Download - II: Progress of EDMF, III: comparison/validation of convection schemes I: some other stuff
II: Progress of EDMF,III: comparison/validation of
convection schemesI: some other stuff
Sander Tijm (HIRLAM)Contributions of: Siebesma, De Rooy, Lenderink,
De Roode, Sass, Calvo, Ivarsson, Bengtsson, Malardel, Rontu
Hirlam fog problem (2006)
Fog improvement Fog problem over the sea Too much fog and too low temperature Especially in spring and summer Runaway effect of cloud top cooling ->
more cloud water -> larger emissivity -> stronger cloud top cooling
Until in equilibrium with surface flux
Fog improvement
Fog improvement Improvement of behaviour through (Bent
Hansen Sass): Reduction of cloud water due to raining out of
fog layer (fall speed of cloud droplets) Reduction of cooling through adjustment of thin
cloud layer emissivity Less cloud top cooling and cloud
water formation Less fog and not so cold
Fog improvement
LW-radiation problem With new surface scheme (quicker
reaction to radiation) LW-radiation very important for winter conditions
Clear sky, cold and dry LW-down too low Surface LW-up too large Too rapid cooling of surface
LW-radiation problem
EDMF (TKE)
De Roode, De Rooy, Lenderink, Siebesma
h (km)
x(km)
0
5
1
Use LES to derive updraft model in clear boundary layer.
0
Updraft at height z composed
of those grid points
that contain the highest p%
of the vertical velocities:
p=1%,3%,5%:
Development
1. EDMF (ECMWF) (Massflux + K-profile)
2. Moist TKE (KNMI)
3. Merge EDMF + TKE
Problems (RICO case)
1. mean state not too bad, but ….
2. Lots of noise
3. results extremely dependent on parameters
4. Unpredictable
Adjustment of EC-EDMF Strip ECMWF EDMF to basics
1. Some recoding + clean-up
2. Get rid off ECMWF tricks (prescribed entrainment, turn off diffusivity in cloud, etc)
3. Use other closure of MF
Modifications in massflux
1. Dry parcel: Reduce initial updraft velocity (reduces
mass flux contribution at surface)
2. Moist parcel: • Replace massflux profile, by linear profile
subcloud + Rooy/Siebesma in cloud• Moist parcel entrains 10-20% less than dry
one (reduces intermittency)
TKE modifications
Add dissipation massflux as source of TKE Do correction of length scale formulation TKE for
transport massflux dry parcel. Do correction of length scale formulation in case of
no shear. Add small backgroud diffussion to avoid instability in
solver. Apply simple cloud fraction formulation
Results Stable results ! Almost no intermittency.
Good results at least for RICO, Dry CBL + FIRE. More cases to test
RICO
RICO
Dry CBL: ED
Dry CBL: MF
Dry CBL
Dry CBL: T-prof
FIRE
FIRE: ED
FIRE: MF
Developments Test more cases + including transition
cases. Put more ECMWF stuff back ? Make cloud mass flux profile more
flexible ?
Validation and intercomparison of convection schemes
HIRLAM: intercomparison Two convection schemes in HIRLAM Been developed next to each other Development resources necessary for
other tasks (mesoscale) Release of Hirlam reference system 7.2 Intercomparison during summer 2007 to
choose between schemes
Intercomparison: setup 8 months in 4 different seasons Two meteorologically different months
per season (e.g. July and August 2006) Special setups (0.05 degrees, 4D-Var,
new surface scheme) Initial conditions from ECMWF analysis Surface analysis
Objective verification Precipitation (30%), Clouds (20%), Synoptic
(20%), Upper air (10%), Special features (10%), Daily Cycle (10%)
8 months: 60% Special cases: 40%
Other features (sophistication physics, documentation, coding standards, future improvements: independent experts)
Objective verification Use contingency tables for precipitation
(threshold) and cloud cover (correct bin) Calculate BIAS, PC, FAR, POFD, ETS,
HKS, ORSS Translate scores to 0-100 scale, e.g.:
100*1/BIAS if BIAS > 1; 100*BIAS if BIAS < 1
Observations-> F 0.25 0.5 1.0 2.0 4.0 8.0 16.0 32.0 64.0 128.0 >128.0 o 62005 523 607 168 104 63 30 8 1 6 6 r 69250 1160 1225 256 153 85 34 6 2 6 6 e 9647 447 524 128 70 45 11 6 0 0 0 c 6553 562 1084 412 202 83 26 8 0 1 0 a 7449 724 1195 375 195 86 33 3 2 3 0 s 5190 555 1312 677 461 164 44 12 2 1 1 t 4779 746 1801 836 513 199 58 6 2 0 0 3656 426 1310 799 799 432 123 23 1 1 1 2610 411 1386 996 1119 516 149 33 3 0 1 2064 232 887 689 926 696 272 56 5 1 2 1340 153 574 595 1026 982 374 70 7 1 3 1124 95 344 344 649 716 384 122 19 2 1 403 47 144 143 317 537 508 181 18 5 1 387 20 63 71 165 292 347 138 19 3 0 62 3 10 15 36 78 136 119 31 4 0 73 3 14 6 26 40 67 63 20 5 0 6 0 0 0 1 4 4 10 6 1 0 4 1 2 2 0 2 6 8 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Objective verification Use RMS and bias for synoptic scores Best scheme gets 100 for certain score Second scheme gets
100*RMS(best)/RMS(worst) Bias: 100*(1-bias/RMS(worst))
Parameter (weight) STRACO KF-RK Objective scores (months, 60%) Precipitation (30%) 52.1 53.5 Clouds (20%) 33.5 37.5 Synops (10%) 92.0 92.4 Synop upper air (10%) 90.9 93.4 Upper air bias (10%) 83.9 85.9 Daily Cycle (10%) - - Special cases (10%) - Small precipitation amounts - Satellite imagery
54.0
-
47.4
- Objective scores (special runs, 40%) Precipitation (30%) 42.8 44.3 Clouds (20%) 34.2 35.4 Synops (10%) 91.6 90.4 Upper air (10%) 92.3 92.8 Other features Sophistication physics +/- + Cost (lower number is more expensive) 100 88 Future improvements + + Code standard + - Documentation +/- +/- EPS ? ?
Validation of convection schemes
Developments in EDMF important for pbl state, transition to deep convection
Compiling dataset to validate shallow convection in mesoscale model output
Some deep convection cases included also Observations include: Cabauw tower,
Radiosonde, MSG, GPS IWV, 10-min syn obs NL, Radar, PBL from ceilometers
Validation archive Archive stored at ECMWF Open for anyone to use Description at:
http://www.knmi.nl/~tijm/HARMONIE_cases.html
Fair weather cumulus
Example: convection dying inland
Daily cycle of convection
Daily cycle of convection
Convection PBL development
Outlook In addition to shallow convection
validation of deep convection: Strength and depth Physics dynamics interaction Impact of parameterisation on resolved deep
convection Impact of initial and boundary conditions Convection over sea (subtle)
Combination of standard observations over NL ideal for validation work