Predictability of precipitation
determined by
convection-permitting ensemble modeling
Christian Keil and George C .Craig
Meteorologisches Institut, Ludwig-Maximilians-Universität, München
Institut für Physik der Atmosphäre
Motivation
1.Predictability, or forecast uncertainty, of convective precipitation is
influenced by all scales, but in different ways in different meteorological
situations.
2.Forced-frontal convective precipitation associated with synoptic
weather patterns may be predictable for several days and is primarily
governed by lateral boundary conditions in limited area models.
3.Single convective cells developing during air-mass convection
situations, which of themselves are predictable only for a matter of
hours, are frequently triggered by local, small-scale processes
enforced e.g. by mountain ridges and are anticipated to react
sensitively to changes in the model physics.
Institut für Physik der Atmosphäre
Tools and Ingredients
1. Forecasts of experimental COSMO-DE-EPS of DWD
2. Period in August 2007 during COPS characterized by
different meteorological situations
3. COPS country
4. Observational data of the
European Radar composite
5. Radar reflectivity
dBZ_850 > 19 dbz @ xx:15 UTC
360 x 360 km2
Institut für Physik der Atmosphäre
COSMO-DE-EPS: a cloud resolving EPS @ DWD
Institut für Physik der Atmosphäre
COSMO-DE-EPS: 4 global models
ECMWF
DWD
NCEP
UKMO
Institut für Physik der Atmosphäre
Institut für Physik der Atmosphäre
maximal turbulent length scale
laminar layer depth
laminar layer depth
cloud cover at saturation
Entrainment rate
COSMO-DE-EPS: 5 physics perturbations
Measures and Methods
hits + false alarms
1. BIAS Score: BIAS=
hits + misses
to indicate whether the forecast system has a tendency to
underforecast (BIAS<1) or overforecast (BIAS>1) events.
2. Concept of convective timescale
to describe two mechanisms for control of convection
3. Displacement and Amplitude Score DAS
a novel spatial verification measure employing an areal image
matcher using classical optical flow technique
Keil, C. and G. C. Craig, 2009: A displacement and amplitude score employing an
optical flow technique. Wea. and Forecasting, 24, 1297-1308.
Institut für Physik der Atmosphäre
BIAS of COSMO-DE-EPS 8 - 13 Aug 2007
8.8.07
9.8.07
10.8.07
11.8.07
12.8.07
13.8.07
COPS IOP15
ECMWF DWD NCEP UKMO
Z > 19 dbz
Institut für Physik der Atmosphäre
Convection on 12 Aug 2007 observed by Radar
Institut für Physik der Atmosphäre
COSMO-DE-EPS on 12 Aug 2007 17:15 UTC
Radar
Institut für Physik der Atmosphäre
COSMO-DE-EPS on 12 Aug 2007 23:15 UTC
Radar
Institut für Physik der Atmosphäre
BIAS of COSMO-DE-EPS on 12 Aug 2007
ECMWF DWD NCEP UKMO
Z > 19 dbz
Institut für Physik der Atmosphäre
2 regimes on 12 Aug
ECMWF DWD NCEP UKMO
• BIAS score allows the
separation between locally
forced precipitation
(triggered situations) and
synoptically forced
precipitation
(equilibrium)
Institut für Physik der Atmosphäre
triggered
equilibrium
Measures and Methods
hits + false alarms
1. BIAS Score: BIAS=
hits + misses
to indicate whether the forecast system has a tendency to
underforecast (BIAS<1) or overforecast (BIAS>1) events.
2. Concept of convective timescale
to describe two mechanisms for control of convection
Institut für Physik der Atmosphäre
Two mechanisms for control of convection
- convection removes CAPE
rapidly in comparison to the rate
it is being generated
Height
1. Dynamical production of
CAPE: Equilibrium
CAPE
2. Local perturbations to
overcome CIN: Triggered
- large amounts of CAPE can
CIN
build up if triggers not present
Temperature
To identify regime, consider
timescale over which convection
removes CAPE
Convective timescale (Done et al. 2006)
CAPE
CAPE
tc =
~
dCAPE/dt
Precip. rate
Institut für Physik der Atmosphäre
Time series of convective timescale
12-20UTC: ~25 hrs
Triggered convection
20-24UTC: ~ 5 hrs
Equilibrium convection
Institut für Physik der Atmosphäre
Convective timescale vs ensemble spread
triggered
equilibrium
convection
Institut für Physik der Atmosphäre
Convective timescale vs ensemble spread
triggered
Institut für Physik der Atmosphäre
equilibrium
Mean bias score vs ensemble spread of all days
Institut für Physik der Atmosphäre
Mean bias score vs ensemble spread of all days
triggered
Institut für Physik der Atmosphäre
equilibrium
Summary
Two predictability regimes can be distinguished
depending on the control of convection as measured by
• mean and spread of BIAS score
• the convective adjustment timescale
tc
1. Triggered convection: forecast sensitive to changes in
the model physics, tc of more than a few hours
2. Equilibrium convection: control of convection by
synoptic forcing determining the creation of instability,
short convective timescale tc
Institut für Physik der Atmosphäre
Questions?
Institut für Physik der Atmosphäre
COSMO-DE-EPS
20 Member Ensemble at horizontal resolution of Δx=2.8km
4 Global Models:
Member 1-5: EZMW
Member 6-10: GME
Member 11-15: NCEP
Member 16-20: UKMO
5 physics perturbations:
Member 1,6,11,16: entrscv=0.002
Member 2,7,12,17: clc_diag=0.5
Member 3,8,13,18: rlam_heat=50
Member 4,9,14,19: rlam_heat=0.1
Member 5,10,15,20: tur_len=150
Institut für Physik der Atmosphäre

Sources of uncertainty examined by high resolution ensemble