COST-TOPROF J-CAL Joint microwave CALibration

experiment

Lindenberg, 25-28 August 2014

Goals of J-CAL• Standards and recommendations for optimum absolute

(LN2) and automatic calibrations of passive microwave radiometers– Radiometrics MP-profiler– HATPRO (RPG)

• Learning by doing! Hands on instruments...• Error characterization of LN2 calibrations

– Repeatability– Drift– Estimation of random error (covariances)– Stability of calibration parameters– comparison with high-quality GRUAN radiosondes

Instruments

• Two Radiometrics profiler(DWD Lindenberg)

• 3 HATPRO instruments ofdifferent series:– G1 (Schneefernerhaus, 2005)– G2 (IMGW Poland, 2008)– G4 (RPG demonstrator, 2014)

• extra radiosondes (in additionto 4 daily sondes)

Instruments

• Two Radiometrics profiler(DWD Lindenberg)

M1: JürgenM2: Nicolae, Marta

• 3 HATPRO instruments ofdifferent series:– G1: Uli– G2: Bernhard– G4: Harald

Phone

Jürgen Güldner DWD Local host +49 157 72959624

Bernhard Pospichal Uni Leipzig Organizer +49 176 25342344

Ulrich Löhnert Uni Cologne +49 179 6753627

Nils Küchler Uni Cologne

Jana Preissler NUI Galway

Dariusz Wilk IMGW

Jerzy Jutarski IMGW

Nicolae Ajtai IMAA

Rémi Guillot Météo France

Krispin Bisek DWD (Local host)

Martha Nelson Radiometrics Manufacturer

Harald Czekala RPG Manufacturer

Participants

Time lineToday (Monday) / tonight

• Set-up of instruments, check if everything is running• Parallel operation of instruments during night

– also in case they are not re-calibrated yet.– Only Vertical sky observation, standard calibration routines (gain

calibration)– G4 azimuth scans at zenith– recording of BT, voltages– No sky-tipping

• Identification of reference instrument (HATPRO G1 from Cologne); LN2 calibration?

• general: At start of additional radiosondes all instruments should look vertically (10, 16 UTC?)

Time lineTomorrow (Tuesday)• if no rain: LN2 calibrations, repeatability (details see next

slide)• Discussion on outcome of LN2 calibration (see later)

Night Tuesday/Wednesday and/or Wednesday/Thursday:• Determination of error covariances by several hours (min. 3)

of hot-load observations (night)– All instruments– record TBs, hot-load temperatures and voltages

• If conditions are good: parallel multi angular observations – comparison of atmospheric inhomogeneities (sky-tipping)

Time line

Wednesday• Long-term stability of calibration parameters in K-band (HATPRO)

– hot-load observations with/without noise (Nils Küchler)

Wednesday discussion• Evaluation of nighttime measurements• First data evaluation, results of campaign• How to proceed? Who is doing which analyses?

First draft of MWR operation guidelines should be presented at next TOPROF workshop (November)

Software („live“)

1. Manufacturers perform one LN2 calibration cycle according to their usual practice (HATPRO G1, reference)

2. Users perform calibration experiments with their instruments

3. Radiometrics and HATPRO sequentially; turn off other instruments

LN2 calibration details

– Repeatability of LN2 calibration (4 cycles with Hatpro G2):• in order to avoid mixing of oxygen into LN2, we

should change LN2 3 times, at the end take „old“ LN2

• constantly measuring physical bath temperature

– Repeatability of LN2 calibration (MWP)• 4 cycles with LN2 tank on top (LN2 change or

not?)

– Stability of closed new LN2 target • (min. 2 cycles with HATPRO G4, also measuring

bath temperature) – maybe also other HATPRO?

– Reference instrument is constantly pointing vertically (G1?)

LN2 calibration details

• At every calibration cycle the following points should be considered:

• Every cycle consists of – 2min zenith view, 2min hot load view– 2min view on cold load, also measuring physical LN2

temperature – LN2 calibration (according to standard software

procedure)– 2min view on cold load, also measuring physical LN2

temperature – 2min zenith view, 2min hot load view

• Issues: – Influence Radiometrics/RPG? – rain?

LN2 calibration details

• All operators, please in detail note what experiments were carried out and when– time synchronization??– cell phone camera pictures (instrument and sky!)– note exact times of mdfs and mbfs run– save mdfs and mbfs!– ...

• We need a detailed protocol from every MWR operator on what, why and when things have been done in order to evaluate the calibartions

Important: Documentation!

• After calibrations discussion of: – What was carried out successfully, what problems

occurred?– Identification of error sources at LN2 calibration– How to analyze data from today’s LN2

calibrations? Who does what?– Outlook on night/Wednesday experiments– ….

