04.11.2014 Views

The Impact of Scanning Pattern Strategies on Uniform Sky ... - JACH

The Impact of Scanning Pattern Strategies on Uniform Sky ... - JACH

The Impact of Scanning Pattern Strategies on Uniform Sky ... - JACH

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Doc No: SC2/ANA/S210/008<br />

Vers: 0.2<br />

Category: Report<br />

Doc Type: L A TEX<br />

State: Draft<br />

Author: Jennifer Balfour, Douglas Scott & Edward Chapin<br />

Date: 2006/11/16<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong><br />

<strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large Maps<br />

Summary<br />

This document discusses the impact <str<strong>on</strong>g>of</str<strong>on</strong>g> the SCUBA-2 (Holland et al. 2006) array design <strong>on</strong><br />

the ability to fully sample large maps with the proposed scanning strategies. In particular we<br />

study how the sampling <str<strong>on</strong>g>of</str<strong>on</strong>g> maps depends <strong>on</strong> the orientati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the SCUBA-2 array relative to<br />

the scan pattern, also c<strong>on</strong>sidering the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> the gaps between sub-arrays and the possibility<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> blocks <str<strong>on</strong>g>of</str<strong>on</strong>g> n<strong>on</strong>-functi<strong>on</strong>ing bolometers. Pseudo-code for scanning strategies are added in<br />

appendices.<br />

1 SCUBA-2 : A submillimetre camera for the JCMT<br />

SCUBA-2, due to arrive at the James Clerk Maxwell Telescope <strong>on</strong> Mauna Kea early in 2007,<br />

will be used to map large areas <str<strong>on</strong>g>of</str<strong>on</strong>g> the submillimetre sky at 450 µm and 850 µm. One <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

primary science goals <str<strong>on</strong>g>of</str<strong>on</strong>g> SCUBA-2 is to map the sky in an efficient manner, using filled arrays<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> bolometers and scanning patterns designed to cover large areas in minimum time. With<br />

an 8 × 8 arcminute field <str<strong>on</strong>g>of</str<strong>on</strong>g> view and over 10,000 bolometers in two arrays, SCUBA-2 will be<br />

able to map the sky up to 1000 faster than the original SCUBA instrument. For example, a<br />

proposed 180 ◦ × 2 ◦ Galactic Plane survey will take 50 hours, as opposed to over 5 years with<br />

SCUBA. Wide-field surveys performed by SCUBA-2 will be invaluable to the next generati<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> submm interferometers such as ALMA. In additi<strong>on</strong>, SCUBA-2’s high sensitivity at multiple<br />

wavelengths will allow observers to exploit periods <str<strong>on</strong>g>of</str<strong>on</strong>g> good weather at the JCMT. In order to<br />

do so, s<str<strong>on</strong>g>of</str<strong>on</strong>g>tware is being developed which will provide the user with the best scanning soluti<strong>on</strong>s<br />

to cover the chosen area <str<strong>on</strong>g>of</str<strong>on</strong>g> sky to the required depth, while ensuring adequate coverage <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

the entire map.<br />

2 SCUBA-2 Array Design<br />

SCUBA-2 is comprised <str<strong>on</strong>g>of</str<strong>on</strong>g> 2 bolometer arrays, <strong>on</strong>e operating at 450 µm, and <strong>on</strong>e at 850 µm.<br />

Each <str<strong>on</strong>g>of</str<strong>on</strong>g> these arrays is divided into 4 subarrays <str<strong>on</strong>g>of</str<strong>on</strong>g> 32 by 40 detectors, arranged in a “pinwheel”<br />

formati<strong>on</strong> (Figure 1).


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 4 SCANNING PATTERNS<br />

Figure 1: SCUBA-2 subarray layout. Each sub-array c<strong>on</strong>sists <str<strong>on</strong>g>of</str<strong>on</strong>g> 32 rows <str<strong>on</strong>g>of</str<strong>on</strong>g> 40 detectors with a gap<br />

corresp<strong>on</strong>ding to the extent <str<strong>on</strong>g>of</str<strong>on</strong>g> approximately 4 detectors.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> sub-arrays themselves are filled areas <str<strong>on</strong>g>of</str<strong>on</strong>g> multiplexed transiti<strong>on</strong> edge sensors, which provide<br />

a fully sampled image at 850 µm, and an undersampled image at 450 µm. Due to the<br />

complex fabricati<strong>on</strong> process <str<strong>on</strong>g>of</str<strong>on</strong>g> the high number <str<strong>on</strong>g>of</str<strong>on</strong>g> bolometers, it is estimated that perhaps<br />

10–20% <str<strong>on</strong>g>of</str<strong>on</strong>g> the bolometers will not functi<strong>on</strong> correctly. Possible causes are manufacturing defects<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> individual bolometers, shorts in patches <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays, and wiring failures in the<br />

multiplexing which could cause an entire row or column <str<strong>on</strong>g>of</str<strong>on</strong>g> bolometers to malfuncti<strong>on</strong>.<br />

Several different scan patterns are being c<strong>on</strong>sidered. 11 <str<strong>on</strong>g>The</str<strong>on</strong>g>se scanning patterns will be used<br />

to map out areas larger than the field <str<strong>on</strong>g>of</str<strong>on</strong>g> view <str<strong>on</strong>g>of</str<strong>on</strong>g> the arrays, while also providing fully sampled<br />

450 µm array data, filling the gaps between the sub-arrays, and covering areas missed by<br />

dead and faulty bolometers.<br />

3 SCUBA-2 Operating Modes<br />

SCUBA-2 will have 3 primary modes <str<strong>on</strong>g>of</str<strong>on</strong>g> operati<strong>on</strong>, STARE, 6 DREAM, 7 and SCAN 4, 5 . STARE<br />

is a simple ‘point-and-shoot’ mode in which the camera stares at a specified area <str<strong>on</strong>g>of</str<strong>on</strong>g> sky for a<br />

period <str<strong>on</strong>g>of</str<strong>on</strong>g> time. In DREAM (Dutch REal-time Acquisiti<strong>on</strong> Mode) the sec<strong>on</strong>dary mirror (SMU)<br />

is rapidly moved in a star-like pattern such that each bolometer observes multiple points <strong>on</strong><br />

the sky. SCAN patterns such as P<strong>on</strong>g, Boustrophed<strong>on</strong>, and Lissajous are used to map large<br />

areas. Each scanning pattern determines a soluti<strong>on</strong> for x(t) and y(t) with which to drive the<br />

telescope while c<strong>on</strong>tinuously collecting data at 200 Hz. In additi<strong>on</strong> to these basic imaging<br />

and survey modes, SCUBA-2 will include imaging polarimetry with POL-2 8 and medium<br />

resoluti<strong>on</strong> spectroscopy with FTS-2. 2<br />

4 <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g>s<br />

Due to the shape and layout <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays, scanning patterns must be chosen which not<br />

<strong>on</strong>ly overcome the challenges <str<strong>on</strong>g>of</str<strong>on</strong>g> observing in the submillimetre, but also provide acceptable<br />

(i.e. ∼close to uniform) sampling across an entire map, despite the ‘gaps’ between the subarrays.<br />

Several factors must be c<strong>on</strong>sidered when designing a scanning pattern, the following<br />

being some <str<strong>on</strong>g>of</str<strong>on</strong>g> the most crucial 11 :<br />

