INSTYTUT METEOROLOGII I GOSPODARKI WODNEJ
description
Transcript of INSTYTUT METEOROLOGII I GOSPODARKI WODNEJ
INSTYTUT METEOROLOGIII GOSPODARKI WODNEJ
INSTITUTE OF METEOROLOGYAND WATER MANAGEMENT
TITLE : IMPLEMENTATION OF MOSAIC APPROACH IN COSMO AT IMWM
AUTHORS: Grzegorz Duniec, Andrzej Mazur
DATE: 05.09.2010
2
1. Validation of SUBS – done for ver. 4.8 of COSMO MODEL
2. Estimation of the influence of SUBS on model results for different parameterization of physics and numerics
StudiesIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
3
Numerical schemes• Leapfrog: 3–timelevel HE-VI (Horizontally Explicit – Vertically Implicit)
Integration
• Leapfrog: 3–timelevel semi–implicit
• Runge – Kutta: 2–timelevel HE–VI Integration with irunge_kutta=1 (3-order standard scheme)
• Runge – Kutta: 2–timelevel HE–VI Integration with irunge_kutta=2 (variant scheme with Total Variation Diminishing)
Convection schemes• Kain – Fritsch (1992)
• Tiedtke (1989)
• Modified Tiedtke scheme for shallow convection
Tests setupIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
4
Domain (113x101 grid points, 40 levels, resolution 7km, forecast time horizon 3 hours)Tests setup (cntd.)
IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
5
• Orig – original COSMO ver. 4.8
• Ctrl – control (subs-modified code of COSMO ver 4.8 with no account for subs: nsubs=0)
• Twins – subs modified COSMO ver. 4.8 with nsubs=4, low resolution input data only (nsubs=4, identical subpixels)
• Subs – modified COSMO ver. 4.8, nsubs=4, high–resolution data included
Explanation of abbreviationsIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
6
• T2m – air temperature at 2m above groud level
• TD – dew point temperature
• TSO – soil temperature at 0 cm down (surface temp.)
• U10m – zonal wind component, 10m above ground level
• V10m – meridional wind component, 10m above ground level
• Water+ice content of soil layers 1cm down
Resulting fields (selected)IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Dates chosen for tests IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Six different synoptic situations chosen for tests (domain marked with black frame in next six slides)
2009.II.01 (00UTC) low temperature, ground was frozen solid
2009.IV.22 (12 UTC) sunny/fair day
2009.X.16 (00 UTC, 06 UTC) ground covered with snow
2009.XI.04 (12 UTC) windy day with intensive precipitation
2009.XI.21 (06 UTC) dense fog
and data for previous case with correct formula for derivation of mean qvs to mimic flux averaging 2006.VIII.01 (12 UTC)
8low temperature, ground was frozen solid
Synoptic situation – 01 II 2009 IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
9sunny/fair day
Synoptic situation – 22 IV 2009 IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
10ground covered with snow
Synoptic situation – 16 X 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
11windy day with intensive precipitation
Synoptic sytuation – 04 XI 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
12dense fog
Synoptic sytuation – 21 XI 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
MethodologyIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Comparison(for all combinations of
convection and numerical schemes)
• orig vs ctrl
• orig vs subs
• orig vs twins
• twins vs ctrl
• subs vs ctrl
• twins vs subs
Statistics(for all combinations of
convection and numerical schemes)
• correlation
• standard deviation
• covariance
• variance
The ”best” configurations and resultsIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
The „best” configuration (concerning correlation coefficient) was obtained forall combinations – results for water in soil (WSO) for 01 II, 22 IV, 16 X, 21 XI (with and without shallow convection).
correlation coefficient – WSO
leapdef leapdef1 leapsemi leapsemi1 RungeKutta1
RungeKutta2
orig-twins 1 1 1 1 1 1
orig-subs 1 1 1 1 1 1
orig-ctrl 1 1 1 1 1 1
ctrl-twins 1 1 1 1 1 1
ctrl-subs 1 1 1 1 1 1
subs-twins 1 1 1 1 1 1
The ”best” configurations and resultsIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
One of the „best” configuration (concerning correlation coefficient) was obtained forzonal wind component (U10) for 22 IV 2009 (without shallow convection).