LN2 calibration summary

Analysis and reporting• central storage of J-CAL data (ftp server)

– ftp://gop.meteo.uni-koeln.de/pub/upload– HATPRO: BRT, LV0, HKD, cal.log, abscal.his, mdf files– Radiometrics: csv + level0 (vlt), level1 (TB), file description– radiosondes– photos, measurement protocols– please use tar and zipped file for every instrument

• time line– raw data and protocol upload until Sept. 4– data analysis until end of October– presentation of main results in Roskilde (TOPROF MC/WG, Nov. 18-20)

Analysis and reportingfor the following build teams of two? (one from each MWR man.)• Repeatability analysis (Nils, Marta)

– after each calibration calculate from zenith views (2min) and all channels Δ( TBinst-TBref) MEAN and STD

– use closest channel in frequency from reference instrument– plot as a function of frequency and calibration number carried out– calculate of Δ (MEAN( TBinst-TBref)) measure of repeatability – differentiate between cases with “fresh” LN2 and “refilled” LN2 and

refer to the measured LN2 temperatures

Analysis and reportingfor the following build teams of two? (one from each MWR man.)• Covariance analysis from HL views (Bernhard, Nicolae)

– plot time series of THL and smooth if necessary

– between gain calibrations (5 min): subtract (linear) trend from TBHL,

ichannel in order to account for systematic differences

– analyze time series of (THL,smoothed-TBHL,ichannel) for all channels on given TB resolution calculate variances and co-variances

– plot covariances and save them into an ascii file– create short recipe for operationalists– are the achieved results the similar for LN2 views … target temp. and

frequency dependence?

Analysis and reportingfor the following build teams of two? (one from each MWR man.)• comparisons with clear-sky radiosondes (part of absolute error)

(Jürgen, Uli, Bernhard, Harry)– use DWD sondes to calculate TBsim

– intercomparison of MWR TB and TBsim – can we achieve an absolute error characterization?

– …

Analysis and reportingfor the following build teams of two? (one from each MWR man.)• Summary of recommendations & Drafting of report (Bernhard

and Uli)– ...

List of recommendations/questions• Which are optimal conditions for LN2 calibration? • What has to be taken into account when performing a LN2

calibration• How often must LN2 calibrations be performed? • How can the drift between two LN2 calibrations be characterized?

• part of automatic LN2 calibration routine: BTs with view on cold load with old calibration parameters should be stored by software!

• What are the optimal settings for automatic calibrations?• How often should gain calibrations be performed? Depending on

instrument type!• Automatic V-band noise switching calibration? • ...• ...• ...

Name of institution IMGW Istituto di Metodologie per l'Analisi Ambientale - Consiglio Nazionale delle Ricerche (CNR-IMAA)

OMSz (Hungarian Meteorological Service)

National University of Ireland, Galway (NUIG), Centre for Climate and Air Pollution Studies (C-CAPS)

RPGLeibniz Institut für Troposphärenforschung (TROPOS)

Contact person Jerzy Jutarski, Jerzy.Jutarski@imgw.pl

Dr. Fabio Madonna Bálint Varga (varga.b@met.hu)

Jana Preißler (jana.preissler@nuigalway.ie)

Harald Czekala, czekala@radiometer-physics.de

Patric Seifert, Bernhard Pospichal

Where is/are the instrument(s) located?

Wroclaw (51°06’47” N, 16°52’52” E) Roof of the institute Szeged – N 46.256 E

20.090

Mace Head Atmospheric Research Station, Carna, Co. Galway, Ireland, 53.33° N, 9.9° W

Leipzig, N51°21', E12°26;mobile ocean facility² (Polarstern)

Which instrument(s) are operated? RPG HATPRO MP3014 Radiometrics MP-3000A RPG-HATPRO Demo HATPRO unit, and

other acceptance test units RPG HATPRO

Which construction year is the instrument(s)? 2008 2004 (upgraded in 2005) 2010 2008

HATPRO-G4 (LAN, Vaisala WXT-520, improved IRR mounting)

2011, (2006)²

)Is the instrument continuously (24/7) operated?

Yes Yes Yes Yes yes yes, (no)²

Which type(s) of retrieval algorithm is used? (Statistical, neuronal network, combined, other?)

statistical QR, LR (TBL)Neural network, Cloudnet for liquid water

Neural networkStatistical (software version 8.44 provided by RPG)

Only NN from now on Statistical algorithms

How often are gain/noise diode/sky tipping/LN2 calibrations performed?

G: 10min, ND: 60 min, T: 6 hour, LN2: 1/year

6 Months, Calibrated by the manufacturer – 2 years

LN2 only at installation time (May 2010.), TIP calibration 2 years ago

G: 5min (10 s integration), N: 30 min, T: 2hours, LN2 ~6 months

Sky tip tried every 1 hours, during night time only, LN2 every 6 months

G: 10min, ND: 30 min, T: 1hour, LN2 ~6 months

If applicable, which thresholds are used (sky tipping, LN2)?

Sky tipping: chi², lin. corrsky tipping R=0.995, LN2 within 2 % of the previous calibration.

sky tipping: 80K (zenith), 4K (bilateral tip), 0.1 (Chi^2), 0.9999 (correlation)

0.99995 and 0.1, plus 80K zenith threshold for TB, only night time

0.9999 and 0.1, plus 80K zenith threshold for TB, only night time no skytip on ocean

Are any quality control measures taken regularly? (e.g. comparison to radiosondes, models, etc.)

Comparison to radiosondes – 2012/2013

Comparison with GPS, Raman lidar, radiosondes

Radiosonde temp/RH profiles, not regularly No Visual comparison to Essen

radiosonde

radiosonde, not regularlyradiosonde on Polarstern²