1. Move the telescope across the sky as quickly as possible, taking data at a rate which is<br />

Page 2 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 4 SCANNING PATTERNS<br />

much faster than the changes in the atmosphere.<br />

2. Maximize the number <str<strong>on</strong>g>of</str<strong>on</strong>g> different detectors which collect data from each sky pixel in order<br />

to correct for instrumental variati<strong>on</strong>s including faulty bolometers, correlated noise,<br />

and gain fluctati<strong>on</strong>s.<br />

3. Cover each sky pixel in different directi<strong>on</strong>s while avoiding periodic patterns in order<br />

to separate astr<strong>on</strong>omical structure from temporal variati<strong>on</strong>s (i.e. varying atmospheric<br />

emissi<strong>on</strong> and gain fluctuati<strong>on</strong>s).<br />

In additi<strong>on</strong>, a scanning pattern should strive to provide uniform sky coverage. Given the<br />

design <str<strong>on</strong>g>of</str<strong>on</strong>g> the array and the wide variety <str<strong>on</strong>g>of</str<strong>on</strong>g> map shapes and areas a user might choose to scan<br />

(as well as the likelihood <str<strong>on</strong>g>of</str<strong>on</strong>g> n<strong>on</strong>-functi<strong>on</strong>ing bolometers), absolutely uniform coverage will<br />

rarely be possible. However, the scanning pattern should be designed to maximize sampling<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> every pixel while covering an area in a minimal time.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> simplest scanning strategy to map an area larger than the field-<str<strong>on</strong>g>of</str<strong>on</strong>g>-view <str<strong>on</strong>g>of</str<strong>on</strong>g> the array<br />

is a ‘boustrophed<strong>on</strong>’ or raster pattern, such as that previously implemented by the original<br />

SCUBA instrument. 3 In the simple example below this pattern is used to map a rectangular<br />

regi<strong>on</strong> at an angle parallel to <strong>on</strong>e side <str<strong>on</strong>g>of</str<strong>on</strong>g> the regi<strong>on</strong>. Additi<strong>on</strong>al details regarding the<br />

BOUSTROPHEDON scanning pattern are given in Appendix A.<br />

Figure 2: Simple Boustrophed<strong>on</strong> scanning pattern. <str<strong>on</strong>g>The</str<strong>on</strong>g> diagram indicates the path <str<strong>on</strong>g>of</str<strong>on</strong>g> the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

array as it scans across the sky. Successive sweeps should be no farther apart than the full array size<br />

to ensure that every regi<strong>on</strong> is covered, and it is probably desirable for the spacing to be less than half<br />

the field-<str<strong>on</strong>g>of</str<strong>on</strong>g>-view <str<strong>on</strong>g>of</str<strong>on</strong>g> the array in order to provide overlap in the two directi<strong>on</strong>s.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> Boustrophed<strong>on</strong> pattern violates the third <str<strong>on</strong>g>of</str<strong>on</strong>g> the guidelines above: the pixels are revisited<br />

in a periodic fashi<strong>on</strong>, and in additi<strong>on</strong>, each pixel is visited in <strong>on</strong>ly two directi<strong>on</strong>s (which are<br />

just 180 ◦ from each other).<br />

In order to overcome this, the regi<strong>on</strong> can be scanned at two different angles, neither <str<strong>on</strong>g>of</str<strong>on</strong>g> which<br />

are parallel to a side <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle. In Fig. 3 below, the boustrophed<strong>on</strong> pattern is again<br />

used to scan a rectangular regi<strong>on</strong>, but the time taken to cross the width <str<strong>on</strong>g>of</str<strong>on</strong>g> the pattern and<br />

turn around varies between sweeps across the sky. <str<strong>on</strong>g>The</str<strong>on</strong>g> pattern could then be repeated at an<br />

angle approximately 90 ◦ different from the first pattern (right panel <str<strong>on</strong>g>of</str<strong>on</strong>g> Fig. 5).<br />

For SCUBA-2, a scanning strategy called P<strong>on</strong>g, similar to the ‘boxscan’ used by SHARC-II<br />

<strong>on</strong> the CSO, 1 is proposed. In this pattern, a rectangular regi<strong>on</strong> is filled by ‘bouncing’ the<br />

telescope <str<strong>on</strong>g>of</str<strong>on</strong>g>f the limits <str<strong>on</strong>g>of</str<strong>on</strong>g> the box, covering each pixel in multiple directi<strong>on</strong>s and without periodic<br />

revisitati<strong>on</strong>. Similarly to a Lissajous pattern, the number <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘nodes’ (where the pattern<br />

Page 3 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 4 SCANNING PATTERNS<br />

Figure 3: Boustrophed<strong>on</strong> scanning at an angle (45 ◦ here). <str<strong>on</strong>g>The</str<strong>on</strong>g> right panel shows how a repeat <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

scan with a 90 ◦ rotati<strong>on</strong> provides cross-linking.<br />

bounces <str<strong>on</strong>g>of</str<strong>on</strong>g> the edges <str<strong>on</strong>g>of</str<strong>on</strong>g> the box) in the x and y directi<strong>on</strong>s must not share comm<strong>on</strong> factors,<br />

and <strong>on</strong>e must be even while the other is odd. This ensures that the pattern c<strong>on</strong>tinues until<br />

the entire box is filled, with relatively even coverage over the entire area (with a shallower<br />

border, about half the array wide). Additi<strong>on</strong>al details regarding the P<strong>on</strong>g scanning pattern<br />

are given in Appendix B.<br />

Figure 4: ‘Straight’ P<strong>on</strong>g scan with a map width ≃4000 arcsec, height ≃3200 arcsec, and sweep spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 500 arcsec. This spacing is measured between the parallel scan lines as indicated by the bold arrow.<br />

Two possible P<strong>on</strong>g soluti<strong>on</strong>s are proposed. In the first, the map is scanned in straight line<br />

segments between the vertices, with abrupt 90 ◦ turns at the boundaries <str<strong>on</strong>g>of</str<strong>on</strong>g> the box (Fig. 4).<br />

This strategy ensures that the sweeps across the sky are parallel and have c<strong>on</strong>stant spacing.<br />

Pseudo-code for this ‘straight’ P<strong>on</strong>g scanning pattern can be found in Appendix C. Note<br />

that in principle, an angle other than 45 ◦ can be selected for the paths c<strong>on</strong>necting the P<strong>on</strong>g<br />

vertices. In this case, the grid would be composed <str<strong>on</strong>g>of</str<strong>on</strong>g> rhombuses as opposed to squares, and<br />

the vertical and horiz<strong>on</strong>tal vertex spacing would not be equal. For simplicity, all P<strong>on</strong>g scans<br />

Page 4 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 4 SCANNING PATTERNS<br />

c<strong>on</strong>sidered in this document will be those which fill the box dimensi<strong>on</strong>s with 45 ◦ angles, providing<br />

orthog<strong>on</strong>al cross-linking.<br />

In the event that abrubt changes in velocity at the turnaround points are not preferable, <strong>on</strong>e<br />

could simple change the velocity in some way near the turnarounds. An alternative is that<br />

the corners can be ‘rounded-<str<strong>on</strong>g>of</str<strong>on</strong>g>f’ by approximating the P<strong>on</strong>g scan with a Fourier expanded<br />

Lissajous scan instead (Fig. 5). In this case, x(t) and y(t) are approximati<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> triangle waves<br />

with 5 terms each. Pseudo-code for this ‘straight’ P<strong>on</strong>g scanning pattern can be found in<br />

Appendix E. <str<strong>on</strong>g>The</str<strong>on</strong>g> specific formulae are :<br />

x(t) = 8α x<br />

π 2<br />

y(t) = 8α y<br />

π 2<br />

∑<br />

n=1,3,5,7,9<br />

∑<br />

n=1,3,5,7,9<br />

(−1) n−1<br />

2<br />

n 2<br />

(−1) n−1<br />

2<br />

n 2<br />

( ) 2πnt<br />

sin ; (1)<br />

β x<br />

( ) 2πnt<br />

sin . (2)<br />

β y<br />

Where α x and α y are the amplitudes <str<strong>on</strong>g>of</str<strong>on</strong>g> the x and y moti<strong>on</strong>, respectively, and β x and β y are<br />

the periods <str<strong>on</strong>g>of</str<strong>on</strong>g> the x and y moti<strong>on</strong>, respectively.<br />

