Slide1: Review of WVRs in Astronomy Alan Roy MPIfR (Wiedner)


Slide2: The Troposphere as Seen from Orbit Method: Synthetic Aperture Radar (Earth Resources Satellite) Frequency: 9 GHz Region: Groningen Interferograms by differencing images from different days 5 km 5 km Internal waves in a homo- genously cloudy troposphere A frontal zone Convective cells 0 mm -100 mm 100 mm Hanssen (1997)


Slide3: Coherence Loss due to Troposphere Pico Veleta – Onsala baseline Source: BL Lac Frequency: 86 GHz Coherence Function 7 min 360° VLBI phase time series


Slide4: Phase Referencing Errors due Troposphere


Slide5: WVR Performance Requirements Phase Correction Aim: coherence = 0.9 requires   / 20 (0.18 mm rms at  = 3.4 mm) after correction Need: thermal noise  14 mK in 3 s Need: gain stability 3.9 x 10-4 in 300 s Zenith Delay for Phase Referencing Aim: transfer phase over 5o with 0.1 rad error at 43 GHz Need: absolute ZWD with error andlt; 1 mm (?)


Slide6: WVR Performance Requirements Opacity Measurement Aim: correct visibility amplitude to 1 % (1 ) Need: thermal noise  2.7 K Need: absolute calibration  14 % (1 )


Slide7: Phase Correction Methods Use a nearby strong calibrator a) Interleave source and calibrator observations BUT: must cycle fast -andgt; short integrations -andgt; few calibrators strong enough b) Dual beam: observe simultaneously calibrator and source (VERA) BUT: need duplicate moveable receiver c) Dual frequency: observe target source at lower frequency scale up phase to calibrate the higher frequency BUT: scaling up multiplies the phase noise; need very good low-frequency observation d) Paired antennas: one observes target, one observes calibrator (Asaki 1997) Measure the water vapour and infer the phase a) Total power method b) Radiometric phase correction (eg at 22 GHz, 183 GHz or 20 um)


Slide8: Telescope Technique Freq Path Residual / mm dG/G dT in 1 s VLA WLM 22 GHz cooled 0.81 0.6x10-4 (100 s) 20 mK Plateau de Bure WLM 22 GHz uncooled 0.031 7.5x10-4 (30 min) Plateau de Bure TP 230 GHz cooled 0.041 2x10-4 Pico Veleta TP 230 GHz cooled 0.24 OVRO WLM 22 GHz uncooled 0.16 10 mK BIMA TP 90 GHz cooled 0.17 BIMA WLM 22 GHz uncooled 0.1 5x10-3 CSO-JCMT WLM 183 GHz uncooled 0.06 SMA TP 230 GHz cooled 0.09 2x10-4 SMA WLM 183 GHz uncooled ATCA WLM 22 GHz cooled 0.3 12 mK Effelsberg WLM 22 GHz uncooled 0.24 5x10-4 (100 s) 12 mK VLBA TP 86 GHz cooled 0.6 Chatnantor WLM 183 GHz uncooled 0.08 2x10-3 (100s) DSN WLM 22 GHz uncooled 0.21 25 mK (8 s) IRMA WLM 15 THz cooled WVR Phase Correction Performance Comparison = represented at this meeting = lowest rms phase demonstrated


Slide9: Total Power Phase Correction Plateau de Bure Total power at 230 GHz Correction applied to simultaneous 90.6 GHz Bremer 1995, 2000 3 mm 30 min Phase correction Observed phase: rms = 0.623 mm Corrected phase: rms = 0.167 mm


Slide10: Total Power Phase Correction: VLBI demo Pico Veleta - Onsala Total power at 230 GHz Correction applied to simultaneous 86 GHz VLBI Bremer et al. 2000 4.7 mm 6 min Observed phase: rms = 0.71 mm Corrected phase: rms = 0.45 mm Phase correction


Slide11: Owens Valley Radio Observatory (Caltech) (Array before moving to Cedar Flat) Frequencies: 86 - 115 GHz 210 – 270 GHz Antenna diam: 10.4 m Altitude: 1220 m


Slide12: Owens Valley Radio Observatory Woody, Carpenter, Scoville 2000, ASP Conf Ser 217, 317 Uncooled LNA (Tsys = 200 K) Downconvert to 4 GHz to 12 GHz (cheaper components, better characterized) Triplexer separates 2 GHz Bands on line and off-line 18.2 to 20.2, 21.2 to 23.2, 24.2 to 26.2 GHz Analog sum of wing channels for continuum Analog difference of line and continuum channels Alternate L and C every 1.7 ms to 16-bit A/D 363 K load Ambient load Cold load (optional)


