Figure 1
The MetCoOp domain.
Figure 2
Amount of available AMSU-A radiance observations (all channels) over the MetCoOp domain on the 10 February 2021, for each satellite and assimilation window (left panel). Map of available observations at 12 UTC (right panel).
Table 1
AMSU-A and MHS characteristics, with corresponding equivalent AWS channels. Observe that we here use a running number from 1 to 19 for the AWS channel ids, as it is used in RTTOV, see, e.g., the WMO OSCAR database at https://space.oscar.wmo.int/instruments/view/mwr_aws.
| INSTRUMENT | CHANNEL IDX | FREQUENCY (GHz) | SENSITIVITY | AWS (CHANNEL IDX) |
|---|---|---|---|---|
| AMSU-A | 1 | 23.8 | surface | no |
| 2 | 31.4 | surface | no | |
| 3 | 50.3 | surface | yes (1) | |
| 4 | 52.8 | temperature | yes (2) | |
| 5 | 53.596±0.115 | temperature | yes (4) | |
| 6 | 54.4 | temperature | yes (5) | |
| 7 | 54.9 | temperature | yes (6) | |
| 8 | 55.5 | temperature | yes (7) | |
| 9 | ν=57.290 | temperature | yes (8) | |
| 10 | ν±0.217 | temperature | no | |
| 11 | ν±0.322±0.048 | temperature | no | |
| 12 | ν±0.322±0.022 | temperature | no | |
| 13 | ν±0.322±0.010 | temperature | no | |
| 14 | ν±0.322±0.0045 | temperature | no | |
| 15 | 89 | surface | yes (9) | |
| MHS | 1 | 89 | surface | yes (9) |
| 2 | 157 | humidity | eq* (10, 165.5 GHz) | |
| 3 | 183±1 | humidity | eq* (15, 182.311 GHz) | |
| 4 | 183±3 | humidity | eq* (13, 180.311 GHz) | |
| 5 | 190.311 | humidity | eq* (11, 176.311 GHz) |
[i] eq*: equivalent channels to AWS.
Figure 3
AMSU-A and MHS (plain lines) together with AWS (dashed lines) weighting functions for a 64-level reference profile at nadir over sub-Arctic for temperature channels (left panel) and water vapor (right panel) channels.
Table 2
Mean (standard deviation) of atmospheric transmittance at surface for radiances at AMSU-A and MHS window channels (data from 10 February 2021-09 UTC and 10 June 2021-09 UTC).
| AMSU-A CHANNEL 3 | MHS CHANNEL 1 | MHS CHANNEL 2 | |
|---|---|---|---|
| Summer | 0.66 (0.03) | 0.77 (0.06) | 0.46 (0.11) |
| Winter | 0.66 (0.03) | 0.90 (0.02) | 0.84 (0.05) |
Figure 4
Expected geolocation lat/lon shift of observation as seen from the boresight locations on ground for five consecutive scan lines of all four AWS feed horns. Based on real data 4 May 2025, around 22:15 UTC. Scan position 73 (the middle position—not identical to the sub-satellite track) of the first scan line of each feed horn is marked by a large dot.
Figure 5
IFOV representation for a single AMSU-A scanline and using a microwave radiance footprint operator. The retrieved surface emissivity is plotted inside the IFOV, i.e., footprint area indicating the heterogeneous surface conditions over the MetCoOp domain.
Table 3
Experimental design.
| DYNAMIC EMISSIVITY | LOW-PEAKING CHANNELS** | CONDITIONS | |
|---|---|---|---|
| ctl-3D-Var /ctl-4D-Var | No | No | N/A |
| ldyn-3D-Var/ ldyn-4D-Var | Yes | Yes | Land*/sea/sea-ice |
| nolwp-3D-Var/nolwp-4D-Var | Yes | No | N/A |
[i] *: except where orography is higher than 500 m.
**: channel 5 of AMSU-A and MHS.
Figure 6
Default (which would have resulted from the previously used system) land surface emissivity map (left), backup atlas emissivity maps (middle panel) compared to a “snapshot” of the dynamical retrieved emissivity (right panel) at AMSU-A Channel 3 (50 GHz) on 10 February 2021, 09 UTC.
Figure 7
Histograms of land and sea surface emissivity at AMSU-A channel 3 (50 GHz) over (a) snow-covered land surface, (b) snow-free land surfaces, and (c) sea-ice in February 2021. HARMONIE-AROME default emissivity (ctl-4D-Var) in red, dynamical retrievals (ldyn-4D-Var) in blue, and averaged ATLAS emissivities in green.