Figure 5: ‘Curvy’ P<strong>on</strong>g scan created by approximating triangle waves with 5 terms in x(t) and y(t).<br />

This approach results in almost straight paths across the central regi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan while<br />

avoiding sharp corners at the vertices. It can be seen from Fig. 6 that with an approximati<strong>on</strong><br />

using 5 terms, the scans are almost identical in the path that they take to cover the map,<br />

with the excepti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the outer edges.<br />

In both Boustrophed<strong>on</strong> and P<strong>on</strong>g scanning, several factors must be c<strong>on</strong>sidered to ensure that<br />

the scanning area is fully sampled, and that there is not a wide variati<strong>on</strong> in sky coverage.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> distance between successive sweeps across the sky should be no farther apart than half<br />

the total field <str<strong>on</strong>g>of</str<strong>on</strong>g> view <str<strong>on</strong>g>of</str<strong>on</strong>g> the array to ensure that each pixel is visited in at least two directi<strong>on</strong>s<br />

and by multiple passes <str<strong>on</strong>g>of</str<strong>on</strong>g> the array. Also, the angle <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan relative to the array itself<br />

should be chosen such that the gap between sub-arrays (and undersampling at 450 µm) is<br />

Page 5 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 5 IDENTIFYING ‘BAD ANGLES’ IN SINGLE STRAIGHT SCANS<br />

Figure 6: ‘Straight’ and ‘Curvy’ P<strong>on</strong>g patterns compared.<br />

filled <strong>on</strong> a single sweep. <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> at an angle parallel or close to parallel to the gap between<br />

sub-arrays will cause there to be undersampled ‘stripes’ <str<strong>on</strong>g>of</str<strong>on</strong>g> lower or zero sampling <strong>on</strong> each<br />

sweep across the sky.<br />

5 Identifying ‘Bad Angles’ in single straight scans<br />

For SCUBA 3 there were ‘Nyquist angles’ chosen for scanning, such that successive rows filled<br />

in gaps between bolometers. However, the large number <str<strong>on</strong>g>of</str<strong>on</strong>g> rows (and columns) in SCUBA-<br />

2 means there are many ‘good angles’, and hence it is more appropriate to be c<strong>on</strong>cerned<br />

with avoiding ‘bad angles’ than with selecting ‘good angles’. To determine the impact <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

the chosen angle, the SCUBA-2 simulator was used to first map out the hits-per-pixel for<br />

simple straight scans across the sky at various angles, with four subarrays in their correct<br />

c<strong>on</strong>figurati<strong>on</strong> and with the assumpti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> no malfuncti<strong>on</strong>ing bolometers. In this scenario, the<br />

telescope accelerates to a maximum velocity <str<strong>on</strong>g>of</str<strong>on</strong>g> 600 arcsec s −1 , at which point the lowest pixel<br />

hitcount within a regi<strong>on</strong> surrounding the gap was measured. For these tests the pixel size<br />

is 6 arcsec and a ‘hit’ is counted simply through determining the pixel in which the centre <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

each bolometer lies at every 200 Hz sample.<br />

One can see from Fig. 7 that when the angle is 0◦ relative to the symmetry axis, there are<br />

zero hits in the gap and a great deal <str<strong>on</strong>g>of</str<strong>on</strong>g> structure around the edges. Figs. 8, 9, 10 and 11<br />

show the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> changing the scan angle to 10 ◦ , 20 ◦ , 30 ◦ , and 40 ◦ . One can see the gap<br />

filling up and the structure smoothing out as the angle is increased.<br />

NOTE : All the maps in this study were created from data generated by the SCUBA-2 simulator,<br />

using a pixel size <str<strong>on</strong>g>of</str<strong>on</strong>g> 6 arcsec<strong>on</strong>ds.<br />

In order to determine the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan angle <strong>on</strong> hits-per-pixel, the central regi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

scan was measured for minimum, median, and maximum hits-per-pixel. <str<strong>on</strong>g>The</str<strong>on</strong>g> area chosen<br />

for these measurements included the path <str<strong>on</strong>g>of</str<strong>on</strong>g> all four sub-arrays and excluded the start and<br />

endpoints <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan as shown in Fig. 12. Of course the detailed results will depend <strong>on</strong> the<br />

Page 6 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 5 IDENTIFYING ‘BAD ANGLES’ IN SINGLE STRAIGHT SCANS<br />

Figure 7: Hits per pixel for a scan at 0 ◦ (relative to <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the array edges) for a 1000 arcsec scan length<br />

with 600 arcsec s −1 scan velocity.<br />

Figure 8: Hits per pixel for a scan at 10.0 ◦ .<br />

specific choice for this area; we need to chose something and our choice (Fig. 12) minimizes<br />

effects from the ‘edges’, so we can focus <strong>on</strong> the ‘well-sampled’ centre <str<strong>on</strong>g>of</str<strong>on</strong>g> the mapped regi<strong>on</strong>.<br />

Fig. 13 shows the minimum, median, and maximum hits-per-pixel as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> angle,<br />

clearly indicating the severe drop-<str<strong>on</strong>g>of</str<strong>on</strong>g>f in the minimum sampling rate as the angle approaches<br />

4 ◦ . From examining the pixel count in the regi<strong>on</strong> surrounding the less sampled secti<strong>on</strong> caused<br />

by the gap, it is clear that below an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 4 ◦ , some pixels are not visited even <strong>on</strong>ce in a<br />

single path across the sky. Because gaps exist in both the horiz<strong>on</strong>tal and vertical directi<strong>on</strong>s,<br />

there is no guarantee that these missed pixels would be visited when the scanning pattern<br />

returns <strong>on</strong> the sec<strong>on</strong>d scan directi<strong>on</strong>, depending <strong>on</strong> the choice <str<strong>on</strong>g>of</str<strong>on</strong>g> overlap between successive<br />

scans. As a result, a scanning pattern chosen with minimum overlap (less than half the full<br />

array width), and scan angles <str<strong>on</strong>g>of</str<strong>on</strong>g> less than 4 ◦ from being parallel to the gaps may yield some<br />

pixels in the map which have zero hits.<br />

This value can be understood geometrically from the ratio <str<strong>on</strong>g>of</str<strong>on</strong>g> the gap size to the full array size<br />

(i.e. tan −1 (gapwidth / arraywidth) where the array width is twice the length <str<strong>on</strong>g>of</str<strong>on</strong>g> the side <str<strong>on</strong>g>of</str<strong>on</strong>g> a<br />

sub-array plus the width <str<strong>on</strong>g>of</str<strong>on</strong>g> the gap. Because <str<strong>on</strong>g>of</str<strong>on</strong>g> the n<strong>on</strong>-symmetrical ‘pinwheel’ orientati<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays, the arraywidth is either 2×32 + 4 or 2×40 + 4 detectors, depending <strong>on</strong> the<br />

directi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan). In either case, the answer is ∼4 ◦ .<br />

As the angle is increased above 4 ◦ , the gap is quickly filled in by the coverage provided by multiple<br />

bolometers. <str<strong>on</strong>g>The</str<strong>on</strong>g> minimum pixel sampling increases until reaching a maximum value at<br />

∼7 ◦ , and remains relatively c<strong>on</strong>stant to ∼33 ◦ . Bey<strong>on</strong>d this angle, the area with the minimum<br />

hits-per-pixel becomes the undersampled outer edges <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan path. At this point, the<br />

overlap <str<strong>on</strong>g>of</str<strong>on</strong>g> the scans across the sky becomes the primary factor in determining the minimum<br />

pixel count, as the area around the gap is well covered by the sub-array geometry.<br />

