Requirement on SKA Low calibration transfer - further work needed

There are several gaps and questions surrounding an important Low calibration requirement. Resolving these will likely require some analysis / simulation.

The relevant L1 requirement is:
SKA1-SYS_REQ-3542 'SKA1_Low calibration transfer' reads:
When commanded, SKA1_Low shall transfer calibration seamlessly across frequency changes, band changes, and/or source changes, such that the any change in calibration due to the telescope (as compared to the environment or the source) shall be less than 0.1% in amplitude and 0.001 radians in phase.

The child LFAA requirement SKA1-LFAA-47 'Calibration transfer' reads:
When transferring calibration between beams and pointing directions, the change due to the telescope (as compared to the environment or the source) shall be less than 0.7% (TBC) in amplitude and 0.07 (TBC) radians in phase.

Further work that is required is as follows:

• The TBCs in SKA1-LFAA-47 need to be resolved. In particular, it has been suggested that the 0.07 radians in phase should be a factor of 10 smaller. In its current form the requirement allows the phase errors to be a factor of ~10 higher than the amplitude errors.
• Secondly, the current wording of SKA1-LFAA-47 assumes that its parent SKA1-SYS_REQ-3542 should be modified (to 1% in amplitude and 0.01 radians in phase, that could potentially be relaxed to 2%/0.02 rad). It also needs to be clarified which part of the beam is being referred to (e.g. within halfpower radius), as this requirement is clearly inappropriate for large radii.
• M. Waterson has indicated there might be a missing requirement for amplitude stability. This should be analysed.

Regarding the third bullet. In an email from M. Waterson:

"Robert L and I have also been discussing flux transfer vs stability, coming to the conclusion that a budget allocating 2.5% to the source flux error and 2.5% to amplitude stability seems like a reasonable target budget (under the SKA situation, where we only expect precision performance above 45 degrees...). My own test scenario is for the case where the calibrator is (say) at 2 hours past transit and the target 2 hours ahead, so that your reference measurement target will be descending into larger and larger error space while your target field is coming into "precision" territory - this would be the case where the maximum reliance on system stability is needed.

I think there needs to be a separate analysis of how accurately you can make any individual flux/amplitude measurement (which depends on the pattern error and SNR), and how long it can be trusted (which depends on the amplitude stability of the system including whatever dynamic calibration adjustments are made). We should know both the un-adjusted drifts as well as what we could expect from a reference measurement considering that both cases will eventually be used in someone's observations, and also given the limited number of southern-sky calibrators for direct measurements."