Improving radar Doppler wind information extraction Yong Kheng Goh, Anthony Holt University of Essex, U. K. Günther Haase, Tomas Landelius SMHI, Sweden.

Improving radar Doppler wind information extraction

Yong Kheng Goh, Anthony HoltUniversity of Essex, U. K.

Günther Haase, Tomas Landelius SMHI, Sweden

Radar Observations

● Some radars used in operational forecasting only provide Reflectivity data.

● Some others provide Radial Velocity data from Doppler measurements. They can suffer from velocity ambiguity due to folding.

● In this study we also make use of the reasonably close proximity of two Doppler radars in the Po Valley in Italy.

CARPE DIEM WP2 objectives

● Improve the use of Doppler wind data via:

1) Super-observation product (SMHI)

2) Operational dual-Doppler wind retrieval

(ESSEX)

Dual Doppler Wind Retrieval

Procedure:1) Terrain analysis – establish areas

amenable to dual-Doppler analysis.2) Data gridding – interpolating polar

data into Cartesian data.3) Calculating wind field.4) Verifying wind field (a) by re-constructing PPI and

comparing with original PPI; (b) comparing “along-track” components.

Italian Po Valley

Problems with terrain and radar orientation

Height Profile

Italian terrain

Location of Po Valley study region

Data Gridding● Typical dimension:

– 60 x 60 cells x 4 layers– 0.5 km x 0.5 km x 0.9

km● Search and average

method.● Resource hungry

process.● E.g. 60x60x50x50x4

= 36,000,000 times per data set.

Example of polar to Cartesian conversion

● Data type : Doppler velocity● 60 x 60 grid, lattice length = 0.5 km.

Calculating wind field

● Fundamental equations:

iiti

r

iiiiii

Vwv

wvu

sin

sincoscoscossinˆ)(

vr

0

vvv

tt

velocityterminal velocityradialnet

radar at t measuremenlocity Doppler ve

vector(velocity) wind),,()(

ti

ir

wV

iv

wvuv

Numerical procedures● Iterative method:

– horizontal components

– vertical component

● Boundary conditions: – zero velocity on ground

● Typical convergent factor

11, nnnn wDCvwBAu

dz

dw

z

w

y

v

x

u 1

where0

030.0,,

)1(,,

)(

,,)(

,,)1(

max

kjin

kjin

kjin

kjin

ww

ww

Comparison of reconstructed radial velocity field and radar measurement

measurement Reconstructed

“along-track” componentsv

r1

r2

r12

v . (r12 + r2 – r1) = 0

= v(cal).r12 + vr2 (obs) r2 – vr1 (obs) r1

Typical relative deviation, /(v .r12) < ±1%

Assimilation into NWP models

Quality control(e.g. de-aliasing)

WP2: de-aliasing & super-observations (SMHI)

Radarwinds

VVP profiles andsuper-observations

• Innovation: radar observations are mapped onto the surface of a torus (assuming linear winds)

• Advantage: no need for additional wind data from other instruments or NWP models

• Performance: accurate and robust tool for eliminating multiple folding

• Assimilation: benefits through improved quality of wind profiles and super-observations

De-aliasing algorithm

Vantaa (Finland): 4 December 1999, 12:00 UTC

Wind velocity profiles (VVP)

Wind direction profiles (VVP)

Vantaa (Finland): 4 December 1999, 12:00 UTC

Super-observations

Vantaa (Finland): 4 December 1999, 12:00 UTC

tow

ard

s

a

wa

y

De-aliased Doppler measurements

Gattatico (Italy): 31 July 2003, 17:34 UTC

tow

ard

s

a

wa

y

Summary

● Real time dual Doppler wind retrieval can provide useful 3D wind velocity field information to the weather radar operators.

● De-aliased Doppler winds can be assimilated into NWP models through super-observations.

● To-do:– Comparison with NWP model.– Triple Doppler wind retrieval.