Page 7 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 5 IDENTIFYING ‘BAD ANGLES’ IN SINGLE STRAIGHT SCANS<br />

Figure 9: Hits per pixel for a scan at 20.0 ◦ .<br />

Figure 10: Hits per pixel for a scan at 30.0 ◦ .<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> median sampling (middle curve in Fig. 13) gradually decreases as the angle is increased<br />

from 0 ◦ to 45 ◦ . This is due to the rise in undersampling that occurs towards the outer edges <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

the scan. As the sub-array layout is tilted towards a 45 ◦ angle, the outer edge is sampled by<br />

a smaller area (the corner <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays). However, as expected, the maximum sampling<br />

(upper curve in Fig. 13 remains relatively c<strong>on</strong>stant, because for all angles there exist pixels<br />

which are ‘hit’ at almost every sample interval.<br />

When scanning at 450 µm, additi<strong>on</strong>al care must be taken to ensure that the scan angle fills in<br />

the undersampled pixels caused by having the same pixel size <strong>on</strong> the sky, but about half the<br />

beamsize. While the gap will have essentially the same effect <strong>on</strong> scans at 850 µm and 450 µm,<br />

there are more potential ‘bad’ angles for 450 µm scanning. For example, a scan at 45 ◦ would<br />

fail to compensate for the undersampling, because the centres <str<strong>on</strong>g>of</str<strong>on</strong>g> the detectors would pass<br />

directly over the same map pixels, leaving undersampled areas between the detectors. Crosslinking<br />

soluti<strong>on</strong>s such as P<strong>on</strong>g should be able to eliminate this problem, and in particular a<br />

scan which does not proceed in a straight line (for example, the ‘Curvy’ P<strong>on</strong>g soluti<strong>on</strong>) will<br />

not suffer from these additi<strong>on</strong>al ‘bad’ angles.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> basic c<strong>on</strong>clusi<strong>on</strong> from this part <str<strong>on</strong>g>of</str<strong>on</strong>g> our study is that <strong>on</strong>e should avoid having a substantial<br />

fracti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> data for a map being gathered at scan angles below about 7 ◦ , but otherwise the<br />

choice <str<strong>on</strong>g>of</str<strong>on</strong>g> angle makes no dramatic difference to the sampling.<br />

Page 8 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 5 IDENTIFYING ‘BAD ANGLES’ IN SINGLE STRAIGHT SCANS<br />

Figure 11: Hits per pixel for a scan at 40.0 ◦ .<br />

Figure 12: Area measured for hits-per-pixel<br />

Figure 13: Minimum (bottom line), median (central line), and maximum (top line) pixel sampling as a<br />

functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan angle for a single scan path, all measured within the box delineated in Fig. 12.<br />

Page 9 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 6 THE IMPACT OF FAULTY BOLOMETERS<br />

6 <str<strong>on</strong>g>The</str<strong>on</strong>g> impact <str<strong>on</strong>g>of</str<strong>on</strong>g> faulty bolometers<br />

It is guessed that SCUBA-2 may have 100–200 ‘dead’ bolometers per sub-array. Such bolometers<br />

may record incorrect values <str<strong>on</strong>g>of</str<strong>on</strong>g> flux, or may be unable to collect any data at all. In any<br />

case, malfuncti<strong>on</strong>ing bolometers must be flagged so as not to be used in any map-making<br />

procedure. <str<strong>on</strong>g>The</str<strong>on</strong>g> scanning patterns chosen should be able to provide a fully sampled and reas<strong>on</strong>ably<br />

uniform image at both 450µm and 850µm, regardless (within reas<strong>on</strong>) <str<strong>on</strong>g>of</str<strong>on</strong>g> the layout <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

the bad bolometers.<br />

Unlike the original SCUBA, which employed separated feedhorn-coupled arrays (Fig. 14),<br />

SCUBA-2 will have 4 filled sub-arrays <str<strong>on</strong>g>of</str<strong>on</strong>g> detectors (Fig. 15) which can provide a fully sampled<br />

image (at 850µm) without requiring ‘jiggling’ <str<strong>on</strong>g>of</str<strong>on</strong>g> the sec<strong>on</strong>dary mirror. <str<strong>on</strong>g>The</str<strong>on</strong>g> highly sensitive<br />

transiti<strong>on</strong>-edge sensors used to detect sub-mm radiati<strong>on</strong> must be kept at temperatures below<br />

1 K. In order have such a close-packed array <str<strong>on</strong>g>of</str<strong>on</strong>g> detectors, multiplexing is used to share comm<strong>on</strong><br />

wiring between detectors in a single column or row <str<strong>on</strong>g>of</str<strong>on</strong>g> a sub-array. As a result <str<strong>on</strong>g>of</str<strong>on</strong>g> this<br />

multiplexing, the possibility exists <str<strong>on</strong>g>of</str<strong>on</strong>g> losing an entire row (40 detectors) or column (32 detectors)<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> bolometers to a wiring or electr<strong>on</strong>ic fault. Blocks <str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>tiguous detectors could also<br />

be ‘dead’ due to fabricati<strong>on</strong> problems. On top <str<strong>on</strong>g>of</str<strong>on</strong>g> that it is likely that some isolated individual<br />

bolometers will not meet performance requirements. In perhaps the worst case scenario,<br />

groups <str<strong>on</strong>g>of</str<strong>on</strong>g> bolometers near the gaps will be affected, causing the undersampling effect <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

gap to have an impact at a wider range <str<strong>on</strong>g>of</str<strong>on</strong>g> angles.<br />

In order to examine the increase in angle required to compensate for dead rows <str<strong>on</strong>g>of</str<strong>on</strong>g> bolometers,<br />

simulati<strong>on</strong>s were run with the outer rows and columns <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays flagged as ‘bad’. In<br />

the first simulati<strong>on</strong>, <strong>on</strong>ly <strong>on</strong>e sub-array had bad bolometers added to it, effectively reducing<br />

the field <str<strong>on</strong>g>of</str<strong>on</strong>g> view <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-array by <strong>on</strong>e row/column <str<strong>on</strong>g>of</str<strong>on</strong>g> bolometers <strong>on</strong> all four sides. This<br />

creates a gap in the full array which, for half <str<strong>on</strong>g>of</str<strong>on</strong>g> its length, is <strong>on</strong>e bolometer wider than the<br />

regular gap. In this case, the three uncompromised subarrays are able to provide enough<br />

coverage that the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> the gap is not significantly worse, as we show in Fig. 16.<br />

In the next simulati<strong>on</strong>, the edges <str<strong>on</strong>g>of</str<strong>on</strong>g> all four sub-arrays were flagged as bad, essentially increasing<br />

the total width <str<strong>on</strong>g>of</str<strong>on</strong>g> the gaps from 4 to 6 bolometers for their entire length. This may<br />

be regarded as a worst case scenario, since it seems unlikely that isolated groups <str<strong>on</strong>g>of</str<strong>on</strong>g> dead<br />

bolometers within a sub-array could be as large as the gap between the sub-arrays.<br />

As expected, the minimum hits-per-pixel over the gap increases in a similar fashi<strong>on</strong> to scans<br />

without dead bolometers, however the angles at which some areas are totally missed include<br />

those within about 5 ◦ <str<strong>on</strong>g>of</str<strong>on</strong>g> the directi<strong>on</strong> parallel to the gap. This indicates that the angles<br />

Figure 14: SCUBA feedhorn-coupled array.<br />

Figure 15: SCUBA-2 filled array.<br />

Page 10 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 6 THE IMPACT OF FAULTY BOLOMETERS<br />