Slide13: Owens Valley Radio Observatory Woody et al. (2000) Two levels of Dicke switching reduce effects of gain and offset drifts: 1) PIN-diode attenuators adjust the Line-Continuum output to be zero for blackbody loads; output measures deviation from a flat spectrum. Transfer switch reverses assignment of Line and Continuum to the detectors every 1.7 ms; demodulation is performed in software -andgt; removes DC offsets and most of the gain drifts in detectors and following electronics Results: Allan Variance -andgt; noise in L - C andlt; 10 mK for 20 s to 20 min while noise in L andamp; C andgt; 30 mK -andgt; analog L – C differencing and transfer switch modulation valuable C1 andamp; C2 channels derived from -10 dB coupler have 10x more noise -andgt; standard radiometer noise is not the dominant noise White noise to 1 s in L or C channels separately White noise to 10 s in L-C channel


Slide14: Owens Valley Radio Observatory Woody et al. (2000) Calibration Once per hour hot andamp; ambient load Solve for gain, Tsys, and drift in offset of L-C channel Accuracy of gain determination: 1 % Noise in offset determination: 20 mK


Slide15: Owens Valley Radio Observatory 26 min 3 mm interferometer path at 100 GHz WVR predicted path RMS before correction = 0.53 mm RMS after correction = 0.16 mm Woody et al. (2000)


Slide16: Owens Valley Radio Observatory Woody et al. (2000) Path Length Retrieval Observe a strong calibrator -andgt; conversion factor Typically use a fixed 12 mm/K cf calculated conversion factor of 8 mm/K Difference is 'within the uncertainties of the triplexer bandpass shapes and atmospheric model assumptions'


Slide17: Owens Valley Radio Observatory Woody et al. (2000)


Slide18: Owens Valley Radio Observatory Woody et al. (2000) Transferring phase between calibrator and source: hard! (due to gradient in sky brightness) must normalize gains among the WVRs using the step due to elevation change Average L-C from all WVRs / K L-C from each WVR / K


Slide19: Owens Valley Radio Observatory Woody et al. (2000) 0309+411 at 100 GHz for 5 h Cycle: 6 min source, 6 min calibrator (0.7 degrees away) WVR phase is transferred from calibrator to source Before WVR correction After WVR correction (weather degraded) (good weather) 28 Jy 40 Jy 36 Jy 42 Jy 13 Jy 34 Jy


Slide20: Owens Valley Radio Observatory Woody et al. (2000) Conclusion Can correct tropospheric phase fluctuations down to andlt; 0.2 mm. Allows 3 mm observations in previously unusable weather. Not sufficient for improving images during typical conditions Or for routine use during 1 mm observations. Developing a cooled version to decrease noise to reach 0.05 mm. Staguhn et al. 2001, ASP Conf: First light on prototype Cooled 22 GHz WVR Double sideband heterodyne 0.5 GHz to 4 GHz IF 16 channel analogue lag correlator (APHID) (see Alberto Bolatto’s talk)


Slide21: JCMT – CSO Interferometer Frequencies: 210 – 270 GHz 318 – 360 GHz Higher than OVRO 460 – 500 GHz Antenna diam: 10.4 m andamp; 15 m Altitude: 4092 m Higher than OVRO Location: Hawaii James Clark Maxwell Telescope (JCMT) Caltech Submillimeter Observatory (CSO)


Slide22: JCMT – CSO: 183 GHz WVRs Line pivot points: least sensitive to altitude of water vapour Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036


Slide23: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 The three double-sideband frequency channels of the WLM


Slide24: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 Advantages of 183 GHz over 22 GHz: - line is 10 x stronger than 22 GHz. -andgt; can build uncooled systems - optics are small -andgt; easier to install in existing telescopes Disadvantages of 183 GHz: - line saturates easily -andgt; suitable only for dry sites - retrieval coefficient depends on amount of water vapour and conditions


Slide25: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036


Slide26: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 Calibration - Loads at 30 C and 100 C - Load stability: 10 mK - Flip mirror cycles every 1 s between sky and loads 10 mK 5 min Load temperature vs time Sectioned drawing of load


Slide27: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 Hot load Warm load Mirror 2 Mirror 1 Corrugated horn (facing away)


Slide28: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 Uncooled mixer (Tsys = 2500 K) Coupler Mixer Filter Detector V/F Power splitter 1.2 GHz 4.2 GHz 7.8 GHz Oscillators Gunn oscillator 91.655 GHz 183.31 GHz +/- 8 GHz Double-sideband mixing makes measurement insensitive to filter shape Used coupler + power splitter since no suitable triplexer exists


Slide29: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 A small shift in the centre frequency of a filter makes a big change in the measured brightness temperature since the line is steep. Thus, need filter shape within 5 MHz of spec. No triplexer matched this.


