This project is funded
by the European Union
Evaluation of the ability of progressively finer
MNWP models to reproduce wind regimes
in complex terrain
Mario Hrastinski, Kristian Horvath, Iris Odak,
Stjepan Ivatek-Šahdan and Alica Bajić
Workshop on advances in meso- and micro-meteorology
Jezerčica thermae, Donja Stubica, Croatia
4. November 2014.,
1
Contents
•
Introduction
•
Data and methods
•
Results
•
Conclusion
2
Introduction
8
Most Pag
Sibenik
Osijek
Wind speed (ms-1)
7
6
5
4
3
2
1
J
F
M
A
M
J
J
A
S
O
N
D
Time (months)
Most Pag
Šibenik
Osijek
ms-1
3
Introduction
•
Objectives:
1) to evaluate the overall model performance and to quantify
among different sources of errors
2) to provide the scale dependent measure of model performance
4
Data and methods
ALADIN 8 km:
• 37 levels
• 240 x 216 grid points
• 72-hourly forecast
• 3 hours output
• hydrostatic
• initialized at 00 UTC
DADA 2 km:
• 15 levels
• 450 x 450 grid points
• 72-hourly forecast
• 3 hours output
• hydrostatic
• initialized at 00 UTC
ALADIN 2 km:
•
Covered period: 2010-2012.
•
10 min wind speed time series
•
Modeled series with 1-3 hours output
• 37 levels
• 450 x 450 grid points
• 24-hourly forecast
• 1 hours output
• non-hydrostatic
• Initialized at 06 UTC
5
Data and methods
6
Results – Statistical verification
MBIAS
Most Pag
Osijek
1.6
1.6
1.6
1.2
1.2
1.2
0.8
0.8
0.8
0.4
RMSE
Šibenik
2
4
6
8
10
12
0.4
2
4
6
8
10
12
0.4
4.6
4.6
4.6
3.3
3.3
3.3
2
2
2
0.7
2
4
6
8
10
12
0.7
2
Time (months)
4
6
8
10
12
0.7
AL8
DA2
AL2
2
4
6
8
10
12
2
4
6
8
10
12
Time (months)
Time (months)
Table1. 2010-2012. period average continuous statistical scores at Most Pag, Šibenik and Osijek stations
MBIAS
RMSE
Mpa
Šib
Osi
Mpa
Šib
Osi
AL8
0.78
1.21
1.22
3.16
2.12
1.22
DA2
1.08
1.12
1.26
2.92
1.84
1.31
AL2
1.24
1.10
1.28
3.36
1.90
1.26
7
Results – RMSE decomposition
ALADIN 8 km
ALADIN 8 km
4
RMSE comp.
RMSE comp.
6
4
2
0
-2
-4
J
F
M
A
M
J
J
A
S
O
N
3
2
1
0
-1
-2
D
J
F
M
A
M
DADA 2 km
RMSE comp.
RMSE comp.
2
0
-2
J
F
M
A
M
J
J
A
S
O
N
S
O
N
D
A
S
O
N
D
S
O
N
D
3
2
1
0
-1
-2
D
J
F
M
A
ALADIN 2 km
M
J
J
ALADIN 2 km
6
4
RMSE comp.
RMSE comp.
A
4
4
4
2
0
-2
-4
J
DADA 2 km
6
-4
J
J
F
M
A
M
J
J
A
S
O
N
D
3
2
1
0
-1
-2
J
F
M
A
M
J
J
A
Time (months)
Time (months)
Most Pag
Šibenik
8
Results – Kinetic energy spectra
•
Qualitative tool for model evaluation
•
Large scale spectra ~ k-3, 'turbulent' spectra ~ k-5/3 D mesoscale EF)
•
Effective resolution ≈ 4-5 ∆x
∼ 400 hPa
∼ 600 hPa
∼ 1000 hPa
9
Results – Vorticity and divergence spectra
•
All spectra comparable for scales larger than 200 km
•
Vorticity is more energetic for scales > 200 km
•
Divergence domintes for scales < 100 km
Vorticity
Divergence
10
Results – Power spectral density
•
Spectra of components rather than of wind speed
•
ALADIN 2 km forecasts improve simulation of DIU and SUB motions
Most Pag
Šibenik
Osijek
11
Results – PSD in spectral ranges
Zonal and cross-mountain components
•
Major improvement for DIU and SUB cross-mountain motions
•
Along-mountain SUB motions not adequatly resolved
3
3
2.5
2.5
2
2
DIU/DIU obs
SUB/SUBobs
•
1.5
1
0.5
0
0.1
OBS-co
OBS-ct
AL8-co
AL8-ct
DA2-co
DA2-ct
AL-co
AL-ct
1.5
1
0.5
0.15
0.2
0.25
0.3
0.35
SUB/total
0.4
0.45
0.5
0
0
0.1
0.2
0.3
0.4
0.5
DIU/total
12
Conclusions
•
ALADIN 8 km model is sufficient for wind forecast in continental
part of Croatia
•
In coastal part forecast generally improves with increasing the
model resolution
•
The largest portion of errors can be atributed to phase errors
•
KE spectra follow the k-3 at larger scales in upper troposphere and
flatten towards the surface
•
The most significant increase of accuracy was found for diurnal
periods of motions in cross-mountain direction
13
Apendix – A
DADA 2 km
•
ME builds up faster in the upper
troposphere
•
spin up time is shortest for
ALADIN 2 km model
ALADIN 8 km
ALADIN 2 km
14
Appendix - A
ALADIN 8 km
DADA 2 km
ALADIN 2 km
15
Appendix-B
16
Appendix-C
3.5
2.5
3
2
DIU/DIU obs
SUB/SUBobs
2.5
1.5
1
2
1.5
1
0.5
0.5
0
0.05
0.1
0.15
0.2
0.25
SUB/total
0.3
0.35
0.4
0.45
0
0
0.05
0.1
0.15
0.2
0.25
0.3
DIU/total
17
0.35