tropics: the “wiggles” issue (see below) can be seen in red stdev. South Pole: excellent match.
Fig. 2. Averaged vertical profiles: comparison of MIPAS with GEM, BASCOE and ECMWF Tropics (30°S-30°N)
South Pole (90°S-90°N) -1
-1
10
10
Temperature 0
0
Pressure [hPa]
1
1
10
10
2
2
10
10
3
10
Pressure [hPa]
10
10
3
190
200
210
220
230 T (K)
240
250
260
270
-8
-6
-4
-2 0 O-A for T (K)
2
4
6
-10
-5
0
5
10
15
180
200
220
240 T (K)
O-A for T (K)
260
280
300
10
-1
10
-1
10
Ozone 0
10
Pressure [hPa]
Temperature Assimilation of TOVS radiances corrects well the warm bias of GEM model (grey) in strato. Analyses have +/- same quality as ECMWF (pink). Problem: bias “zig-zags” with altitude. Note: MIPAS temperatures are not assimilated yet.
Environment Canada
0
10
1
1
10
10
Pressure [hPa]
1
2
2
10
10
3
10
0
0.1
0.2
0.3
0.4
0.5 O3 (vmr)
0.6
0.7
0.8
0.9
1
-1
-0.5
0 0.5 O-A for O3 (vmr)
-5
x 10
1
3
1.5
-2
-6
-1.5
-1
-0.5
0 0.5 O-A for O3 (vmr)
x 10
1
1.5
2
0
1
2
3
4 O3 (vmr)
-6
x 10
5
6
7
8
10
-6
x 10
-1
10
-1
10
Methane 0
1
10
Pressure [hPa]
10
1
10
2
2
10
Water vapor Model performs quite well below 1hPa,assim should be easy.
NO2
0
10
Pressure [hPa]
Methane Assim corrects well the bias, we get results similar to BASCOE
10
Average from 2003/09/01 to 2003/09/04 (1100 profiles in tropics, 600 around South Pole) 3
10
3
0
0.5
1
1.5 CH4 (vmr)
2
2.5
3
-2
-1.5
-1
-0.5
-6
x 10
0 0.5 O-A for CH4 (vmr)
1
1.5
2
0
-7
x 10
0.2
0.4
0.6
0.8
MIPAS obs. (ESA offline v4.61)
1 NO2 (vmr)
1.2
1.4
1.6
1.8
2
10
-8
x 10
standard deviations -1
10
Water vapor
GEM, free dyn & free chem 1-month dyn & chem forecast (FMR21I01)
0
10
Pressure [hPa]
NO2 above South Pole Downdraft in polar night vortex => very high NO2 in strato due to NO productions specific to MLT. These are not modelled => model severely underestimates NO2 (QC in BASCOE
GEM, assim dyn & free chem 3D-FGAT dyn assim, 1-month chem forecast (V3ES0307) 1
10
GEM, assim dyn & assim O3 , CH4 6h refresh by trial dyn fields, 3D-VAR assim of MIPAS O and CH (A08S0302) 3
4
ECMWF operational analyses 4D-VAR dyn assim, free O and H O with param chem
2
3
10
2
BASCOE assim chem by CTM 4D-VAR assim of MIPAS into ECMWF-driven CTM, see ASSET paper (Geer et al., 2006) (v3q33) 3
2
3
4
5 H2O (vmr)
6
7
8
-2
-6
x 10
-1.5
-1
-0.5
0 0.5 O-A for H2O (vmr)
1
1.5
2 -6
x 10
Fig. 3. Zonally averaged ozone on 2003/09/04
Standing issues Horizontal oscillation of ozone (”wiggles”) in tropical lower strato See fig.3. Wiggles are present when advection is driven by assimilated wind fields, but disappear when advection is driven by trial fields (6h forecasts) from an offline assimilation of dynamics. Wiggles contribute to “zig-zags” in vertical profiles of bias with MIPAS (fig. 2). Modifying T stdev (in B) and/or removing p0-T balance operator in tropics could solve this issue (ask Y.R.).
10
Free chemistry with dyn refreshed by trial fields from offline assimilation of dyn
1
10
1
1
1
1
1 1
10
3
4
7 10
9
8
3
4
4
9
7
3 4
3
10
4
5
8
6 4
5
4 1
2
2
10
6
5
2
-40
8
0 Latitude [deg]
20
10
6
7
80 V3ES0307
3
2
10
10
4
7
6
4
5
6
10
4
60
5
8
3
4
2
40
6
3
9
5
1
-20
7
2
4
4
Min: 0.02 ; Max: 9.8 ; Contour from 0 to 9.5 by 0.5
-60
3
4
5
3
1
-80
3
5 5
6 7
8
2
2
4
5
6 7
8
10
3
3
4
6
2
2
2
2
2
5
Pressure [Pa]
Horizontal length of correlations in B and vortex edge See fig.1. Horizontal correlations in B have a circular shape with default length = 600 km. When the observation is close to the polar vortex, the analysis increment runs over the edge and injects ozone into the vortex. This is why a freerunning simulation delivers better (more) ozone depletion than the MIPAS analysis.
Free chemistry with online assimilation of dynamics
Pressure [Pa]
10
2
9 8 6 3
4
7
5
3
4
3
2
1
1
1
5
-80
-60
-40
-20
0 Latitude [deg]
Environment Canada - Environnement Canada
20
40
60
80 A08S0302
