Robin Hogan Ewan OConnor Anthony Illingworth Nicolas Gaussiat Malcolm Brooks Cloudnet Evaluating the clouds in European forecast models.

Robin HoganEwan O’Connor

Anthony IllingworthNicolas GaussiatMalcolm Brooks

CloudnetCloudnetEvaluating the clouds in Evaluating the clouds in

European forecast modelsEuropean forecast models

OverviewOverview• Motivation

– Representation of clouds in GCMs

• About the Cloudnet project• Cloud products

– Instrument synergy and target categorization– Cloud fraction– Liquid water content– Ice water content

• Evaluation of models– Long-term means– Skill scores– PDFs

Representation of clouds in Representation of clouds in modelsmodels

• Reality– Structure on all scales– 3D interaction with radiation

• Typical GCM gridbox– Horizontal size: 1-100 km (forecast model)

~300 km (climate model)– Vertical size: ~500 m– Holds cloud fraction & mean water content– Clouds assumed to be horizontally uniform– Cloud phase and particle size are usually

functions of temperature

• How accurate is cloud fraction and water content in models?

Heig

ht

1-300 km

~500 m

Cloud water content in GCMsCloud water content in GCMs

14 global models (AMIP)

90N 80 60 40 20 0 -20 -40 -60 -80 90S

0.05

0.10

0.15

0.20

0.25

Latitude

Ver

tical

ly in

tegr

ated

clo

ud w

ater

(kg

m-2) But all

models tuned to give about the same top-of-atmosphere radiation!

• Water content in models varies by factor of 10!• Current satellites provide information at cloud top• Need instrument with high vertical resolution…

The EU Cloudnet projectThe EU Cloudnet projectApril 2001 – October 2005April 2001 – October 2005

• Aim: to retrieve continuously the crucial cloud parameters for climate and forecast models– Three sites: Chilbolton (GB) Cabauw (NL) and Palaiseau (F)– Soon to include all the US and Tropical ARM sites + Lindenberg

• To evaluate a number of operational models– Met Office (mesoscale and global versions)– ECMWF– Météo-France (Arpege)– KNMI (Racmo and Hirlam)– Swedish RCA model (…Coming soon: German & Canadian

models)

• Crucial aspects– Report retrieval errors and data quality flags– Use common formats based around NetCDF to allow all

algorithms to be applied at all sites and compared to all models

The three Cloudnet sitesThe three Cloudnet sites

• Core instrumentation at each site:– Cloud radar, cloud lidar, microwave radiometers, raingauge

Cabauw, The Netherlands1.2-GHz wind profiler + RASS (KNMI)3.3-GHz FM-CW radar TARA (TUD)35-GHz cloud radar (KNMI)1064/532-nm lidar (RIVM)905 nm lidar ceilometer (KNMI)22-channel MICCY radiometer (Bonn)IR radiometer (KNMI)

Chilbolton, UK3-GHz Doppler/polarisation radar (CAMRa)94-GHz Doppler cloud radar (Galileo)35-GHz Doppler cloud radar (Copernicus)905-nm lidar ceilometer355-nm UV lidar22.2/28.8 GHz dual frequency radiometer

SIRTA, Palaiseau (Paris), France5-GHz Doppler Radar (Ronsard)94-GHz Doppler Radar (Rasta)1064/532 nm polarimetric lidar10.6 µm Scanning Doppler Lidar24/37-GHz radiometer (DRAKKAR)23.8/31.7-GHz radiometer (RESCOM)

Basics of radar and lidarBasics of radar and lidar

Radar/lidar ratio provides information on particle size

Detects cloud base

Penetrates ice cloud

Strong echo from

liquid clouds

Detects cloud top

Radar: Z~D6

Sensitive to large particles (ice, drizzle)

Lidar: ~D2

Sensitive to small particles

(droplets, aerosol)

Level 0-1: observed quantities | Level 2-3: cloud products

The Instrument synergy/The Instrument synergy/Target categorization Target categorization

product product • Makes multi-sensor data much easier to use:

– Combines radar, lidar, model, raingauge and -wave radiometer– Identical formatIdentical format for each site (based around NetCDF)

• Performs common pre-processing tasks:– Interpolation on to the same grid– Ingest model data (many algorithms need temperature & wind)– Correct radar for attenuationattenuation (gas and liquid)

• Provides essential extra information:– Random and systematic measurement errorsmeasurement errors– Instrument sensitivitysensitivity– Categorization of targets: droplets/ice/aerosol/insectsdroplets/ice/aerosol/insects etc. – Data quality flags:Data quality flags: when are the observations unreliable?

