identification of snow and rain at the surface using

Identification of Snow and Rain at the Surface using Polarimetric Radar

Martin Hagen and Alice Dalphinet*

DLR Institut für Physik der AtmosphäreDLR Institut für Physik der AtmosphäreOberpfaffenhofen, Germany

* MétéoFrance, Villeneuve d'Ascq, France

Hydrometeor Classification

Why again (and again)?

• Radars use different wavelengths• Radar operators have different preferences• Radars can provide different sets of radar products• Radars provide radar products with different quality• Radars provide radar products with different quality• Hydrometeor classifiers can be combined with additional

products (like temperature profiles, … )• …

� Different membership functions and relative weights areused for fuzzy-logic hydrometeor classifiers

www.DLR.de • Chart 2 > Martin Hagen > ERAD2012 > Toulouse > 25 - 29 June 2012


DLR’s POLDIRAD

uses most times and seldomalternate H V hybrid (STAR)switching mode mode

� rhoHV(1) is lower than rhoHV(0)

� LDR is available


Identification of Rain and Snow at the Surface

Aviation and road authorities want to know:

• when does it start to snow?(when does rain turn in snow?)

• how long does it snow?

• how much snow will fall?

How can we contribute withpolarimetric weather radar?

What additional tools ormeasurements would berequired?


Identification of Rain and Snow at the Surface

Challenges

• What happens between the radar beam and the surface?

1.0°

0.5°

1.5°

950

1500

2050

?Poldirad at 603 m ASLMUC at 50m range445 m ASL

• Light rain and snow are hard to distinguish

• How to identify freezing rain or drizzle?


0 15 30 45 60 75400?

From Classification to Precipitation at Ground

Bright band identification

2nd Hydrometeor classification

1st Hydrometeor classification


2nd Hydrometeor classification

Weather situation classification

Precipitation at ground

Hydrometeor Classification Process

1st Step

fuzzy logic

2nd Step

inputZH LDR ρHV

ZDR VH

output weather not weather

input output

ZH Light raininput output

ZH Clutter2nd Step

fuzzylogic

ZH

ZDR

LDRρHV

BB altitudeIs-it-stratiform?

Light rainModerate rainHeavy rainDry snowWet snowHailWet hailGraupelCrystals

ZH

ZDR

ClutterInsects, birds

Fuzzy-logic Classification

• POLDIRAD can use LDR and ρHV for the identification of melting hydrometeors

• The current classification is optimized for winter conditions.

• 1-D and 2-D membership functions are used:

78

2D Zh - Zdr Parameter Space for Rain

light rain 01.0

mem

bers

hip

func

tion light rain

ρHV Membership for Rain and Snow

• Definition of membership functions and weights are empirical


-101234567

10 0 10 20 30 40 50 60Z

dr (

dB)

Zh (dBZ)

light rain 0light rain 1moderate rain 0moderate rain 1heavy rain 0heavy rain 1

0.80 0.84 0.88 0.92 0.96 1.000.0

0.2

0.4

0.6

0.8

mem

bers

hip

func

tion

ρHV

light rainmoderate rainheavy raindry snowwet snowcrystals

Bright Band Identification

Modelized rain / dry snow limit (MRSL)

Estimation of MRSL for volume scan sectors

drysnow

freezinglevel

MRSL

psnow

wet

Classes:wet snowothers


rain/snowlimit 0 1

memb. fnct.

MRSL

prain

wetsnow

rain... ...

... 2.1

2.0

1.91.9

1.92.5 2.6

2.52.6

MRSLheights

Verification of Estimated Bright Band Height

• Micro Rain Radar MRR (located at Lichtenau 27 km SW) provides good indication of bright band height (change of fallspeed)

Fall speed

Reflectivity


Hei

ght (

m)

3000

2500

2000

1500

8 10 12 14 16 18Time (UTC)


• Final classification

• Reflectivity


Weather Situation Classifier

Is it stratiform or convective?

• Indicator for convective: ZH > 40 dBZ; ZH > 55 dBZ; season

• Use of hydrometeor classification

convective

favouringhydrometeors

hail, graupel

adversehydrometeors

snow Hei

ght

snow

rain


convective

stratiformrainstratiformsnowstratiformfront

hail, graupel

snowrain

snow

snow

snow

hail, graupel

hail, graupelrain

hail, graupel

Hei

ght

rain

snow

snow

snow rain

Hei

ght

Hei

ght

Precipitation at Surface

• For stratiform precipitation events:• ZH < 20 dBZ• Season• Is the melting layer visible; is it at ground?

situationcon-vect.

stratiform

• Without further observations an extrapolation of radar hydrometeor classification towards the ground is limited to weather situations with standard linear temperature profile


situationvect. ML front no ML

hydro-met.

rain, hail, graupel

rain, snow

snow


• Example of a cold front approaching from the north

DWD radar composite Poldirad reflectivity12:30 UTC

Observationsat radar site

www.DLR.de • Chart 14> Martin Hagen > ERAD2012 > Toulouse > 25 - 29 June 2012

13:30 UTC

-8°C

12°C

924hPa

932hPa

∆t = -5°C

∆p = 2.5 2°C


• Example of a cold front approaching from the north

snow atground

Verification using MRR


rain atground

Verification using MRRprecipitation fall velocities

snowrain

13 14 UTC

2500

AGL

0 m

Summary

Characteristics

• Globally robust method

• Correct detection of the non-meteorological echoes

• Acceptable restitution of the

Limits

• Robust detection of snowfalls

• Detection of a front delineating an area of rain and of snow at the ground• Acceptable restitution of the

melting layer

• Quite good detection of the snowfalls

the ground

• Limited precision in the hydrometeors classes

• Validity of the membership functions

Conclusions – Perspectives

• Importance and performance of radar measurements for observation and nowcasting on mesoscale

• The results of the detection of the bright band and snowfall are encouraging

• Up to now only standard linear temperature profiles are coveredcovered

Perspectives / Future work• Improvement of the membership functions and weights

• Include surface temperature and temperature profiles from aircraft observations (AMDAR) or NWP temperature fields

identification of snow and rain at the surface using

Documents