ecmwf cloud scheme: validation and direction adrian tompkins
DESCRIPTION
ECMWF cloud scheme: Validation and Direction Adrian Tompkins. The MP Question: “What have ECMWF ever done for us”? ECMWF’s minor role in Cloudnet: To provide data and await feedback…? Due to my lack of time, this puts the data in the “slow feedback loop”. Model parametrization. 1. Development. - PowerPoint PPT PresentationTRANSCRIPT
Cloudnet meeting, A.Tompkins, ECMWF 11
ECMWF cloud scheme: Validation and DirectionAdrian Tompkins
The MP Question: “What have ECMWF ever done for us”?ECMWF’s minor role in Cloudnet: To provide data and await
feedback…?Due to my lack of time, this puts the data in the “slow
feedback loop”
Data
Model parametrization
2. Validation1. Development
Cloudnet meeting, A.Tompkins, ECMWF 22
ValidationExample: Validation of model versus Meteosat Brightness
Temperatures“Expensive” (human resources) validation for a fixed period
But what if (validation) >> (model cycle updates) ?
i.e. When results arrive they refer to “old” cycle
70°S 70°S
60°S60°S
50°S 50°S
40°S40°S
30°S 30°S
20°S20°S
10°S 10°S
0°0°
10°N 10°N
20°N20°N
30°N 30°N
40°N40°N
50°N 50°N
60°N60°N
70°N 70°N
60°W 40°W 20°W 0° 20°E 40°E 60°E
METEOSAT 7 First Infrared Band Thursday 14 October 2004 0600UTC
70°S 70°S
60°S60°S
50°S 50°S
40°S40°S
30°S 30°S
20°S20°S
10°S 10°S
0°0°
10°N 10°N
20°N20°N
30°N 30°N
40°N40°N
50°N 50°N
60°N60°N
70°N 70°N
60°W 40°W 20°W 0° 20°E 40°E 60°E
RTTOV generated METEOSAT 7 First Infrared Band (10 bit)Tuesday 12 October 2004 12UTC ECMWF Forecast t+42 VT:Thursday 14 October 2004 06UTC
Courtesy of F. Chevallier
Cloudnet meeting, A.Tompkins, ECMWF 33
Uses of ARM
ARM data has been used as a validation toolCloud cover, Cloud ice retrievals from radar
(Janiskova) Simulated Z (Morcrette)Surface radiative fluxes and liquid water paths (JJM)
2D-Var assimilation of radar data to test future cloudsat use (Bennedetti and Lopez)
SGP data used to validate new turbulence model (Neggers and Koehler)
Cases studies and “one-offs”, no routine use in model cycle development
Cloudnet meeting, A.Tompkins, ECMWF 44
DevelopmentDevelopment can mean “using the data to derive / develop
/ tune a parametrization”e.g. Tompkins and Di Giuseppe use cloudnet data to tune
and test a new SW cloud overlap parametrization for solar zenith angle effects on cloud geometry
0 20 40 60 80 100SZA (deg)
-0.15
-0.10
-0.05
-0.00
0.05
0.10
0.15T
OA
Re
flect
ion
Err
or
New SchemeHogan (2.1km)RandomMaximumMax-Ran
ECMWF SW albedo error with respect to
a TIPA benchmark calculation using
over 100 cloud scenes taken over
Chilbolton
Cloudnet meeting, A.Tompkins, ECMWF 55
0 20 40 60 80 100SZA (deg)
-0.15
-0.10
-0.05
-0.00
0.05
0.10
0.15T
OA
Re
flect
ion
Err
or
New SchemeHogan (2.1km)RandomMaximumMax-Ran
DevelopmentHogan Length-scale tuned to give correct Cloud Cover
over Chilbolton, then used for 600 Palaiseau scenes as independent “test”
Experience: Data extremely easy to useReprocessing of ARM site data extremely welcome!!!
