SKA Key Science Requirements Matrix 2005
This document and the associated matrix have been compiled from input from members of the ISWG. Both this discussion document and the matrix will be updated during 2005, as the Key Science workshops refine the parameters for the key experiments, to create a consolidated description of the SKA key science goal SKA requirements.
In an earlier SKA memo (#29, www.skatelescope.org) a simple parameter space analysis was compiled, based on the SKA science case of 1998 (Taylor et al.). As that science case made no prioritisation of the individual goals, all science drivers were treated with equal weighting. By identifying the requirements of the individual science drivers with respect to a wide range of simple specifications (e.g. frequency requirements, requirement for multiple fields of view, continuum or channel sensitivity etc) we illustrated that even a simple analysis could offer insight into the ‘best-fit’ SKA specification.
Since 1998 the SKA science case has developed markedly – both with the advent of what might be considered far more challenging experiments (e.g. ‘Dark Energy’ and ‘the Epoch of Reionization’) and the identification of the ‘Key’ science drivers: these being the head-line experiments which the SKA would be capable of addressing (Gaensler et al, SKA memo #44).
This new analysis has reviewed the parameters in the matrix, using the SKA specification goals set out by Dayton Jones et al (SKA memo #45). This matrix and explanation document build on the content of memo #45 as here we break down the requirements per KSG experiment. We can clearly see where the demanding overall SKA specification originates. An important point to remember is that 4/5 of the “Key Science Goals” are themes, and whilst they collect together highly-related topics, the requirements of their 3 experiments can be radically different – producing an ‘SKA specification’ that detractors like to quote as being from ‘DC to daylight’.
Matrix entries for each key science goal experiment
This section defines the columns/sections and values adopted in this matrix.
1 SKA Frequency range requirements
The frequency columns are the driving values in the specification for an experiment. If an experiment grades a particular frequency range as “0” (not required/useless) then there are no entries in this frequency range for any other parameter in the matrix.
Frequency ranges are assigned values of
Also has a second value for each frequency, showing the bandwidth required - "S" for the 'default SKA spec' of 25% of maximum band freq, or 4 GHz if above 16 GHz).
2. A/T sensitivity
A/T is to capture sensitivity requirement, obviously tempered by the galactic background (the top line of the matrix indicates this likely ‘base’ value).
Currently specified as the continuous FOV imaging size for each FOV (BUT see discussion of next section).
4. Spatial Resolution
Indicate the size of the imaged sources.
5. Minimum baseline
Length between the shortest baselines in km.
The next set of columns indicates the physical extent/distribution of the SKA collecting area over a limited size scale:
6. Array core %
The fraction of the SKA collecting area within "array core extent" (see item 7 below).
7. Array core extent
The physical dimension of the SKA close-packed "core" in km.
8. Mid range %
The fraction of the SKA collecting area within "mid range extent" (see item 9 below).
9. Mid range extent
The physical dimension of the SKA at its intermediate scale in km.
10. Large range %
The fraction of the SKA collecting area within the "large range extent" (see item 11 below).
11. Large range extent
The physical dimension of the SKA at the largest scale in km.
Note that the straw-man science requirements specification adopts values of
20% within 1km = “core”
50% within 5km = “mid-range”
75% within 150km = “large range”
with the rest (25%) distributed out to the "max baseline" (item 12).
12. Maximum baseline
The length of the longest baseline in km.
13. # pencil beams from core array
The number of simultaneous summed (phased array) beams within a single FOV formed from the innermost core (columns 6-7).
14. # pencil beams from mid-range array
The number of simultaneous summed (phased array) beams within a single FOV formed from the mid-range core (columns 7-8) of the SKA.
15. # separate FOVs
The number of separate FOVs required at the full sensitivity of the SKA.
16. # simultaneous frequency bands
The number of simultaneous independent observing bands required.
17. Dynamic range
The dynamic range requirement in either (I) imaging or (S) spectra.
18. # channels
The number of channels per observing band per baseline.
