Content page earth

1
Volker Matthias, GKSS Research Center,
Geesthacht, Germany
The influence of MM5 nudging schemes on CMAQ
simulations of benzo(a)pyrene concentrations and
depositions in Europe
2
Outline

• Introduction
• MM5 nudging schemes
• Comparison to measurements
• CMAQ model results for B(a)P and PM10
• Conclusions

3
Why PAHs ?

• PAHs imperil humans and ecosystems
• highly bioaccumulative (food chain)
• persistent in various environmental compartments
  
• significant adverse effects already at low doses
• carcinogenic
• impair immune system
• impair reproduction

B(a)P
Object of international reduction
conventions (Target values for air concentrations
- EU 1 ng/m3 (annual average), UK 0.25 ng/m3)
4
Model domains
30 vertical layers up to 100 hPa
54x54 km2 grid
18x18 km2 grid
5
MM5 parameterizations

• The use of MM5 at GKSS
• Kain Fritsch 2 convective cloud scheme
• MRF (Hong and Pan) PBL scheme
• Reisner 2 microphysics scheme
• Noah LSM
• ERA40 6-hourly boundary conditions
• (FDDA grid nudging with ERA40 fields)

6
Checking meteorological fields
Comparison to ascents from 88 radiosonde station
from the igra data base (Durre et al., 2006). RS
data is interpolated to model levels. Mean
deviation and correlation are calculated for each
profile.
7
Test of nudging schemes

• Cases
• No nudging (no nudging)
• No nudging, periodic restart (periodic)
• Nudge only winds (nudge wind)
• Nudge winds and temperature (nudge uvt)
• Nudge winds, temperature and humidity (nudge all)
• Nudge all, but use no LSM (nudge no LSM)
• Nudge all with variable SST (nudge vsst)
• Use subsets of layers (nudge all, 9L (12 L))

8
Test of nudging schemes temperature
3
5

0
-3
nudge all periodic nudge wind no nudging nudge
uvt nudge no LSM nudge vsst nudge all, 9L (12 L)
root mean square error / K mean difference
/ K
9
Test of nudging schemes rel. humidity
0.4
0.2

-0.2
0
nudge all periodic nudge wind no nudging nudge
uvt nudge no LSM nudge vsst nudge all, 9L (12 L)
root mean square error mean difference
10
Monthly mean of correlation and bias sorted by
station Rel. Humidity
Year 2000, mean correlation (Rel. Hum.)
11
Annual mean of vertical wind profiles
U-wind
V-wind
high temporal correlations at all heights (0.85
- 0.99)
12
Comparison to wind-profiler data (CWINDE) mean
vertical profiles (year 2000)
Pendine/UK
Lindenberg/D
Data from Myles Turp, UK Met Office
13
U-wind spectra from wind profilers vs. model
results
Model/oberservations
Maximum-Entropy-spectra of the U-wind component
14
Effects on B(a)P concentrations
full nudging
15
Effects on B(a)P depositions
full nudging
16
Annual PM10
Time series of daily means at Neuglobsow/Germany,
year 2000
mean bias -7.4 mg/m3 corr coeff 0.52
17
Chemical composition at Melpitz/Germany
Sec. Organic Aer. underestimated, particularly in
summer
18
PM10 frequency distribution function
Neuglobsow / Germany
Payerne/Switzerland

• Frequency distribution of PM10 measurements is
  log-normal.
• Model results reveal the same distribution
  function.

19
Conclusions

• MM5 results were tested with different nudging
  schemes
• Full nudging of T, RH, U and V permits close
  agreement with observations
• Depending on horizontal grid scale, intraday
  variability of wind comp. is not resolved by the
  model
• Deposition patterns can be highly sensitive to
  nudging/no nudging in MM5
• PM10 results imply significant underestimation
  of organic aerosol in summer
• Good results for sec. inorganic aerosols in
  Central Europe