ngVLA Memo No. 108

Feasibility of Self-cal for ngVLA Dynamic Range Requirements · 2025

Read ngVLA Memo No. 108 by Feasibility of Self-cal for ngVLA Dynamic Range Requirements (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-14441.txt.

Introduction

A planned radio telescope array, the ngVLA, promises images far sharper in contrast than today's, and keeping that promise hinges on self-calibration. Sridharan and Bhatnagar argued in 2025 that the array's own sensitivity makes the trick feasible as a routine baseline strategy.

Self-calibration uses bright sky sources in the field to correct atmospheric wobbles on short timescales. The memo derives a general limit on achievable contrast that depends only on antenna count and target noise, not on band or dish performance. That clean bound frames whether the required forty-five and thirty-five decibel ranges at two bands can be met. The authors conclude the strategy is sound.

Catalog record

CREATORFeasibility of Self-cal for ngVLA Dynamic Range Requirements
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE21 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-14441

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

ngVLA Memo No. 108 Feasibility of Self-cal for ngVLA Dynamic Range Requirements Abstract T. K. Sridharan & Sanjay Bhatnagar 1 Background and Motivation The ngVLA science goals require the ability to detect weak emission in the presence of strong peak emission at levels that demand high dynamic ranges in the constructed images. The emission occurring within an observing field in general and their peaks can be exploited to meet the specified dynamic range requirements through self-calibration. When sufficient signal to noise ratio is realized in an observation, self-calibration is almost always used to improve image quality. The feasibility of this approach can be assessed by looking at source counts of background sources serendipitously occurring within the fields of view for each band in relation to achievable noise levels, even in the absence of strong emission from the science target. An alternative approach is to recognize the broader nature of the question - requiring a certain dynamic range, DR , at a targeted science noise level \sigma_{science} presupposes the presence of bright emission in the field at a corresponding level of detected interferometric flux of \sim DR\times\sigma_{science} , by definition. In fact, it is emission from such a background source that drives the most stringent dynamic range requirement for ngVLA observations in the first place. The 45 dB requirement in Band 2 (SCI0113, SYS6103) results from the key science goal targeting deep-field continuum studies of MW-like galaxies at 8 GHz (KSG03), in the presence of a brighter background galaxy source in the field. Would that source allow self-cal? With the problem posed broadly in this way, one can derive a general limit without recourse to background source counts and the attendant Poisson fluctuations of their occurrence, and which depends only on the number of antennas in the array and the science noise level of the observation, independent of the observing band, primary beam size and antenna SEFDs. In the simple case of a point source, this implies a limit on the solution interval at which self-cal can be carried out. This limit is derived below. The approach is not fundamentally new, known to many in one form or another and is captured in the maxim roughly stated as “ …if you are DR limited, you can selfcal and selfcal will help; if you benefit from selfcal you are (were) DR limited; conversely, if there is not enough SNR for selfcal, you have reached the DR limit your data allows… ” ( e.g. Bhatnagar, private communication). This synthesis imaging lore and its implications are placed on a formal footing here. 2 Solution Interval Limit With DR= dynamic range demanded N_{A}= Number of antennas SNR_{A\_selfcal\_threshold}= required SNR per antenna, including all baselines to it, to allow self-cal t_{sol}= self-cal solution interval lower limit (for a given DR,SNR_{A\_selfcal\_threshold} and N_{A} ) t_{int}= integration time to reach a science rms of \sigma_{science} needing to fulfill a demanded dynamic range DR implies a peak flux in the field of S_{p}=DR\times\sigma_{science}=DR\times SEFD\times\sqrt{2}/\sqrt{N_{A}(N_{A}-1)\Delta\nu t_{int}} (1) Similarly, to achieve a SNR of SNR_{A\_selfcal\_threshold} for one antenna for self-cal to be