correlation coefficient – U10 (without shallow convection)
leapdef leapdef1 leapsemi leapsemi1 RungeKutta1
RungeKutta2
orig-twins 1 1 1 1 1 1
orig-subs 0,999 0,998 0,992 0,991 0,999 0,999
orig-ctrl 1 1 1 1 1 1
ctrl-twins 1 1 1 1 1 1
ctrl-subs 0,999 0,999 0,992 0,991 0,999 0,999
subs-twins 0,999 0,999 0,992 0,991 0,999 0,999
The ”best” configurations and resultsIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
The „best” configuration (concerning correlation coefficient) was obtained fordew point temperature (TD) for 01 II 2009 (with shallow convection).
correlation coefficient – TD (with shallow convection)
leapdef leapsemi RungeKutta1
RungeKutta2
orig-twins 1 1 1 1
orig-subs 0,999 0,998 0,999 0,999
orig-ctrl 1 1 1 1
ctrl-twins 1 1 1 1
ctrl-subs 0,999 0,998 0,999 0,999
subs-twins 0,999 0,998 0,999 0,999
The ”worst” configurations and resultsIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
One of the „worst” configuration (concerning correlation coefficient) was obtained fororig vs. twins, ctrl vs. twins, subs vs. twins for air temperature for 04 XI 2009:
correlation coefficient – T2M (without shallow convection)
leapdef leapsemi RungeKutta1
RungeKutta2
orig-twins 0,026 0,049 0,026 0,026
orig-subs 0,999 0,992 0,999 1
orig-ctrl 1 1 1 1
ctrl-twins 0,026 0,049 0,026 0,026
ctrl-subs 0,999 0,993 1 1
subs-twins 0,026 0,057 0,027 0,028
The ”worst” configurations and resultsIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
One of the „worst” configuration (concerning correlation coefficient) was obtained for orig vs. twins, ctrl vs. twins, subs vs. twins for air temperature for 04 XI 2009:
correlation coefficient – T2M (with shallow convection)
leapdef leapsemi RungeKutta1
RungeKutta2
orig-twins 0,020 0,037 0,024 0,024
orig-subs 1 0,997 1 1
orig-ctrl 1 1 1 1
ctrl-twins 0,020 0,037 0,024 0,024
ctrl-subs 1 0,997 1 1
subs-twins 0,020 0,044 0,026 0,026
The ”worst” configurations and resultsIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
The „worst” configuration (concerning correlation coefficient) was obtained forleapsemi – orig, ctrl and subs vs. twins – results for soil temperature for 16 X 2009:
correlation coefficient – TSO (without shallow convection)
leapdef leapdef1 leapsemi leapsemi1 RungeKutta1
RungeKutta2
orig-twins 1 1 0,85 1 1 1
orig-subs 0,998 0,998 0,997 0,997 0,998 0,998
orig-ctrl 1 1 1 1 1 1
ctrl-twins 1 1 0,85 1 1 1
ctrl-subs 0,998 0,998 0,997 0,996 0,998 0,998
subs-twins 0,997 0,998 0,85 0,996 0,998 0,998
Two bugs in cosmo-subs detected (Corrected by Annika Schomburg) IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
1. The major bug is in the computation of the effective qv_s (specific humidity at the surface):
atmh
h
atmisnhs qvK
KqvqvK
qv
,,
atm
h
atmisnhs qv
KqvqvK
qv
,,
wrong formula
correct formula
2. The second error is not important for the model run itself, only for some of the output variables. In the module near_surface.f90 there is a loop over ns=1:nsubs1, which, by mistake, includes some of the output variables like rain_con, which are summed up too many times due to this loop. To correct this bug one can just exchange the single large loop by smaller loops only for those variables where necessary.