Figure 16: Simulati<strong>on</strong> with bad bolometers al<strong>on</strong>g all four edges <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>on</strong>e sub-array. Minimum, median,<br />

and maximum hits are plotted as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan angle, again measured within the box shown in<br />

Fig. 12.<br />

Figure 17: Simulati<strong>on</strong> with bad bolometers al<strong>on</strong>g all four edges <str<strong>on</strong>g>of</str<strong>on</strong>g> every sub-array. Minimum, median,<br />

and maximum hits are plotted as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan angle.<br />

Page 11 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 6 THE IMPACT OF FAULTY BOLOMETERS<br />

Figure 18: Black indicates approximate area<br />

flagged as ‘bad’ for simulati<strong>on</strong> in Figs 20 and<br />

22.<br />

Figure 19: Black indicates approximate area<br />

flagged as ‘bad’ for simulati<strong>on</strong> in Figs 21 and<br />

23.<br />

c<strong>on</strong>sidered to be ‘bad’ will need to be modified if multiple rows/columns al<strong>on</strong>g the gap are<br />

faulty, although the change is not dramatic. <str<strong>on</strong>g>The</str<strong>on</strong>g> changes to the median and maximum in<br />

each case are fairly insignificant.<br />

A third test was performed to check the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> a c<strong>on</strong>tinguous block <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘bad’ bolometers. In<br />

the following two simulati<strong>on</strong>s, a 10×10 area in the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays was flagged as<br />

‘bad’ and the hits-per-pixel were again measured. <str<strong>on</strong>g>The</str<strong>on</strong>g> setups are shown in Figs. 18 and 19<br />

and the resulting sampling curves are shown in Figs. 20 and 21.<br />

Let us focus <strong>on</strong> a specific scan angle, say 30 ◦ . Comparing Figs. 22 and 23 with our earlier scan<br />

which had no dead bolometers (Fig. 10) we can see the immediate effect <str<strong>on</strong>g>of</str<strong>on</strong>g> these bad regi<strong>on</strong>s.<br />

While the minimum hits-per-pixel at lower angles does not differ drastically from the results<br />

with no dead bolometers, at greater angles (especially above 20.0 ◦ ) the difference is more<br />

noticeable. In such a scenario, where blocks <str<strong>on</strong>g>of</str<strong>on</strong>g> dead detectors are present, the scan angles<br />

should be chosen such that these blocks cover the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> each scan path, and are thus<br />

compensated for by the multiple sub-array geometry. Alternately, the spacing (or ‘overlap’) <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

the scan paths could be adjusted to fill in the less-sampled regi<strong>on</strong>s.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> c<strong>on</strong>clusi<strong>on</strong> from this part <str<strong>on</strong>g>of</str<strong>on</strong>g> our study is that modest-sized regi<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘bad’ bolometers will<br />

have no dramatic effect <strong>on</strong> the choice <str<strong>on</strong>g>of</str<strong>on</strong>g> scan angle. Clearly, if there are large ‘dead’ regi<strong>on</strong>s<br />

then <strong>on</strong>e would have to be more careful about scan angle choice and scan overlaps (which we<br />

discuss in the next secti<strong>on</strong>), although this will be less <str<strong>on</strong>g>of</str<strong>on</strong>g> an issue for full P<strong>on</strong>g scan patterns.<br />

Page 12 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 6 THE IMPACT OF FAULTY BOLOMETERS<br />

Figure 20: Simulati<strong>on</strong> with a c<strong>on</strong>tiguous 10×10 block <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘bad’ bolometers in the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>on</strong>e sub-array<br />

(see Fig. 18). Minimum, median, and maximum hits are plotted as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan angle.<br />

Figure 21: Simulati<strong>on</strong> with a c<strong>on</strong>tiguous 10×10 block <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘bad’ bolometers in the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> all four<br />

sub-arrays (see Fig. 19). Minimum, median, and maximum hits are plotted as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan angle.<br />

Page 13 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 6 THE IMPACT OF FAULTY BOLOMETERS<br />

Figure 22: Hits-per-pixel for a scan including <strong>on</strong>e array with a 10×10 regi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘bad’ bolometers in <strong>on</strong>e<br />

sub-array. <str<strong>on</strong>g>The</str<strong>on</strong>g> scan angle here is 30.0 ◦ . This can be compared with Fig. 10, where there were no ‘bad’<br />

bolometers.<br />

Figure 23: Hits-per-pixel for a scan including 10×10 regi<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘bad’ bolometers in all four subarrays(see<br />

Fig. 19). <str<strong>on</strong>g>The</str<strong>on</strong>g> scan angle here is 30.0 ◦ .<br />

Page 14 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 7 SPACING OF PARALLEL SCAN PATHS<br />

7 Spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> parallel scan paths<br />

In the scanning strategies proposed for SCUBA-2, not <strong>on</strong>ly is it necessary to c<strong>on</strong>sider the<br />

angle <str<strong>on</strong>g>of</str<strong>on</strong>g> the paths across the sky, but also how close together these paths are when filling an<br />

area for which both dimensi<strong>on</strong>s exceed the field <str<strong>on</strong>g>of</str<strong>on</strong>g> view <str<strong>on</strong>g>of</str<strong>on</strong>g> the array. A simple Boustrophed<strong>on</strong><br />

scan was simulated (with no bolometers flagged as ‘bad’) while varying the spacings between<br />

the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan paths from 100–400 arcsec<strong>on</strong>ds.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g>se simulati<strong>on</strong>s were c<strong>on</strong>ducted at angles <str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ , 30 ◦ , and 45 ◦ relative to the orientati<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays. As the angle approaches 45 ◦ , the gradient in sampling towards the outer<br />

edge <str<strong>on</strong>g>of</str<strong>on</strong>g> each individual scan path becomes wider and more gradual, while a wider total area<br />

is covered in each single path.<br />

As seen in Figs. 24 through 35, lower angles will result in maps with more abrupt changes in<br />

sampling rate, caused by the overlap <str<strong>on</strong>g>of</str<strong>on</strong>g> individual sub-arrays. Care should be taken to avoid<br />

scans in which the centres <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-arrays are ‘lined-up’ in successive paths across the sky<br />

(as seen in Fig. 26), if the goal <str<strong>on</strong>g>of</str<strong>on</strong>g> even coverage across the map is to be achieved. Ideally, the<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g>fset between scan paths should be approximately half the width <str<strong>on</strong>g>of</str<strong>on</strong>g> an individual sub-array<br />

in order to fill in the less sampled areas <str<strong>on</strong>g>of</str<strong>on</strong>g> the previous paths. One can see this by comparing<br />

the sampling for <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets <str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds (about half the size <str<strong>on</strong>g>of</str<strong>on</strong>g> a sub-array) versus the<br />

sampling for larger <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets. <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fset which makes the sampling most uniform will depend<br />

<strong>on</strong> scan angle, but <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets around half the sub-array width are generally good.<br />

Figure 24: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 75 arcsec<strong>on</strong>ds.<br />

Figure 25: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds.<br />

Figure 26: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 250 arcsec<strong>on</strong>ds.<br />

Figure 27: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> arcsec<strong>on</strong>ds.<br />

Statistics extracted from Figs. 24–35 are shown in Figs. 36, 37, and 38, for scan angles <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

15 ◦ , 30 ◦ , and 45 ◦ , respectively. <str<strong>on</strong>g>The</str<strong>on</strong>g> variati<strong>on</strong> in sampling as a functio fo <str<strong>on</strong>g>of</str<strong>on</strong>g>fset (the patterns<br />

apparent in Figs. 24–35) are clearly seen here. <str<strong>on</strong>g>The</str<strong>on</strong>g> absolute values are not important when<br />

comparing different <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets (since the maps may cover different areas for different <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets) –<br />

but the important discriminator is the spread am<strong>on</strong>g the minimum, median and maximum<br />

hit rates. One can see that there is little relative difference for scan spacings around 100-150<br />

arcsec<strong>on</strong>ds, and the spread is much larger for <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets above about 200 arcsec<strong>on</strong>ds (i.e. roughly<br />

a sub-array size).<br />

Page 15 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 7 SPACING OF PARALLEL SCAN PATHS<br />

