Automated Operational Validation of Meteorological Observations in the Netherlands

Automated Operational Validation of Meteorological Observations in the

Netherlands

Wiel Wauben,KNMI, The Netherlands

Introduction QA/QC chain Measurement system and users Status

Introduction

Automated network for synop and climatological observations.

Data near real-time available to internal and external users every 10-minutes.

Observers at airports only for aeronautical reports, but 12 second wind and RVR data provided continuously.

Automated network requires automated validation in real-time.

QC chain

Sensor

validation

Station

validation

MetNet spatial

validation

Export manual

validation

User

reports

ECMWF HIRLAM

black lists

Pre- and

post calibration

MetNet maintenance

Off-line

On-line

External

Site surveys

& inspection

6 months, technical & station

Calibration period 8-24 months or problems, allowed

range for deviation

Instrument selection

Procedures

Range, jump persistency, basic inter-

relation

Inter-relations, temporal,

spatial

Off-line, daily

Reporting vs sensor errors,

Handling of quality

information

Real-time?

Data flow (MetNet)

ADCM airport airbase

CIBIL central system

KMDS OMWA

real-time database

VIVID

extraction

FTP

server

SIAM

Aviation

Climatological database

Sensor

Platforms RMI

Intranet applications

External clients

MSS

message switch

Lightning & radar

APL

application suite

Sensor 12”10’

Station 12” 30’

National 10’1d

International 1h1d

Internal 1’10’

Sensor 1’5’

Basic assumptions

24*7 considered usefull and reduces manual labour

“No” delay in data flow QC does not change values Result of QC check in binary Q-flag Manual input (link to

technical/environmental changes) Alarm Validation results should be embedded in

QA/QC chain with suitable actions to eliminate causes

Follow up

Overview current QC at various places Details of methodes and usefullness

(number, importance) Optimal location of QC (OMWA, 10min) Q indicators traceable throughout data

flow (sensor-interface BUFR report) Follow up (e.g. single jump in

temperature) User should use data AND quality (mask

applied for the users Start with MetNet but keep general

Ceilometer (NI, QG and statistics)

Ceilometer statistics

Radar versus precipitation gauges

Scatter plot

Daily sums Dependen

t verification since bias is removed

VIMOLA vert. integr. LAM

Quasi geostrofic

P at msl 10m wind currently

short term forcast using hourly data

“any” resolution

indicates suspect P values

Current valiation (daily, non-RT)

Outlook

Make business case for basic 10-min near real-time validation

Investigate other possibilities for temporal, spatial and interrelations in RTV

QC at other NMI’s Start implementation of basic version Allow for extensions/generalisation