Example fromUS ARM site:Need todistinguishinsects fromcloud

Target categorizationTarget categorization

Ice

LiquidRainAerosol Insects

• Combining radar, lidar and model allows the type of cloud (or other target) to be identified

• From this can calculate cloud fraction in each model gridbox

Observations

Met Office

Mesoscale Model

ECMWF

Global Model

Meteo-France

ARPEGE Model

KNMI

RACMO Model

Swedish RCA model

Cloud Cloud fractionfraction

Dual wavelength microwave Dual wavelength microwave radiometerradiometer

• 22 and 28 GHz optical depths sensitive to liquid water path (LWP) and water vapour path (WVP)

– Coefficients assumed constant, calibration drifts significantly

LWP - initialLWP - corrected

22222222 WVPLWP cba 28282828 WVPLWP cba

Lidar observes noliquid cloud in profile

• Improve by adding lidar and model (Gaussiat et al.)– Coefficients calculated from cloud temperature information – Use lidar to recalibrate in clear skies when LWP should be

zero

Liquid water contentLiquid water content• Can’t use radar Z for LWC: often affected by drizzle

– Simple alternative: lidar and radar provide cloud boundaries– Model temperature used to predict “adiabatic” LWC profile– Scale with LWP (entrainment often reduces LWC below

adiabatic)

Radar reflectivity

Liquid water contentRain at ground:

unreliable retrieval

• Ice water content from reflectivity and temperature

• Error in ice water content

• Retrieval flag

Mostly retrieval error

Mostly liquid attenuation correction error

No retrieval: unknown attenuation in rain

Ice waterIce water

Observations

Met Office

Mesoscale Model

ECMWF

Global Model

Meteo-France

ARPEGE Model

KNMI

RACMO Model

Swedish

RCA Model

Cloud fraction - Met Office Cloud fraction - Met Office MesoscaleMesoscaleSample level

3 output

Commonly frequency of occurrence is OK but mean amount when present is wrong.

UM has difficulty

predicting 100% cloud

fraction

LWC - Met Office MesoscaleLWC - Met Office Mesoscale

Frequency of occurrence is again OK but amount when present too low

BL height too low?

Supercooled liquid water occurrence is much too low

(kg m-3) (kg m-3)

IWC - Met Office MesoscaleIWC - Met Office Mesoscale

(kg m-3) (kg m-3)

Mean ice water content somewhat too high

Need to be careful due to radar sensitivity and because retrievals not carried out in rain

Model cloud

Model clear-sky

A: Cloud hit B: False alarm

C: Miss D: Clear-sky hit

Observed cloud Observed clear-sky

Comparison with Met Officemodel over ChilboltonOctober 2003

Contingency tablesContingency tables

Equitable Equitable threat threat scorescore

Change in Météo France cloud scheme April 2003Note that cloud fraction and water content in this model are entirely diagnostic

• Definition: ETS = (A-E)/(A+B+C-E)• 1 = perfect forecast, 0 = random forecast

Skill Skill versus versus heightheight

• Model performance:– ECMWF, RACMO, Met Office models perform similarly– Météo France not so well, much worse before April 2003– Met Office model significantly better for shorter lead time

• Potential for testing:– New model parameterisations– Global versus mesoscale versions of the Met Office model

Other Cloudnet productsOther Cloudnet products

• Radar/lidar drizzle flux and drizzle drop size– Crucial for lifetime of stratocumulus in climate models

• Radar/lidar ice particle size and optical depth• Turbulent kinetic energy dissipation rate• Incorporate US ARM data into Cloudnet analysis

– Agreed recently at new GEWEX working group: will enable these algorithms to be applied in tropical and polar climates

• Lots of work still to do in evaluating models!

Quicklooks and further information may be found at:

www.met.rdg.ac.uk/radar/cloudnetwww.met.rdg.ac.uk/radar/cloudnet