Figure 28: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 75 arcsec<strong>on</strong>ds.<br />

Figure 29: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds.<br />

Figure 30: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 250 arcsec<strong>on</strong>ds.<br />

Figure 31: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> arcsec<strong>on</strong>ds.<br />

Figure 32: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 75 arcsec<strong>on</strong>ds.<br />

Figure 33: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds.<br />

Figure 34: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 250 arcsec<strong>on</strong>ds.<br />

Figure 35: Boustrophed<strong>on</strong> scan with an angle<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> arcsec<strong>on</strong>ds.<br />

Page 16 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 7 SPACING OF PARALLEL SCAN PATHS<br />

Figure 36: Minimum (bottom line), median (middle line), and maximum (top line) hits-perpixel<br />

as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan spacing for a 15 ◦ Boustrophed<strong>on</strong> scan.<br />

Figure 37: Minimum, median, and maximum hits-per-pixel as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan spacing for<br />

a 30 ◦ Boustrophed<strong>on</strong> scan.<br />

Page 17 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 38: Minimum, median, and maximum hits-per-pixel as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan spacing for<br />

a 45 ◦ Boustrophed<strong>on</strong> scan.<br />

8 P<strong>on</strong>g scanning for uniform sky coverage<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> P<strong>on</strong>g scanning soluti<strong>on</strong> proposed for SCUBA-2 should take into c<strong>on</strong>siderati<strong>on</strong> the gaps,<br />

faulty bolometers, and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> the scans (see Fig. 4), while striving to cover large maps in<br />

an efficient manner. <str<strong>on</strong>g>The</str<strong>on</strong>g> cross-linking pattern will not <strong>on</strong>ly aid in sky-removal, but will also<br />

reduce the deviati<strong>on</strong> in hits-per-pixel, thus providing more uniform sky coverage.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> P<strong>on</strong>g pattern covers the chosen map area by scanning in paths 45 ◦ from the sides <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

the ‘box’. As such, when avoiding ‘bad angles’, this 45 ◦ correcti<strong>on</strong> will need to be applied.<br />

For example, if we wish to avoid the ‘bad angle’ range <str<strong>on</strong>g>of</str<strong>on</strong>g> 0–4 ◦ as found in Secti<strong>on</strong> 5, the<br />

corresp<strong>on</strong>ding ‘bad’ P<strong>on</strong>g angles will be 41–49 ◦ (45±4 ◦ ).<br />

Figures 39 through 54 show the results <str<strong>on</strong>g>of</str<strong>on</strong>g> a similar study to that d<strong>on</strong>e for Boustrophed<strong>on</strong><br />

scans in Secti<strong>on</strong> 7. It is important to notice the dynamic range <str<strong>on</strong>g>of</str<strong>on</strong>g> these plots, even when<br />

the pattern looks dramatic (<strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> these should be adopted as the <str<strong>on</strong>g>of</str<strong>on</strong>g>ficial SCUBA-2 tartan!).<br />

In this case we ran several P<strong>on</strong>g simulati<strong>on</strong>s at three P<strong>on</strong>g angles, 15 ◦ , 30 ◦ , and 45 ◦ , while<br />

again varying the spacing between 100 and 400 arcsec<strong>on</strong>ds. By comparing Figs. 55, 56, 57<br />

and 58 (where we show statistics for sampling rates), we can see how a well-chosen P<strong>on</strong>g<br />

angle <str<strong>on</strong>g>of</str<strong>on</strong>g> say 15 ◦ (i.e. 30 ◦ from being parallel to the gap) does better even at wider <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets<br />

between the sweeps, effectively ‘filling in’ a larger area with each pass across the map. We<br />

can also see that the hits-per-pixel varies much more at a ‘bad’ angle (e.g. 45 ◦ , Fig 58), even<br />

at smaller spacings when every pixel is hit at least <strong>on</strong>ce. This suggests that provided <strong>on</strong>e<br />

ignores explicitly ‘bad’ angles (P<strong>on</strong>g angles around 45 ◦ ) the P<strong>on</strong>g pattern gives fairly uniform<br />

coverage. It will never be exactly uniform, <str<strong>on</strong>g>of</str<strong>on</strong>g> course, but the c<strong>on</strong>trast between minimum and<br />

maximum hits is very modest, particularly for pattern spacings <str<strong>on</strong>g>of</str<strong>on</strong>g> around half a sub-array<br />

width.<br />

Page 18 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 39: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 0 ◦ (45 ◦ relative to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 75 arcsec<strong>on</strong>ds.<br />

Figure 40: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 0 ◦ (45 ◦ relative to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds.<br />

Figure 41: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 0 ◦ (45 ◦ relative to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 250 arcsec<strong>on</strong>ds.<br />

Figure 42: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 0 ◦ (45 ◦ relative to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 400 arcsec<strong>on</strong>ds.<br />

Figure 43: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ (30 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 75 arcsec<strong>on</strong>ds.<br />

Figure 44: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ (30 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds.<br />

Page 19 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 45: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ (30 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 250 arcsec<strong>on</strong>ds.<br />

Figure 46: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 15 ◦ (30 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 400 arcsec<strong>on</strong>ds.<br />

Figure 47: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ (15 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 75 arcsec<strong>on</strong>ds.<br />

Figure 48: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ (15 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds.<br />

Figure 49: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ (15 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 250 arcsec<strong>on</strong>ds.<br />

Figure 50: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 30 ◦ (15 ◦ relative<br />

to the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 400 arcsec<strong>on</strong>ds.<br />

Page 20 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 51: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ (parallel to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 75 arcsec<strong>on</strong>ds.<br />

Figure 52: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ (parellel to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 100 arcsec<strong>on</strong>ds.<br />

Figure 53: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ (parallel to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 250 arcsec<strong>on</strong>ds.<br />

Figure 54: Hits-per-pixel <str<strong>on</strong>g>of</str<strong>on</strong>g> the central regi<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g scan at an angle <str<strong>on</strong>g>of</str<strong>on</strong>g> 45 ◦ (parallel to<br />

the sub-array geometry) and pattern spacing<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> 400 arcsec<strong>on</strong>ds.<br />

Figure 55: Minimum (bottom line), median (middle line), and maximum (top line) hits-perpixel<br />

as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan spacing for a 0 ◦ P<strong>on</strong>g scan.<br />

Page 21 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 56: Minimum (bottom line), median (middle line), and maximum (top line) hits-perpixel<br />

as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan spacing for a 15 ◦ P<strong>on</strong>g scan.<br />

Figure 57: Minimum (bottom line), median (middle line), and maximum (top line) hits-perpixel<br />

as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan spacing for a 30 ◦ P<strong>on</strong>g scan.<br />

Page 22 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 58: Minimum (bottom line), median (middle line), and maximum (top line) hits-perpixel<br />

as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> scan spacing for a 45 ◦ P<strong>on</strong>g scan.<br />

