Date: Mon, 20 Mar 2000 16:22:56 GMT From: John Richer ALMA Phase Correction Project Dual Interferometer and 183 GHz Analysis Summary ------- As part of the UK's in kind contribution to ALMA, we are working on various aspects of the site characterisation and phase correction problems for ALMA. Yasmin Robson (YR)is a half-time PDRA doing most of the work, and is based in Oxford; she is supervised by John Richer and Richard Hills in Cambridge. We have received generous help from Simon Radford, Guillermo Delgado and Angel Otarola. The ultimate aim is to assess the quality of phase correction achievable with the 183 GHz interferometers which were built by the Cambridge/Onsala collaboration. Guillermo Delgado will report on that topic. In the meantime, we have begun a reassessment of the 11 GHz interferometer data to measure the height of the turbulent layer, which is an important criterion in setting the requirements on the beam offset between 183 GHz and astronomical beams. The initial results seem to indicate that most of the turbulence causing the phase fluctuations is at an altitude of a few hundred metres. Upgrades to the Chajnantor radiometers are planned for April 2000, including software modifications to allow recovery from power failures, and network changes to allow future software modifications to be made more easily. We need to discuss in detail how to proceed with the collaboration on the Chajnantor 183 GHz data, and on plans for a second-generation 183GHz receiver. Web Page -------- YR has established an ALMA site where our results can be posted as neccessary. http://www.mrao.cam.ac.uk/~yr/ Data Analysis ------------- We have purchased a dedicated PC runing Linux for the analysis. Data are being collected by two pairs of test interferometers on the proposed Chajnantor site, at an altitude of 5000m in the Atacama desert of Northern Chile. Each interferometer consists of two 1.8 m offset feed antennas, with an E-W baseline of 300m and beamwidth of 1.2 degrees. The interferometer observes unmodulated beacon signals at 11.198 GHz, broadcast from geostationary satellites and measures the phase difference between the signals received by each antenna pair. Communications channels have been established with both the U.S. and ESO partners and access gained to the two sets of interferometer data via their web pages (hereafter the interferometers will be refered to as LSA and NRAO). Data is brought over as required and converted from raw data files (LabView datalogs) to text time series of the interferometer phase, using a program (dlg2text.c) provided by Simon Radford (nrao). Files typically contain 24 hours of data with samples every second. Baseline Removal ---------------- YR has written programs using IDL to remove second order polynomials from the data. This takes into account the diurnal and other components of the satellite motion which causes large swings in the interferometer phase. Comparison of the data from the different interferometers, over similar times, show excellent agreement. We have resolved various problems associated with timestamps on the data and format conversion. The results are shown on: http://www.mrao.cam.ac.uk/~yr/results/baseline_removal/ The numbers below the plot represent the start date of the profile, the r.m.s deviation and the variance. Height of Turbulence -------------------- It is possible to estimate the height of the turbulent layer causing the phase fluctuations from the the time delay between the two phases measured by the two interferometers, and a knowledge of the wind velocity. This technique works because the interferometers are looking in slightly different directions, but we do need to assume that the fluctuations arise in a thin layer of "frozen" turbulence, and that the wind velocity on the ground is a reasonable approximation to that in the turbulent layer. The data from different interferometers is then cross-correlated over similar times to determine any time delays and the implied height of turbulent regions estimated. a)Matching Data Sets The nrao interferometer has been collecting data on the Chajnantor site in Chile since May 1995, while the lsa interferometer on that site has data from June 1998. Possible dates for which matching data sets from both interferometers may be available are given on: http://www.mrao.cam.ac.uk/~yr/results/obs_dates However an overlap in UT is not always available. So, for instance, no data appears on the web for the lsa interferometer from May to September 1999 inclusive. In October 1999 only one day of data is common to both interferometers; while other months may only have a few days of data. * note: the 'day of year' conventions adopted by the lsa and the nrao weather data differs by one day. b)Interpolation YR has written a program using IDL to interpolate any small gaps of one or two seconds once the matching data sets have been found. The output is two sets of matching time and phase series, each 1000 seconds long. *problem: one of the lsa raw data sets (981129a4.dlg) gave an overlap of about four seconds when converted to a time series. Discussions with Simon Radford (NRAO) indicate that there is probably a thermal drift in the sampler clock as the computer warms up. So, on this occassion, the sampler clock has 'run fast' collecting all the necessary data in 56 rather than in 60 seconds. Subsequent integrations show that the clock then slowed down. So far, I have only found one day of data exhibiting this feature. c)Cross-correlation vs Weather Cross-correlating the nrao and lsa data sets for matching times yields a lag time indicative of the height of any turbulence in the atmosphere. The greater the lag-time, the higher the turbulence. IDL was used to process the data and the cross-correlation plots are shown on: http://www.mrao.cam.ac.uk/~yr/results/cross_correl/ The data processed shows a range of wind directions and speeds and the plots are annotated to indicate these parameters. Weather information is taken from: http://puppis.ls.eso.org/lsa/htmls/meteo.html http://www.tuc.nrao.edu/alma/sites/Chajnantor/data.c.html The wind directions are measured clockwise from North and indicate the direction FROM which the wind is blowing. The results of the cross-correlation are tabulated on: http://www.mrao.cam.ac.uk/~yr/results/weather_table For much of the time the wind direction is at about 270 degrees (i.e. from the sea, in an E-W direction, similar to the interferometers' baselines). These plots tend to be well behaved and show a negative lag (e.g. 19990419 - these also show a negative correlation coefficient and are due to an accidental switch during that day's observation - and 19991107). As can be seen from the table, there are problems when assessing the weather conditions on both those days: 19991107 has no available lsa weather data 19991107 + 19990419 the nrao wind direction is often unreliable, sticking at about 49 degrees. Data were also processed where the weather conditions are abnormal or rapidly changing: e.g. 19981027 UT 26-30 hours show the wind direction is about 100 deg (i.e. opposite to normal). As expected, the lag is positive (about +20 seconds) and the correlation well-behaved. Subsequent plots for that date indicate that the wind direction and speed are changing. These plots are less well-defined and a definite lag time is harder to obtain. As the day progresses the wind settles down, eventually blowing in the more usual direction. The lag is then negative, as expected (e.g. 19981027 UT 45.2). d) Interferometer Geometry and Height of Turbulence Following discussions with Simon, Angel and Guillermo, we have now derived the correct geometry of the interferometers, which allows us to derive the scale height (assuming a thin, frozen layer). The final results are shown in the last column of the weather_table file referred to above. In general, the height is a few hundred metres, in a wide range of wind directions and strengths. This suggests that the current specifications on the radiometer beam offset from the astronomical beam (3-10 arcmin) should be adequate. 18/3/2000 Yasmin Robson John Richer Richard Hills