Dynamical Downscaling of surface wind circulations in the Northeast of the

Iberian Peninsula

Pedro A. Jiménez (UCM-CIEMAT)J. Fidel González-Rouco (UCM)

Juan P. Montávez (UM)Elena García-Bustamante (UCM-CIEMAT)

J. Navarro (CIEMAT)

1.- Regional VS local evaluation2.- Wind variability: Regionalization Group together those observational sites

with similar temporal variability3.- Dynamical downscaling 3.1.- Evaluation

3.2.- Inference analysis

Outline

Motivation:

- Analyze the surface wind variability in a complex terrain region

- Evaluate the capability of a dynamical downscaling to

reproduce it

1.- Regional VS local evaluation

The present evaluation uses the Reid and Turner concept of regional evaluation to analyze the accuracy of a dynamical downscaling in reproducing the surface wind variability.

The evaluation of numerical simulations usually compares observations and simulations at the nearest grid point

Cox et al.

1998Hanna and Yang

2001Buckley 2004

Reid and Turner

2001

Problematic: 1.- Local effects in observations 2.- Representativeness errors in simulations

Observational dataWind speed and wind directionFrom 1992 to 200510 minutes resolution10 above ground levelQuality control

Ebro

valley

PyreneesCantabrian

mountains

Iberian

System

Daily averages

Ebro valley

(Broad valley)

2.- Regionalization

Jimenez et al., JAMC; 47,

2008

Methodology:Rotation of the EOF

Northern valleys(Narrow valleys)

Mountain stations

North-to-south oriented stations

3.- Dynamical downscaling

Initial and boundary conditions every 6 hours form the ERA-40 reanalysis

WRF 2.1.2

The whole observational period (1992 to 2005) is simulated at at high horizontal resolution of 2km over the area of study.

Local VS Regional

Ebro valley

(Broad valley)

3.1.- Evaluation

Jimenez et al., JAMC

(submitted)

Observations

Mountain subregion

Regionalization obtained with the spatially masked simulation

3.1.- Evaluation: Regionalization in WRF

Simulation

No

rmal

ized

sp

ectr

al

den

sity

3.2.- Inference analysis: Regionalization obtained with the complete simulation

Conclusions

1.- The regional evaluation seems to provide a more appropriate framework than the traditional comparison with the nearest simulated grid point.

2.- The WRF dynamical downscaling is able to reproduce areas of coherent wind variability.

3.- The simulation can be extended to analyze the wind variability before 1992.

Canonical series 1

Corr.: 0.87

Statistical downscalingStatistical downscaling

Canonical Correlation Analysis: optimal linear

association between large scale circulation and local

wind field

Validation and methodology sensitivity

zonal wind component

Wind climatology reconstruction

Kalnay et al., Bu. Amer. Met. Soc, 21; 1996 Kallberg et al., Rep. Seri. 17, ECMWF; 2004

Luterbacher et al., Cli. Dy., 18; 2002

ERA-40 VS WRF