From Secti<strong>on</strong>s 5 and 7 our tests suggest that scans with angles <str<strong>on</strong>g>of</str<strong>on</strong>g> 8–33 ◦ relative to the gaps<br />

between the sub-arrays, and with a spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> approximately half the width <str<strong>on</strong>g>of</str<strong>on</strong>g> a subarray<br />

should yield a fairly uniform sampling rate across the map. So we now study four P<strong>on</strong>g scans<br />

to compare the uniformity <str<strong>on</strong>g>of</str<strong>on</strong>g> the hits-per-pixel across 4 maps made by fulfilling neither, <strong>on</strong>e,<br />

or both <str<strong>on</strong>g>of</str<strong>on</strong>g> these criteria ( Figs. 59 and 60, Figs. 61–64, and Figs. 65 and 66 respectively). When<br />

choosing ‘bad’ and ‘good’ spacing, we c<strong>on</strong>sider that given a bolometer width <str<strong>on</strong>g>of</str<strong>on</strong>g> 6 arcsec<strong>on</strong>ds,<br />

the width <str<strong>on</strong>g>of</str<strong>on</strong>g> a sub-array is either 240 arcsec<strong>on</strong>ds (when measured across the l<strong>on</strong>g edge) or 192<br />

arcsec<strong>on</strong>ds (when measured across the short edge). In additi<strong>on</strong>, we must bear in mind that<br />

the centres <str<strong>on</strong>g>of</str<strong>on</strong>g> two side-by-side sub-arrays are separated by a further 24 arcsec<strong>on</strong>ds due to<br />

the gaps between the arrays. For this simulati<strong>on</strong>, we have chosen 108 arcsec<strong>on</strong>ds as a ‘good’<br />

spacing (calculated using half the shorter sub-array width plus half the gap width). Scans at<br />

this spacing will ensure that the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> a sub-array passes directly over the undersampled<br />

regi<strong>on</strong> caused by the gap in a previous sweep <str<strong>on</strong>g>of</str<strong>on</strong>g> the pattern. <str<strong>on</strong>g>The</str<strong>on</strong>g> ‘bad’ spacing used here<br />

is 264 arcsec<strong>on</strong>ds, a spacing which causes the sub-arrays to repeat similar paths across the<br />

map and leave undersampled regi<strong>on</strong>s in between (this value is calculated using the width <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

the l<strong>on</strong>ger side <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-array plus the width <str<strong>on</strong>g>of</str<strong>on</strong>g> the gap). <str<strong>on</strong>g>The</str<strong>on</strong>g> ‘good’ P<strong>on</strong>g angle used is 25 ◦ ,<br />

which results in scan paths 20 ◦ from parallel to the sub-array geometry, while the ‘bad’ P<strong>on</strong>g<br />

angle used is 45 ◦ , which results in scan paths parallel to the sub-array geometry. Each scan<br />

has an approximate height and width <str<strong>on</strong>g>of</str<strong>on</strong>g> 1000 arcsec<strong>on</strong>ds.<br />

In order to compare the drastic difference in sampling variati<strong>on</strong> am<strong>on</strong>g these P<strong>on</strong>g scans, we<br />

can take a slice across the centre <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan (parallel to <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the gaps) and measure the<br />

hits-per-pixel at each point. <str<strong>on</strong>g>The</str<strong>on</strong>g> specific slices shown in Figs. 60, 62, 64, and 66 are taken<br />

across the hits-per-pixel maps such that the greatest variati<strong>on</strong> in hits-per-pixel is measured<br />

in each case.<br />

Page 23 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 59: P<strong>on</strong>g scan using a ‘bad’ angle and ‘bad’ spacing.<br />

Figure 60: ‘Slice’ across P<strong>on</strong>g scan in Fig. 59 showing wide variati<strong>on</strong>s in hits-per-pixel due to<br />

the ‘bad’ angle and spacing.<br />

Figure 61: P<strong>on</strong>g scan using a ‘good’ angle and ‘bad’ spacing.<br />

Page 24 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 62: ‘Slice’ across P<strong>on</strong>g scan in Fig. 61 showing variati<strong>on</strong>s in hits-per-pixel due to the<br />

‘bad’ spacing, but with a ‘good’ angle.<br />

Figure 63: P<strong>on</strong>g scan using a ‘bad’ angle and ‘good’ spacing.<br />

Figure 64: ‘Slice’ across P<strong>on</strong>g scan in Fig. 63 showing variati<strong>on</strong>s in hits-per-pixel due to the<br />

‘bad’ angle, but with ‘good’ spacing.<br />

Page 25 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 8 PONG SCANNING FOR UNIFORM SKY COVERAGE<br />

Figure 65: P<strong>on</strong>g scan using a ‘good’ angle and ‘good’ spacing.<br />

Figure 66: ‘Slice’ across P<strong>on</strong>g scan in Fig. 65 showing the relatively uniform coverage in the<br />

central regi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the map when <strong>on</strong>e choses a ‘good’ angle and ‘good’ spacing.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g>se simulati<strong>on</strong>s c<strong>on</strong>firm that both the angle and the spacing play an important role in ensuring<br />

uniform sampling across a scanned map. It would appear from these four simulati<strong>on</strong>s<br />

that the angle has greater impact <strong>on</strong> the c<strong>on</strong>trast in hits-per-pixel than the spacing. However,<br />

this will clearly depend <strong>on</strong> how bad a ‘bad’ spacing is – clearly the coverage would be quite<br />

ugly if the separati<strong>on</strong> was chosen to be larger that the full array size, for example. Additi<strong>on</strong>ally,<br />

while the deviati<strong>on</strong> in hits-per-pixel from the average in a scan with a ‘bad’ angle and a<br />

‘good’ spacing is greater than that observed with a ‘good’ angle and ‘bad’ spacing, it is should<br />

be noted that for the former, the values for the minimum hits-per-pixel across the majority<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the map are significantly higher. In the case <str<strong>on</strong>g>of</str<strong>on</strong>g> this ‘good’ spacing, every pixel is hit approximately<br />

twice as many times as in the case <str<strong>on</strong>g>of</str<strong>on</strong>g> a ‘bad’ spacing due to the overlap <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

array paths in the sweeps across the map. <str<strong>on</strong>g>The</str<strong>on</strong>g> reas<strong>on</strong> for this lies in the time taken to cover<br />

the entire pattern. With a wider spacing (less overlap) the map is covered in a shorter time,<br />

resulting in lower hit counts <strong>on</strong> average across the whole map. <str<strong>on</strong>g>The</str<strong>on</strong>g> important feature to note<br />

in these simulati<strong>on</strong>s is not the minimum value, but the relative difference in the maximum<br />

and minimum hits-per-pixel as compared with the average for each scan.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> basic c<strong>on</strong>clusi<strong>on</strong> from this secti<strong>on</strong> is that P<strong>on</strong>g scans are quite forgiving compared with<br />

Boustrophed<strong>on</strong>. Precisely uniform maps will be impossible to achieve, because <str<strong>on</strong>g>of</str<strong>on</strong>g> the gaps<br />

between the SCUBA-2 sub-arrays (as well as n<strong>on</strong>-functi<strong>on</strong>ing detectors <str<strong>on</strong>g>of</str<strong>on</strong>g> course). However,<br />

maps which have variati<strong>on</strong>s in pixel hit rates <str<strong>on</strong>g>of</str<strong>on</strong>g> ∼ 20% are relatively easy to achieve with<br />

P<strong>on</strong>g, privided that the spacing between sweeps is about half a sub-array size (say 100–150<br />

arcsec<strong>on</strong>ds), and that <strong>on</strong>e tries to avoid scanning at angles <str<strong>on</strong>g>of</str<strong>on</strong>g> around 45±7 ◦ relative to the<br />

array axes.<br />

Page 26 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 9 SCANNING SOLUTIONS AT THE TELESCOPE<br />

9 <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> soluti<strong>on</strong>s at the telescope<br />