19. Total power
If the total power measurement is required (Y) or (N).
Defines criticality of dual polarization measurements – using same grades as for frequency bands (i.e. 3=critical to 0=not required/useless).
Definition of FOV specification
Before further effort is put into the entries of this matrix there is a key requirement that the SKA specification is made clearer regarding the meaning of
- Is this instantaneous or the total sum of all available beams?
- Must be specified for the KSGs across the frequency range as there is no viable SKA concept to provide full sensitivity, multiple FOVs at 10 GHz.
The current matrix has columns for this value but it needs populating.
Future applications & extensions of the matrix
In future the data in this matrix can be used to derive
(i) Figures-of-merit – e.g. survey speed described as, e.g., time to cover 10 square degrees to 100 nanoJy in continuum.
(ii) A full
comparison of the capabilities of both existing (LOFAR,
(iii) An ‘optimal’ SKA specification based upon weighting of the Key Science Drivers, where compromises between concepts can be understood.
(iv) A basis for identifying the best hybrid solutions.
Outcomes – so far
The matrix clearly illustrates that the experiments of the Key Science Goals span the entire range of ‘possible’ SKA frequency space, yet roughly can be mapped into 3 frequency regimes:
The frequency space that maximizes the number of key science experiments is very broad and for this set of goals cannot be narrowed easily (400 MHz – 20 GHz).
For each of the Key Science Goals the highest-scoring frequency range(s) are shown in blue: we find that
The requirement for an ultra-wide field of view is solely from the Equation of State experiment and covers the frequency range from 100 MHz to 1 GHz. A “> 1 degree” requirement for the pulsar searches extending to 2 GHz.
At the bottom of the matrix are 3 rows which sum the total ‘scores’ across the frequency bands:
Other SKA drivers
This matrix only records the specifications requirements for the “Key Science Goals”: as recorded in the new SKA science case there are many other drivers for the SKA, and not all are ‘pure’ astronomy – e.g. geodesy, spacecraft tracking etc.
Notes on Individual Matrix Entries for the Key Science Goals
These notes have, for the most part, been compiled by the authors of the matrix entires.
Key Science Goal A – Probing the Dark Ages.
Goal comprises 3 experiments, each using a different radio characteristic to probe the earliest objects in the Universe: EoR (1.4 GHz HI line), first star formation (110 GHz+ CO lines) and first AGN (sub-GHz continuum). Thus the SKA requirements differ within this goal.
Mapping the EoR: requires high sensitivity from a compact core region.
The first stars: Detect redshifted CO, requires sensitivity and resolution >4GHz. Complementing ALMA, but SKA would have far more sensitivity and resolution (and instantaneous sky coverage for blind surveys).
The first AGN: Requirement is to do an all-sky survey with broad frequency coverage over 200 – 600 MHz. The requirement is to obtain a radio spectrum for the sources so the goal is to obtain 0.1 mJy in 100 MHz sub-bands (or even 0.01 mJy if possible). This survey must be complete in one or two years, hence FOV is coupled to A/T. To follow-up the survey we would like to revisit a candidate list of N sources with VLBI at ~10 mas resolution. (N to be determined).
Key Science Goal B – Galaxy Evolution & Large Scale Structure.
Goal comprises 3 experiments – the determination of the “Equation of State” of the Universe and tracing galaxy evolution via HI and continuum surveys.
The EoS experiment has the highest FOV requirement, at least tens or hundreds of square degrees to derive the required galaxy redshift sample.
Key Science Goal C – Strong Field tests of General Relativity.
Goal comprises 3 experiments based on Pulsar timing:
Key Science Goal D – The Cradle of Life.
Goal comprises 3 aspects of searching for the origin of life:
SETI: the multiple field of view requirement column the requirement is shown as “1?” as it depends on how large the SETI target star list is. If there are more than 10^7 stars then we’ll have enough targets in one FOV to make effective use of all 100 pencil beams at lower frequencies ~1 GHz. FOV shrinks as you go higher in frequency so the target list would have to increase, or we might consider sub-arrays.
