Transcript model_comp_notes

2007 AEW activity in Global models.
- Following Berry, Thorncroft and Hewson (2007; henceforth BTH07) we use
700hPa wind to compute curvature, shear vorticity and Trough/Jet axes in 1x1
degree model analyses and forecasts data from ECMWF, UK Met Office,
MeteoFrance (Arpege) and NCEP (GFS, no forecasts yet)
- Focus on July-September over the African continent. Looking for
characteristics and systematic differences in AEW representations.
- Forecasts are evaluated using the analysis from the same model, as we
don’t know which one is actually correct!
UKMET
Arpege
ECMWF
GFS
Hovmöller spacetime diagrams of
the 700hPa
curvature vorticity,
averaged 5-15N –
showing the
synoptic wave
activity. Only
positive values
shown for clarity.
700hPa curvature vorticity Hovmöller summary.
-All model analyses produce the same number of AEWs (26) that can objectively
tracked in the individual maps, following the conventions and the rules set out in
BTH07.
- Relative activity in each model shows AEW confined west of the Greenwich
Meridian (GM) until mid July, then high activity until mid September. Activity then tails
off with smaller scale AEWs.
- Clear differences in the synoptic character of AEWs on the Hovmollers:
GFS tends to produce lots of localized maxima within the AEWs (blotchiness).
UKMO appears to have the weakest AEWs, with a clear lack of activity east of 10E
compared to the others.
Arpege is similar to UKMO, but with slightly more activity east of 10E
ECMWF is more a blend of the other three, with broad, long-lived AEWs.
- The maps (not shown) indicate the timing at specific locations is essentially the
same for each AEW case (as far as the time resolution allows).
UKMET
Arpege
ECMWF
GFS
700hPa Shear
vorticity = 0
(black contour)
with 700hPa
zonal wind
(coloured),
average 10W10E. Figures
show the
location and
intensity of the
African easterly
jet.
700hPa African easterly Jet summary.
- All models show that the peak AEW activity (mid July- mid September)
coincides with AEJ axis centred near 15N. Seems to oscillate around this position
with a ~2 week frequency (significant?)
-The weak AEW activity to start the season occurs when the AEJ is weaker and
in a variable location.
- The weak AEW at the end of the season occurs after the AEJ has suddenly
shifted south (approximately 10N) and strengthened.
- The AEJ is strongest in the GFS. For most part the AEJ is at the same latitude
in all the analyses, EXCEPT in mid-late September when the GFS is further
south of others.
- The jet analyses are much closer to one another than the AEWs – consistent
with our hypothesis that AEWs are convectively coupled.
GFS
UKMO
ECMWF
MFRANCE
- Maps of our diagnostics show that the trough/jet axes from the different analyses
are generally highly correlated in space and time.
- Largest differences exist to the east of 10E, as suggested by the Hovmöller
diagrams. The UKMO does not appear to be less capable than the others. Looks
like the latitudinal averaging washes out the weaker AEWs in this model.
- Any differences between products do not appear to be systematic.
- All model analyses have some value for the forecaster!
JAS 2007 mean 700hPa shear vorticity.
UKMO
ECMWF
Arpege
GFS
JAS 2007 mean 700hPa curvature vorticity
UKMO
ECMWF
Arpege
GFS
Shear vorticity:
- UKMO has weaker shear vorticity in the region 10-30E.
- All have pronounced localized maxima or couplets near Niamey, Dakar and
Addis.
- Widest curvature vorticity strip is in the GFS analysis, but strongest gradients in
ECMWF/Arpege.
- Very large values in Arpege over Sudan/Ethiopia.
Curvature vorticity:
- Weakest overall in GFS (due to cancellation of +ve and –ve: see timeseries),
strongest in ECMWF.
- In UKMO/Arpege very low 10-30E (consistent with Hovmöller diagrams).
Standard deviation of 700hPa curvature vorticity JAS 2007.
UKMO
ECMWF
Arpege
GFS
Standard deviation of 700hPa curvature vorticity JAS 2007 summary:
- Highest ‘activity’ in GFS due most intense signals (see later), stretching into
middle east.
- ECMWF has core of strongest activity from ~10E westwards with a secondary
peak near Darfur.
- UKMO has weakest activity, starting near 15E.
- Arpege is similar to the GFS, but with a broken ‘storm track’ between the GM
and Darfur.
- UKMO/GFS/ECMWF have a bullseye just downtream of Niamey (2E): is this this
the impact of sounding data?
Curvature vorticity time series analysis.
- Looking at curvature vorticity averaged in 3 boxes within the ‘storm track’:
15W
GM
15E
Time series of curvature vorticity in GM box August 2007.
