We have been able to substantially improve the linear mosaicing algorithm. A revised algorithm is found to produce much better images when care is taken to sample out to regions of zero brightness. Formerly, it was believed that the computationally inexpensive linear mosaics would only be used for very low SNR. The new algorithm results in dynamic ranges exceeding 500:1 when a guard band of 3 HWHM is observed around the image, dynamic ranges of about 100:1 with a guard band of 2 HWHM, and dynamic ranges of about 50:1 with a 1 HWHM guard band.
A more realistic model for pointing errors in an array of antennas is discussed and applied to simulations.
Systematic components of the array pointing such as global offsets and global drifts affect image quality more than the random pointing components. Systematic pointing errors are also easier to calibrate.
For bright sources observed at 230GHz, the dynamic range of a mosaic image will be limited by 1" pointing errors to about 750:1 unless the pointing errors are calibrated. This limit depends strongly on the pointing error model. The 1" pointing specification is required by mosaic to achieve high dynamic range and to accurately measure all spacings, not just the short spacings. With 2" pointing errors dynamic ranges of 400:1 are possible.
With 1" pointing errors, the dynamic range for objects with Tb < 40K will be limited by thermal noise (1MHz bandwidth, 1 minute per pointing). Dynamic range in images of brighter objects will be limited by pointing errors unless pointing calibration is performed.
Images produced from simulated homogeneous array are compared to images produced from simulated single dish data. Noise and pointing errors are added to each. The fidelity of the images on the relevant spatial frequencies is gauged. On all spacings, but particularly from 0- 15 meters, the homogeneous array images outperform the single dish images. This is true even when a more lenient pointing error model is applied to the single dish observations.
Crude simulations of the residual effects of the atmosphere an calibrated data are obtained ading random fluctuations and drifts to the antenna gains. the imaging capabilities of the homogeneous array are found to be robust even with very large errors ( 10-20%).
Good images can be made by obtaiing equal signal to noise rather than equal moise on all spatial frequencies. Hence, total power measurements may take only a few percent of the time of a mosaic observation.
The MMA is able to image planets quite well, with and without noise and errors.
There are no imaging problems which we ave come across so far that require a large total power antenna for the measurement of a short spacings or for any other technial reason. Our simulations will continue with an open mind (see Section 8 for planned simulations) searching for situations in which the homogeneous array fails. There are strong indications that a large central element is more adversely affected by the errors we have simulated to date, so it seems unlikely that any future problems with the homogeneous array could be solved by adding a large central element. Considering all simulations to date, the homogeneous array design is adequate to accomplish the scientific goals which have motivated the proposed Millimeter Array.
View a pdf version of MMA Memo 61.
Last modified: 2002-01-24
alma-memos@nrao.edu