feasible on a solution interval time scale t_{sol} , using all baselines to it, the peak flux needed is S_{p}=SNR_{A\_selfcal\_threshold}\times SEFD/\sqrt{(N_{A}-1)\Delta\nu t_{sol}} (2) combining (1) and (2) t_{sol}/t_{int}=N_{A}\times(SNR_{A\_selfcal\_threshold}/DR)^{2}/2 (3) or equivalently, t_{sol}=t_{int}\times N_{A}\times(SNR_{A\_selfcal\_threshold}/DR)^{2}/2 (4) t_{sol} is the lower limit to the solution interval above which a SNR of SNR_{A\_selfcal\_threshold} for self-cal to be feasible, is achieved. While the discussion above considers point sources and peak fluxes as these cases are more readily appreciated, we note that the arguments are equally applicable and agnostic to other source structures present. What matters for the signal-to-noise ratio is the interferometric flux actually detected, irrespective of the structures from which it arises. A measure of the detected flux in a general image is the quadrature sum of the emission - as adopted in the ngVLA requirements definition. Table 1: Expected SelfCal solution intervals ( N_{A}=214 ; SNR_{A\_SelfCal\_threshold}=3;SNR_{map}=31 ). Band Freq. DR t_{sol}/t_{int} t_{sol1hr} t_{sol10hr} t_{sol16hr} t_{sol100hr} Driving SCI requirement (GHz) (dB) (sec) (sec) (sec) (sec) case, integration time & rms 1 1.2 45 9.6\times 10^{-7} 0.003 0.03 0.06 0.35 2 8 45 9.6\times 10^{-7} 0.003 0.03 0.06 0.35 SCI-0113, 16hr, 0.035 \mu Jy 3 18 40 9.6\times 10^{-6} 0.035 0.35 0.6 3.5 4 27 35 9.6\times 10^{-5} 0.35 3.5 6 35 SCI-0113, 1hr, 0.18 \mu Jy 5 50 32 3.8\times 10^{-4} 1.4 14 22 140 6 70 30 9.6\times 10^{-4} 3.5 35 56 350 6 116 28 2.4\times 10^{-3} 8.6 86 138 860 • (1) The two frequencies with relevant science requirements derived from key science goals are highlighted, vis., 8GHz: SCI-0113, KSG3-011, 16 hr integration, 0.035 \mu Jy and 27 GHz: SCI-0113, KSG3-008, 1 hr integration 0.18 \mu Jy . • (2)The applicable t_{sol} limit for these cases are shown in bold. Table 1 lists t_{sol}/t_{int} and t_{sol} for SNR_{A\_selfcal\_threshold}=3 , for different frequencies for which dynamic range requirements are specified in the ngVLA system requirements and for different values of integration times. The choice of SNR_{A\_selfcal\_threshold}=3 is equivalent to full map SNRs of 22 and 31 for 107 and 214 antennas for an integration time of t_{sol} . The listed science integration times (or equivalently, science noise levels) of 1, 10 and 100 hrs, span the range of integration times in the relevant science goals which drive the DR specifications. The science requirements, and therefore the system requirements, only specify DR for 8 GHz and 27 GHz. Thus, driving DR requirements exist only for these frequencies. For other bands, DR values have been inter-/extrapolated. As an aside, we note that it is formally incomplete to specify a dynamic range without a targeted science noise level (or integration time) – it matters where the specified dynamic range is placed in the range of observed flux levels. This has strong implications for applicable calibration strategies, in particular, if self-cal would be feasible or not. The science requirements may be revised to state the dynamic range requirements along with the noise levels required. As can be seen from the table, as the required dynamic range increases, the feasible solution interval decreases. With a higher demanded dynamic range, the peak emission that demands this dynamic range is higher, allowing a smaller solution interval. For the two frequencies, 8 & 27 GHz, where relevant science requirements derived from key science goals drive the highest DRs, the minimum solution interval limits indicated are <1 s, above which SNR_{A\_selfcal\_threshold}= 3 is achieved, for the noise levels (or equivalently, the integration times) called for by the respective KSGs. In other words, there is sufficient SNR to deliver self-cal on the time scales as short as indicated. This is made possible by the high sensitivity of the ngVLA arising from its large collecting area. Thus, very fine-grained tracking of antenna gains is possible through self-cal. Based on past experience, self-cal is the only time-tested path to achieve high dynamic range imaging and should be a key calibration strategy for the ngVLA. The numbers derived show that this is also feasible: for the time scales implied – as short as 0.06 s and 0.35 s at 8 & 27 GHz – experience indicates a comfortable situation for attaining high dynamic range

Read the complete text