1. Evaluation of TM5 performance for assimilation purpose
E.N. Koffi, P. Bergamaschi, U. Karsten et al.
222Rn measurements: InGOS station PIs
Experimental and model set ups
Daytime variations of both boundary layer heights and 222Rn concentrations
222Rn concentrations versus boundary layer heights
Vertical gradients of 222Rn concentrations at Cabauw
Evaluation of updated slopes treatment and new TM5 convection scheme
Summary and perspectives

2. Boundary layer heights
Experimental and model set ups
Boundary layer heights
IGRA data
IGRA: Boundary layer heights from IGRA data (Seidel et al., 2012)
TM5 data
TM5 : Boundary layer heights as computed in TM5 default version and at INGOS station (Ric =0.3 and both surface wind and stress considered)
TM5_IGRA: As TM5, but at the IGRA station close to the INGOS station
TM5_INGOS: Boundary layer heights computed by TM5 following INGOS protocol (Ric = 0.25 and both surface wind and stress are set to zero) and at INGOS station
TM5_INGOS_IGRA: As TM5_INGOS, but at the IGRA station close to the INGOS station
INGOS and IGRA stations
◊: INGOS stations
●: IGRA stations
○: INGOS stations measuring
222Rn concentrations

3. Experimental and model set ups
TM5 Simulated 222Rn concentrations
F_CONST_D: Simulations by using constant emission maps and the convection scheme based on Tiedtke (1989)
F_INGOS_D: As F_CONST_D, but using INGOS emission maps
F_INGOS_SL : Model runs by using INGOS emission maps, but using updated slopes treatment in TM5
F_INGOS_CON: Model runs by using INGOS emission maps and the convection scheme as used at ECMWF (Berrisford et al., 2011) 
F_INGOS_SLCON: As for F_INGOS_CON and also by using the updated slopes  
INGOS emissions
Constant emission
INGOS emissions
U. Karsten, INGOS meeting, Budapest October, 2013

4. Midday boundary layer heights
For most of the stations, a good agreement is found between IGRA and modelled boundary layer heights
For most of the stations, the differences between the different modelled boundary layer heights are small

5. Daytime minimum 222Rn concentrations
In general, InGOS emission maps give better agreement between simulations and observations
However, significant differences between simulations and observations are found for some stations
Impact of slopes treatment is small
The new convection scheme gives lower concentrations in winter

6. 222Rn concentrations versus boundary layer heights

7. Versus boundary layer heights
222Rn concentrations
Versus boundary layer heights
Again, good agreement between TM5 and IGRA boundary heights
At several stations, simulated 222Rn decrease faster than observed
Fast decrease of simulated concentrations due to sharp increase of BL. Probably related to low temporal resolution of TM5 meteo (3h/6h).

8. TM5 performance for assimilation purpose
● BLH
◊ F_CONST_D
● F_INGOS_D
∆ F_INGOS_CON
The BLHs are in general well reproduced, but the 222Rn concentrations are not for several stations
Limitations of the new 222Rn emission maps and/or also problems in TM5 to simulate vertical gradients correctly (as we see later on) ?

9. TM5 performance for assimilation purpose
● BLH
◊ F_INGOS_D
● F_INGOS_SL
∆ F_INGOS_SLCON
Impact of slope treatment is small
The new convection scheme with the slope treatment gives lower 222Rn concentrations

10. Vertical gradients in radon concentrations at Cabauw (20 versus 200 m)
TM5 mixes up faster than observations at daytime
In general, larger vertical gradients are found from the observations
The differences are reduced during daytime, but still important

11. Evaluation of TM5 performance for assimilation purpose
Vertical gradients in radon concentrations
Cabauw (20 versus 200 m)
00 UTC
12 UTC
Large vertical gradients are still found from the observations at midday, which are not
modelled

12. Evaluation of TM5 performance for assimilation purpose
Conclusions and perspectives
1) Comparison of TM5 boundary layer heights with IGRA radiosonde data
Daytime BLHs are well reproduced (within 10-20% for most InGOS stations)
2) 222Rn simulations with new InGOS emission maps
Overall, improvement compared to the simulations with constant 222Rn emission maps, however significant differences between simulated and observed daytime minimum values (50% of even more for several stations) are found
At several stations, simulations show decrease of 222Rn faster than observed
Faster decrease of simulated 222Rn gradients compared to measurements at Cabauw (20 m versus 200m)

13. Evaluation of TM5 performance for assimilation purpose
Conclusions and perspectives
2) 222Rn simulations with new InGOS emission maps (follow)
These results point to limitations of :
The new 222Rn emissions, but also the problems for TM5 to simulate vertical gradients correctly
Temporal resolution of TM5 meteorological data, which are currently at 3h/6h
For inversions: potential significant impact of exact assimilation time window
3) Evaluation of updated slopes treatment and new TM5 convection
Relatively small impact of slopes, but significant impact of new convection, leading to significantly lower (about 20%) 222Rn during daytime, especially in winter

14. End