Results with wrong formula for 1 VIII 2006 (example)IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
correlation coefficient – zonal wind component
leapdef leapsemi leapsemi1 Runge-Kutta1 Runge-Kutta1.1 RungeKutta2 RungeKutta2.1
orig-twins 0,89 0,89 0,92 0,98 0,91 0,88 0,91
orig-subs 0,88 0,85 0,88 0,41 0,92 0,42 0,92
subs-twins 0,99 0,97 0,98 0,21 0,99 0,21 0,99
Results with correct formula for 1 VIII 2006 (example)IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
correlation coefficient – zonal wind component
leapdef leapsemi leapsemi1 RungeKutta1 RungeKutta1.1 RungeKutta2 RungeKutta2.1
orig-twins 1 1 1 1 1 1 1
orig-subs 0,994 0,979 0,980 0,995 0,996 0,995 0,996
subs-twins 0,994 0,979 0,980 0,995 0,996 0,995 0,996
The ”worst” configurations and results – 16 X 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Soil temperature at 0 cm down (surface temp.)
CORRELATION: 0,85DIFFERENCE BETWEEN ORIG AND TWINS (K)
The ”worst” configurations and results – 16 X 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Soil temperature at 0 cm down (surface temp.)
CORRELATION: 0,85DIFFERENCE BETWEEN SUBS AND TWINS (K)
The ”worst” configurations and results – 16 X 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Soil temperature at 0 cm down (surface temp.)
CORRELATION: 0,85DIFFERENCE BETWEEN CTRL AND TWINS (K)
The ”best” configurations and results – 22 IV 2009 IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Zonal component wind (U10) without shallow convection
CORRELATION: 0,991DIFFERENCE BETWEEN SUBS AND TWINS (K)
The ”best” configurations and results – 01 II 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Dew point temperature (TD) (with shallow convection)
CORRELATION: 0,998DIFFERENCE BETWEEN SUBS AND TWINS (K)
The ”worst” configurations and results – 04 XI 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Air temperature at 2 m (T2M) (without shallow convection)
CORRELATION: 0,026DIFFERENCE BETWEEN SUBS AND TWINS (K)
The ”worst” configurations and results – 04 XI 2009IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Air temperature at 2 m (T2M) (with shallow convection)
CORRELATION: 0,020DIFFERENCE BETWEEN CTRL AND TWINS (K)
Results for 1 VIII 2006 with wrong formulaIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Zonal wind component (U10)
CORRELATION: 0,21DIFFERENCE BETWEEN SUBS AND TWINS (K)
Results for 1 VIII 2006 with correct formulaIMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
Zonal wind component (U10)
CORRELATION: 0,995DIFFERENCE BETWEEN SUBS AND TWINS (K)
Summary
IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
• Extended tests of MOSAIC4COSMO implentation was performed.
• Miscellaneous combinations of various numerical schemes and physical parameterizations were tested for six different synoptic situations
• Except for November 4th, 2009, correlation between results for selected meteorological fields is quite good (from 0.85 to ~1)
• Numerical errors (esp. in data preparation) seem to be a reason for poor correlation for November 4th, 2009
Plans
IMPLEMENTATION OF TILE APPROACH IN COSMO AT IMWM
1. We want to test two new data sets with very interesting meteorological situation(s): - 15 VIII 2008 –very strong thunderstorm,
- May 2010 – very intensive precipitation.
2. We intend to check an influence of subs and twins on: - moisture flux,
- heat flux,
- microstructure of Stratus- and Stratocumulus clouds (e.g. liquid water content or ice content, esp. in winter),
- cloud cover with Stratus- or Stratocumulus clouds,
- precipitation intensity,
- height of cloud base for Stratus- and Stratocumulus clouds,
- microstructure of fog layers,
- structure of boundary layer…
3. If there will be an interest to – maybe to go for tile approach…
THANK YOU FOR YOUR ATTENTIONAUTHORS: Grzegorz Duniec
01-673 Warszawa, ul.: Podleśna 61phone: +48 (22) 56 94 [email protected] Mazur01-673 Warszawa, ul.: Podleśna 61phone: +48 (22) 56 94 [email protected]