4
UKMO
ECMWF
GFS
MFRA
Curvature vorticity (*10^-5/s)
3
2
1
0
210
215
220
225
230
235
240
245
-1
-2
Day
- GFS has very large range, variance >3 times that of others
-UKMO, MFrance and ECMWF well correlated in terms of timing and amplitude (r > 0.6 for
all).
-The major peaks tend to coincide in all models and are generally associated with synoptic
AEWs.
Peak 700hPa curvature vorticity values associated only with objectively identified
AEWs in the three boxes:
2.5
UKMO
ECMWF
GFS
MF
Mean Curvature vort *10^-5
2
1.5
1
0.5
0
15W
GM
15E
- In the GFS analysis AEWs are 2x as strong in the east than at the coast.
- In the ECMWF AEWs are strongest in the east but not massively so. UKMO are
strongest in the GM box. Arpege gets progressively more intense as they move
towards the coast.
- Using the maps, not discernable difference in start location between models.
Forecasts: Correlation coefficients of 700hPa curvature vorticity in the three boxes.
0.9
0.9
15W
GM
15E
0.8
15W
0.8
GM
15E
Arpege
0.7
0.6
0.6
Correlation coefficient
Correlation coefficient
UKMO
0.7
0.5
0.4
0.5
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0
d+1
d+2
d+3
d+4
d+5
D+1
D+2
D+3
D+4
0.7
0.9
15W
15W
GM
15E
GM
15E
0.8
0.6
0.7
0.4
0.5
Correlation coefficient
ECMWF
0.5
0.6
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
0
d+1
d+1
d+2
d+3
d+4
-0.1
d+2
d+3
d+4
d+5
d+5
-0.1
Forecasts are ‘useful’ (r>0.5) up to 2 days only.
At medium range (2-4 days) the GM box has the lowest scores. The 15E box tends to be best – does this reflect the amount of model
used in the analysis (due to sparse observations?).
700hPa curvature vorticity Root Mean Square errors.
0.5
0.6
GM
Arpege
15E
0.5
0.4
0.35
0.4
0.3
RMS Error (*10^-6 /s)
RMS error (*10^-5)
UKMO
15W
0.45
0.25
0.2
15W
0.3
GM
15E
0.2
0.15
0.1
0.1
0.05
0
d+1
d+2
d+3
d+4
d+5
0
mean magn
D+1
D+3
D+4
Mean
1.5
0.5
15W
15W
GM
GM
15E
0.45
15E
1.3
0.4
1.1
0.9
0.3
RMS error (*10^-5)
RMS error (*10^-5)
0.35
0.25
0.2
0.7
0.5
0.15
0.3
0.1
0.05
0.1
0
d+1
d+2
d+3
- RMS errors increase with
time. After ~3 days the RMS
error become as large as the
magnitude of the 700hPa
d+4
d+5
mean magn
-0.1
d+1
d+2
d+3
d+4
d+5
mean magn
GFS
ECMWF
D+2
Mean error of 700hPa curvature vorticity forecasts (looking for persistent bias).
0.15
0.15
15W
15W
GM
15E
0.05
0.05
0
d+1
d+2
d+3
d+4
d+5
-0.05
15E
0
d+1
d+2
d+3
d+4
d+5
-0.05
-0.1
-0.1
-0.15
-0.15
-0.2
-0.2
-In UKMO/ECMWF at 15E the
short term forecasts are too
intense and linearly become too
weak. In Arpege are too intense at
medium range (2-3 days)
0.15
0.1
0.05
0
D+1
D+2
D+3
D+4
15W
GM
15E
Arpege
-0.05
-0.1
-0.15
-0.2
- In UKMO forecast at the GM and
15W have a high bias.
- In ECMWF the biases at 15W
and the GM are generally high
(albeit small at the GM), but
switch to low bias at day 5.
ECMWF
0.1
Mean Error (*10^-5)
Mean error (*10^-5)
UKMO
GM
0.1
Summary (so far….)
- Model analysis do not vary much when looking in maps for
existence/location of AEWs (good news!).
- Models do vary in the intensity of the analyzed AEWs by a significant
amount if the GFS analysis is considered. Is this due to differences in
model physics?
- Quite suspicious dipoles/odd jumps appear in the means near reliable
sounding locations (especially Niamey). Suggest need for data impact
analysis.
- Statistically, the forecasts seem poor past ~2 days.
Next Steps:
- Perform a ‘case study’ of a selected period of strong AEW activity
evaluating the ability (and usefulness) of forecast products in an operational
sense.