The majority of Liverpool Telescope (LT) time is allocated via biannual calls for proposals via committees representing the UK and Spanish user communities. This works well for the existing LT user model, however, for the NRT it has some flaws: the nature of target-of-opportunity science means individual targets cannot be easily assigned to specific users, leading to some duplication of observations for high profile targets. For NRT, on which we anticipate early-time transient classification to be a major activity, there is also the worry that a particular science PI might 'waste' a large fraction of their individual time allocation following up transients which after identification turn out not to match their particular science case, but may well be of use to a different PI. There is clear value to both PIs in coordination, as well as rapid propagation of results to the time domain community. Cenko+ (2020) makes the recommendation: ​‘ToO policies are constructed to encourage groups to pool resources, work together (where sensible), and maximise the science possible from any given dataset’.

​

The final time allocation model will be a matter for discussion by the telescope stakeholders, however, the NRT project will advocate a hybrid model in which the available telescope time is split between a traditional committee-allocated model, and a Key Science Programme to be undertaken as a joint venture between all the partners. We refer to this programme as ‘SPEC’ time, as we imagine it will largely consist of spectroscopic, transient response observations, although other observation modes could potentially be used as part of this programme. The purpose of SPEC is to respond to the challenges of time domain science in the upcoming era. There will be a science oversight committee (informed directly by science working groups), to manage target selection. All data collected via the SPEC programme will be made freely available to NRT stakeholders. The motivation for the SPEC survey is to address the central challenge of target-of-opportunity science, namely that the rate of discovery of new targets of opportunity is extremely high and will only grow. SPEC time will allow the classification of 10,000 targets per year brighter than ​r=20.5​. This represents a factor of 5 increase over the number of spectral classifications (2012) delivered worldwide in 2019 (Kulkarni, S, 2020 arxiv.org/2004.03511) by the combined resources of the entire astronomical community.
 

One of the core strengths of the LT operations model is the ability to schedule new observations on the telescope in real time, and receive the data within minutes of it being obtained. This exploits the potential of robotic astronomy in an era when observing programmes are increasingly collaborative between multiple facilities. Like LT, NRT will be both triggered by other facilities (e.g., responding to a new transient from a wide field survey) as well as triggering further follow-up (e.g. NRT classification spectrum prompts space based or large aperture telescope override). This is particularly powerful when the option is provided to do this without human intervention. It is a design goal of the NRT project to make both automated observation scheduling and data access a straightforward, open and well documented process in order to facilitate the integration of both processes within any current or future transient broker or marshal.

 

Example use cases would be the robotic GOTO survey facility having the capability to autonomously trigger spectroscopic NRT follow-up of a candidate gravitational wave counterpart, or a user of the SPEC programme adding an object of interest into the NRT observing queue directly from the interface of a transient broker such as LASAIR, or the reduced spectrum that is subsequently obtained then being automatically ingested back into that database.

 

Alongside these tools we will continue to maintain a comprehensive archive of all LT and NRT data, with both raw data and high-level data products in a queryable database. We would propose to continue to maintain a proprietary period of no more than 1 year on PI-led data. Our broad approach is in line with the Cenko+ (2020) recommendation that data access should be FAIR (findable, accessible, interoperable, and reusable).