Search for extra-solar terrestrial planet formation requires high resolution.
Key Science Goal E – The Origin & Evolution of Cosmic Magnetism.
Strictly speaking, this goal comprises three separate experiments, namely "large scale structure" (all-sky RMs), "field evolution" (studying individual galaxies and foreground absorbers at a range of redshifts) and "field origin" (very high redshift polarization experiments on GRBs, CMB etc). . However the goal has been specified in one row only, as the goal was presented as a single topic to the ISSC.
So the specs chosen below don't consistently correspond to the same set of goals. (but do they show the extreme specs for all 3?)
1. Frequency coverage: the main frequency will be 1-2 GHz, where we will do surveys to measure all-sky RMs. 0.4-1 GHz and 2-10 GHz are also important for studying individual clusters/galaxies/regions of very low and very high magnetic field strength, respectively. Interesting but less critical science can be done outside this range.
2. A_eff/T_sys: this needs further discussion. For our level 0 proposal, we *assumed* A/T = 20,000 at 1.4 GHz for our all-sky RM grid and calculated everything else based on that. My guess is that other experiments within the magnetism project do not need anything like this in sensitivity, but this is something that needs to be discussed more widely.
3. FOV: The all-sky
RM grid requires 1 deg^2 at 1.4 GHz in order to be able to cover a hemisphere
of sky in ~1 year of observing. (An even larger FOV would obviously be better!)
At low frequencies, a similar FOV would be useful to map objects like nearby
galaxies, clusters and regions of the Milky Way. I don't have much feeling on what would be
important at higher frequencies; I suspect a simple scaling from 1.4 GHz should
be fine, but this should be discussed in
IMPORTANT NOTE: We require imaging of the *full* FOV at 1 arcsec resolution. This is a potentially challenging constraint which perhaps needs to be added to the matrix.
4. Spatial Resolution: most of our goals need 1 arcsec resolution (mainly to overcome beam depolarization). I have pushed this number down to 0.1 arcsec in the 1-4 GHz range to allow for deep confusion-limited imaging of synchrotron emission from high-redshift galaxies. Note that the aim is not to detect polarized emission from these sources, but to detect (& resolve?) them in total intensity. This then gives a handle on field strengths of proto-galaxies in the early Universe.
5. Min baseline: basically as compact as possible, to maximize sensitivity to extended structure in polarization in galaxies & clusters. 20m is a nominal, practical value, but even less would be better. The less often we have to rely on total power, the better.
6. Array core: 50% within 5 km is needed to map synchrotron emission from galaxy clusters.
7. Mid-range: 90% within 35 km gives us almost a km^2 at 1 arcsec resolution at 1.4 GHz, which enables our all-sky RM survey
8. Large-range/max baseline: we don't really need anything beyond 300 km, which corresponds to 0.1 arcsec at 1.4 GHz [see #4 above]
9. Pencil beams: not needed
10. Separate FOVs: only 1 essential (assuming we can image the full 1 deg^2 FOV at arcsec resolution), but more would increase the speed of our all-sky survey
11. Simul freq bands: nominally 2, to allow accurate correction for Faraday effects. (But how widely separated can these be? within a nominal band or anywhere within the SKA's range?)
12. Dynamic range: nominally 10^5 in Stokes I for deep fields? Not really sure about this one; maybe 10^6 is needed?
13. No of channels: 100 per polarization per band per baseline, to avoid bandwidth smearing and allow full-FOV imaging (assuming 350 MHz bandwidth at 1.4 GHz). Also useful for spectropolarimetry experiments and high RMs.
14. Total power: yes, critical for some applications within this project
*IMPORTANT*: not just total power, but properly calibrated Stokes Q/U/V is needed at the zero spacing also.
15. Dual polarization: absolutely essential. We also need high polarization purity (although the spec for this is complicated and probably isn't best given as a single number; different purity requirements are needed for mosaicing/surveys and for single deep pointings).