V. Maurer et al.: Evaluation of coupled and uncoupled simulations 
{a) ROAM-NBS - ERAS 
0° 
MAM 
{b) ROAM-NBS - ICON-CLM 
IF 
50°* 
50°N 
AAO 
MAM 
557 
SON 
LOS 
on | 
4A0°K 
A] 
SCM 
0°N 
50°N 
ON 
O°N 
50°N f 
; 50° 
50°} 
NO) 
Figure 4. Mean seasonal surface temperature difference between ROAM-NBS and ERAS5 (a) and ROAM-NBS and ICON-CLM (b) for the 
whole evaluation period (1979-2020). 
a) tas 
275 
= 
125 . 
£ 0.00 Haar} 4 
8 | | 
5 0.25 X 
FT 
* —0,50 | 
-0.75 
1.00 
a 
‘b) tasmin 
27 
0,50 
X25 
0.00 
0.25 - 
:0,50 ] 
-0,75 
1.00 
Anıntı 
'c) tasmax 
0.75 
50 | 
0.25 
0.00 
0.25 
-0.50 
0.75 
1.00 
kl 
_r————— 
ıCON-CLM-UDAG 
a ICON-CLM 
* "5 ROAM-NBS 
' 
Jul} 
A 
ncntt- 
a 
a 
DI 
1a 
Figure 5. Monthly and spatial mean biases of tas, tasmin, and tasmax against E-OBS for ICON-CLM-UDAG, ICON-CLM, and ROAM-NBS. 
averaged over 1979-2020. 
surface cannot explain all differences; otherwise the fluxes 
would be much higher in JJA than in DJF. The temperature 
and humidity of the overlying air masses, as well as the mean 
wind speed, also have an influence. The sfc Wind differences 
are shown as their interpretation is more straightforward than 
‘hat of the momentum flux, which is given as u and v com- 
ponents. The sign of the sfc Wind differences being in agree- 
ment with the flux differences shows that more mixing due to 
nigher surface temperatures does not only result in stronger 
zurbulent heat fluxes but also in more downward mixing of 
norizontal momentum and, thus, in higher near-surface wind 
speeds (and vice versa). In conclusion, SST biases, which are 
already present in NEMO-NBS and introduced into the cou- 
pled simulation, are directly reflected by the turbulent heat 
fluxes, near-surface wind speed and precipitation. However, 
these biases have no systematic impact on the land areas. 
https://doi.org/10.5194/egmd-19-543-2026 
3.2.3 Wind speed 
For a more detailed evaluation of simulated wind speed, 
DWD station observations along the German coast as well as 
the wind measurements at 100m height at the FINO1 plat- 
form were used. The locations of the stations as well as of 
FINO1 are indicated in Fig. 8a. The surface station evalu- 
ation was done for hourly wind speed for the years 2011— 
2020, which was the period with the most common obser- 
vations. FINO1 was evaluated only up to 2010, as the wake 
effect of the wind farms at this location is too large after 2010 
(e.g. Spangehl et al., 2023). For the surface stations, the nine 
closest grid points were used to capture the varlability due 
to a potential mismatch between the coastline in the model 
and in the real world as well as unresolved geographical fea- 
tures. For FINO1, however, only the closest grid point was 
selected as the platform is located in the North Sea about 
45km away from the German coast and the measurement is 
Geosci. Model Dev... 19. 543578. 2026