Figure 8
(a) Sea-ice surface emissivity maps at MHS Channel 1 (89 GHz) on 10 February 2021-09 UTC. HARMONIE-AROME static default emissivity (ctl-4D-Var) on the left, compared to the retrieved emissivity (ldyn-4D-Var) on the right. (b) First-guess departures at MHS channel 5 (190.311 GHz) with HARMONIE-AROME static default emissivity (left panel), and emissivity retrievals (right panel) used as input to RTTOV.
Figure 9
Histograms of observed (OBS) in gray and simulated Brightness Temperature (BT) in over land (snow or snow-free) surfaces of AMSU-A sounding channels using in the ctl-3D-Var in red or ldyn-3D-Var in black configurations. Statistics use data from February 2021 over the MetCoOp domain.
Figure 10
Same as Figure 9 but for MHS sounding channels over sea (sea-ice or open sea) surfaces.
Figure 11
Time series of the “spin-up” process showing the evolution of uncorrected and corrected averaged First-guess departures (upper plot) and the 10 predictor values (lower plot) for AMSU-A channel 5. Values are extracted from the ctl-3D-Var configuration during the spin-up period (gray shaded background) and from ldyn-3D-Var configuration afterwards. Data are for the period from 15 January to 28 February over the MetCoOp domain.
Figure 12
Count and differences of active radiance observations assimilated in both configurations ctl-3D-Var and ctl-4D-Var per AMSU-A channel and per assimilation cycle over the MetCoOp domain on the 1 February 2021.
Figure 13
Frequency histograms of First-guess departures (not corrected from the VARBC) over land surface using all available observations (from 10 to 15 February 2021) at AMSU-A channel 4 for ctl-3D-Var, ctl-4D-Var and ldyn-3D-Var, ldyn-4D-Var over the MetCoOp domain.
Figure 14
Count and differences of active radiance observations assimilated in both configurations ctl-3D-Var and ldyn-3D-Var per channel and per assimilation cycle over the MetCoOp domain on the period from 1 to 28 February 2021. Upper (lower) plots show the results for the AMSU-A (MHS) observations.
Figure 15
Frequency histograms of First-guess departures over land surface using all available observations (over 5 days from 10 to 15 February 2021) at AMSU-A channel 5 over the MetCoOp domain.
Figure 16
Normalized First-guess departures statistics of the 3D-Var (red) and 4D-Var (black) experiments of Temperature – Radiosonde (TEMP T) and Aircraft (AIREP T), u- and v-wind components – Radiosonde (TEMP U/V) and Aircraft (AIREP U/V), IASI long-wave sensitive (IASI LW) and IASI Water Vapor sensitive channels (IASI WV) by pressure levels and AMSU-A and MHS by channel number. For the period from 1 to 28 February 2021 over the MetCoOp domain.
Figure 17
Averaged analysis increments of humidity (upper row, ) and temperature (lower row, ) at model level 60 for the 3D-Var system: ctl-3D-Var (left), ldyn-3D-Var (center), and the difference ldyn-3D-Var – ctl-3D-Var (right) in the period from 1 to 28 February 2021.
Figure 18
Bias and normalized RMSE (Root Mean Square Error) differences with uncertainty estimates of T2m and RH2m for the 3D-Var and 4D-Var experiments during 1–28 February 2021.
Figure 19
Scorecards of RMSE verification of surface and upper-Air variables for 3D-Var (left) and 4D-Var (right) experiments for period from 1 to 28 February 2021. Three sizes of blue upwards-facing triangles show levels of significant improvements of LDYN vs. CTL (99.7 % for the largest triangles, 95 % and 68 % for the smaller triangles). Vice versa for red downwards-facing triangles, which show degradations. See Appendix for the complete description of the variables.
| CATEGORY | VARIABLE | DESCRIPTION |
|---|---|---|
| Accumulated | AccPcpXh | 3, 6, 12-hour accumulated precipitation |
| Precipitation | ||
| Clouds | Cbase | Cloud base height in metres |
| CCtot | Total cloud cover in oktas (eighths of the sky) | |
| Surface | Pmsl | Atmospheric pressure at mean sea level |
| Variables | T2m | 2-meter air temperature |
| Td2m | 2-meter dew point temperature | |
| Q2m | 2-meter specific humidity | |
| RH2m | 2-meter relative humidity | |
| S10m | 10-meter wind speed | |
| D10m | 10-meter wind direction | |
| Upper-air | Z | Geopotential height at 150, 500, 850 hPa |
| Variables | T | Air temperature at 150, 500, 850 hPa |
| RH | Relative humidity at 150, 500, 850 hPa | |
| Q | Specific humidity at 150, 500, 850 hPa | |
| S | Wind speed at 150, 500, 850 hPa |