ECMWF SW albedo error with respect to
a benchmark calculation using
over 600 cloud scenes taken over
Palaiseau
Cloudnet meeting, A.Tompkins, ECMWF 66
Development
Can also mean a validation tool fast and efficient enough to be included in parametrization tests
ECMWF: T799 L91 medium-range “scores”RMS, AC of Z,T,U
Parametrization Group: “climate suite”3 member 13 month atmosphere only T159L91Validation seasons against: MODIS, ISCCP, Quikscat,
SSMI, TRMM, GPCP, Xie-Arkin, Da-Silva, CERES, ERBE
For parameters of: LWP, TCWV, TCC, 10m winds, rainfall, TOA radn fluxes, surface heat fluxes
Cloudnet meeting, A.Tompkins, ECMWF 77
Example ISCCP Total cloud cover :model cycle 29r1operational early
2005
35
65
65
65
65
65
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
135°W
135°W 90°W
90°W 45°W
45°W 0°
0° 45°E
45°E 90°E
90°E 135°E
135°E
Total Cloud Cover e llu Septem ber 2000 nm onth=12 nens=3 Global Mean: 60.5 50N-S Mean: 58.2
[percent]
5
20
35
50
65
80
93.74
35
65
65
65
65
65
65
6565
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
135°W
135°W 90°W
90°W 45°W
45°W 0°
0° 45°E
45°E 90°E
90°E 135°E
135°E
Total Cloud Cover ISCCP September 2000 nmonth=12 50N-S Mean: 62.2
[percent]
13.50
20
35
50
65
80
95
95.54
10
60°S60°S
30°S 30°S
0°0°
30°N 30°N
60°N60°N
135°W
135°W 90°W
90°W 45°W
45°W 0°
0° 45°E
45°E 90°E
90°E 135°E
135°E
Difference ellu - ISCCP 50N-S Mean err -4.04 50N-S rms 11.3[percent]
-60
-50
-40
-30
-20
-10
10
20
30
37.88
80 60 40 20 0 -20 -40 -60 -80
latitude (deg)
0
20
40
60
80
Zonal Meanmodel obs
-180 -120 -60 0 60 120 180
longitude (deg)
55
60
65
70
75
Extra-Tropicsmodel obs
-180 -120 -60 0 60 120 180
longitude (deg)
40
50
60
70
Tropicsm odel obs
Issue: Cloudnet in slower feedback
loop, but independent and comprehensive
validation (also over points) extremely
important
Cloudnet meeting, A.Tompkins, ECMWF 88
Validation and “tuning”
Data “error”
Model parametrization
Fast validation =“tuned metric”
Slow validation =“Independent” source
Cloudnet meeting, A.Tompkins, ECMWF 99
ECMWF Validation needs: Ice!
Information from cloudnet regarding glaciated clouds is useful
e.g. First comparison of ice water content comparison with microwave limb sounder (Frank Li et al.)
Cloudnet meeting, A.Tompkins, ECMWF 1010
ECMWF validation needs: Higher order moments
Information on subgridscale variability of ice, liquid and water vapour is paramount to developments of statistical cloud cover schemes
Much emphasis has been placed on this, and the Cloudnet results will be central to efforts at ECMWF…
Cloudnet meeting, A.Tompkins, ECMWF 1111
ECMWF Directions, Short term
Numerics have been revised to reduce sensitivity to vertical resolution (moving from T511L60 to T799L91 soon)
Ice sedimentation now a pure advection termIce-to-Snow autoconversion added to modelSimple diagnostic parametrization to allow supersaturation
with respect to iceFinal testing for implementation early 2006
Cloudnet meeting, A.Tompkins, ECMWF 1212
ECMWF Directions, Medium term
Prognostic ice mass mixing ratio
Prognostic ice number concentration
Prognostic moments of total water, with cloud cover derived from a statistical cloud scheme
Interaction between aerosols and microphysics (GEMS)
Attention to numerics
0.5 x Dust
-30 -20 -10 0 10 20 30 40 50 60-15
-5
5
15
25
35
45
latit
ude 6
66
6
612
12
12
12
24
24
30 3036 3648 48 54
g m-2
06121824303642485460
1.5 x Dust
-30 -20 -10 0 10 20 30 40 50 60-15
-5
5
15
25
35
45
latit
ude 6
6
6
6
6 1212
12
12
12
24
24
2430 3030
3636
g m-2
06121824303642485460
Difference
-30 -20 -10 0 10 20 30 40 50 60longitude
-15
-5
5
15
25
35
45
latit
ude
-10-8-6-6-2
-2
-200
000
0
0 0
00
0 00
g m-2
-10-8-6-4-20246810
0.01 IN factor
Reduction in ice water path in response to 3x dust aerosols
over Africa