Slide30: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 DSB mixing to baseband folds water line at oscillator frequency Result is flat water line spectrum Water line spectrum is then same as the calibration load spectrum Calibration factor is then independent of the filter shape


Slide31: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 Gain fluctuations of WVR measured against loads each second 9 min 2x10-4 10x10-4 WVR at JCMT WVR at CSO (outside, so less stable environment)


Slide32: JCMT – CSO: 183 GHz WVRs Wiedner 1998 PhD thesis Wiedner, Hills, Carlstrom, Lay 2001, ApJ, 553, 1036 12 min 1.4 mm Maser source MWC 349 at 356 GHz After correction RMS = 48d = 0.11 mm Before correction RMS = 127d = 0.30 mm WVR correction Atmospheric model: transition strengths from Waters (1976), Ben-Reuven line profile, exponential atmosphere, radiative transfer calculation


Slide33: Sub-Millimeter Array 183 GHz WVRs being installed. (Talks by Ross Williamson andamp; Richard Hills) CSO JCMT SMA


Slide34: Very Large Array 22 GHz WVRs being prototyped. (Talk by Walter Brisken) Image courtesy NRAO/AUI Dave Finley


Slide35: Plateau de Bure 22 GHz WVRs in routine operation. (Talks by Michael Bremer andamp; Aris Karastergiou)


Slide36: Effelsberg 22 GHz sweeping WVR operating. (Talk by Alan Roy)


Slide37: Berkeley-Illinois-Maryland Array 22 GHz sweeping WVRs prototyped. Array relocated to Cedar Flat with OVRO antennas Now called CARMA. (Talk by Alberto Bolatto)


Slide38: VLBI Phase Correction Demo Demonstration by Tahmoush andamp; Rogers (2000) 3C 273 Hat Creek – Kitt Peak 86 GHz VLBI 400 s 4 mm path ● RMS phase noise reduced from 0.88 mm to 0.34 mm after correction. ● Coherent SNR rose by 68 %. VLBI phase WVR phase


Slide39: CARMA (Talk by Alberto Bolatto) Jim Stimson Photography


Slide40: Chajnantor Site Testing Delgado et al. 2001, ALMA Memo 361 Two 183 GHz WVRs 300 m apart Duplicates of JCMT-CSO WVRs (Hills/Wiedner) Co-located with two 11.2 GHz seeing interferometers observing a geostationary satellite


Slide41: Chajnantor Site Testing Delgado et al. 2001, ALMA Memo 361 Correlation coefficient between WVR and interferometers varied. Cause: when turbulence is lower than 300 m it lies in near-field of interferometer antennas causing large beam differences between the instruments (?)


Slide42: Chajnantor Site Testing Delgado et al. 2001, ALMA Memo 361


Slide43: Chajnantor Site Testing Delgado et al. 2001, ALMA Memo 361


Slide44: Australia Telescope Compact Array Frequencies: 1.2 - 106 GHz Antenna diam: 22 m Altitude: 300 m


Slide45: ATCA 22 GHz WVR


Slide46: ATCA WVR Frequencies


Slide47: ATCA Phase Correction Demo


Slide48: NASA Deep Space Network 22 GHz to 32 GHz WVR (Tanner et al.) For Cassini gravity wave experiment Naudet et al. (2000)


Slide49: NASA Deep Space Network Need: 10 mK radiometric stability from 100 s to 10000 s Focus: improve precision and stability of noise diode and Dicke switch Methods: 1) Regulate temperature in radiometer box to 1 mK. 2) bought commercial noise diodes. 3) follow instructions to bias with regulated 28 V. -andgt; poor stability: 20 x 10-4 in 10 s - 100 s 4) try current-regulating bias circuit -andgt; immediate improvement to 1 x 10-4 in 100 s, 5 x 10-4 in 1 day 5) replace magic T power combiner with directional couplers due to extreme sensitivity to mismatch (-40 dB reflection caused 4 % change of noise diode power) -andgt; 1 x 10-4 in 1 day 6) regulate the relative humidity -andgt; 0.3 x 10-4 in 1 day 7) Dicke switch using absorber inserted in slotted waveguide by loudspeaker voicecoil Tanner et al. (1998)


Slide50: NASA Deep Space Network RMS before correction = 0.43 mm RMS after correction = 0.1 mm Naudet et al. (2000) 4 h 2 mm


Slide51: Conclusion Reviewed 5 of 16 WVRs for astronomy (7 radiometers tomorrow) Many clever techniques are available for use Lowest residual path 0.031 mm