When choosing a scanning strategy, it will be important to c<strong>on</strong>sider the gap between subarrays,<br />

any areas <str<strong>on</strong>g>of</str<strong>on</strong>g> faulty bolometers, and the scanning angle, in order to guarantee a map<br />

with relatively uniform sky coverage. Users <str<strong>on</strong>g>of</str<strong>on</strong>g> this instrument should be able to describe an<br />

area to be mapped and be assured that their observati<strong>on</strong> will be close to the best possible<br />

given the array quality, map dimensi<strong>on</strong>s, and time c<strong>on</strong>straints.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> s<str<strong>on</strong>g>of</str<strong>on</strong>g>tware for SCUBA-2 should be resp<strong>on</strong>sible for checking the upcoming observati<strong>on</strong>s<br />

against some minimum acceptable criteria. Each scanning observati<strong>on</strong> should either have<br />

multiple possible soluti<strong>on</strong>s, from which the best is chosen at the time <str<strong>on</strong>g>of</str<strong>on</strong>g> the observati<strong>on</strong>,<br />

or the scheduling system should be able to change the order <str<strong>on</strong>g>of</str<strong>on</strong>g> observati<strong>on</strong>s to avoid poor<br />

scanning.<br />

Example 1) Given a P<strong>on</strong>g observati<strong>on</strong> for a particular area <str<strong>on</strong>g>of</str<strong>on</strong>g> sky, two different soluti<strong>on</strong>s<br />

could be c<strong>on</strong>structed at different angles, with a choice between the two made at the time <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

observati<strong>on</strong> such that the better angle is used. C<strong>on</strong>sider a user who has defined a rectangular<br />

regi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the sky to be mapped. Two P<strong>on</strong>g patterns could be prepared with two different<br />

angles which both include the required regi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> sky. Provided these two angles are no less<br />

14 ◦ apart, we are guaranteed that at least <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the scans will have a pattern with sweeps<br />

which are greater than 7 ◦ from being parallel to the sub-array geometry. Compare Fig. 4 with<br />

Fig. 67 for examples <str<strong>on</strong>g>of</str<strong>on</strong>g> two possible P<strong>on</strong>g patterns which could be c<strong>on</strong>structed to map an area<br />

approximately 3000 × 3000 arcsec<strong>on</strong>ds in size. Given the orientati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the regi<strong>on</strong> relative<br />

to the telescope axes at the time <str<strong>on</strong>g>of</str<strong>on</strong>g> the observati<strong>on</strong>, the pattern which generates the better<br />

angle <str<strong>on</strong>g>of</str<strong>on</strong>g> the two would be used.<br />

Figure 67: ‘Straight’ P<strong>on</strong>g scan with a map width ≃4000 arcsec, height ≃3200 arcsec, and sweep<br />

spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 500 arcsec. <str<strong>on</strong>g>The</str<strong>on</strong>g> P<strong>on</strong>g angle here is 15 ◦ .<br />

Example 2) If <strong>on</strong>ly <strong>on</strong>e soluti<strong>on</strong> is available the scheduling system will identify if this is a<br />

‘bad’ angle and reschedule the scan for a time when the sky has rotated such that the angle<br />

would no l<strong>on</strong>ger be ‘bad’.<br />

Page 27 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps 10 SKY ROTATIONS<br />

10 <strong>Sky</strong> Rotati<strong>on</strong>s<br />

One important factor to c<strong>on</strong>sider in larger maps is the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> sky rotati<strong>on</strong> <strong>on</strong> the scanning<br />

pattern. Over time, the orientati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the array relative to the area to be mapped will clearly<br />

change. <str<strong>on</strong>g>The</str<strong>on</strong>g> sky rotati<strong>on</strong> will modify the results <str<strong>on</strong>g>of</str<strong>on</strong>g> the P<strong>on</strong>g pattern in two ways :<br />

1. As the sky rotates, the paths <str<strong>on</strong>g>of</str<strong>on</strong>g> the P<strong>on</strong>g pattern will curve slightly relative to the sky.<br />

On l<strong>on</strong>g and/or large scans, the sweeps <str<strong>on</strong>g>of</str<strong>on</strong>g> the P<strong>on</strong>g scan will gradually move away from<br />

being orthog<strong>on</strong>al, effectively ‘filling in’ the periodic variati<strong>on</strong> in pixel sampling due to the<br />

parallel sweeps <str<strong>on</strong>g>of</str<strong>on</strong>g> the P<strong>on</strong>g pattern. <str<strong>on</strong>g>The</str<strong>on</strong>g> can be seen in Figs 68–71. In this simulati<strong>on</strong>,<br />

the array performed a sequence <str<strong>on</strong>g>of</str<strong>on</strong>g> 4 P<strong>on</strong>g scans while the sky rotated. Obviously, the<br />

amount by which the P<strong>on</strong>g scan pattern changes relative the sky is dependent <strong>on</strong> where<br />

in the sky we are observing, as well as the P<strong>on</strong>g parameters such as map dimensi<strong>on</strong>s<br />

and the velocity <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan.<br />

2. Over time, the sky rotates relative to the orientati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the SCUBA-2 arrays. Unless we<br />

are P<strong>on</strong>ging in Nasmyth coordinates, sky rotati<strong>on</strong> can be used to overcome the problems<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the ‘bad’ angles. If we find that a planned scan would be at a ‘bad’ angle, simply<br />

waiting until the sky rotates should be sufficient to find a time at which the orientati<strong>on</strong><br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the desired pattern relative to the arrays does not cause the angle to be ‘bad’.<br />

Figure 68: P<strong>on</strong>g scan pattern, showing the effects<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> sky rotati<strong>on</strong>. Here, approximately <strong>on</strong>e<br />

P<strong>on</strong>g pattern has completed.<br />

Figure 69: P<strong>on</strong>g scan pattern, showing the effects<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> sky rotati<strong>on</strong>. Here, approximately two<br />

P<strong>on</strong>g patterns have completed.<br />

Figure 70: P<strong>on</strong>g scan pattern, showing the<br />

effects <str<strong>on</strong>g>of</str<strong>on</strong>g> sky rotati<strong>on</strong>. Here, approximately<br />

three P<strong>on</strong>g patterns have completed.<br />

Figure 71: P<strong>on</strong>g scan pattern, showing the effects<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> sky rotati<strong>on</strong>. Here, approximately four<br />

P<strong>on</strong>g patterns have completed.<br />

Page 28 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps A BOUSTROPHEDON SCAN DESCRIPTION<br />

11 C<strong>on</strong>clusi<strong>on</strong>s<br />

Throughout this study we have used the SCUBA-2 simulator to examine some <str<strong>on</strong>g>of</str<strong>on</strong>g> the anticipated<br />

scanning issues inherent in the sub-array geometry <str<strong>on</strong>g>of</str<strong>on</strong>g> SCUBA-2. We found that scan<br />

angles less that 4 ◦ from parallel to the sub-array geometry leave some regi<strong>on</strong>s <strong>on</strong> each path<br />

completely unsampled, and that angles above about 7 ◦ from parallel to the sub-array geometry<br />

fill the ‘gap’ between the sub-arrays and provide relatively even sampling across the scan<br />

path. Simulati<strong>on</strong>s suggest that most scanning strategies should be able to adequately compensate<br />

for the predicted number <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘dead’ bolometers. A similar study should be c<strong>on</strong>ducted<br />

<strong>on</strong>ce the actual pattern <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘dead’ bolometers <strong>on</strong> the arrays <str<strong>on</strong>g>of</str<strong>on</strong>g> SCUBA-2 is known, especially if<br />

the total number <str<strong>on</strong>g>of</str<strong>on</strong>g> unusable bolometers is much higher than predicted. We also simulated<br />

two proposed scanning patterns, Boustrophed<strong>on</strong> and P<strong>on</strong>g, with a variety <str<strong>on</strong>g>of</str<strong>on</strong>g> pattern spacings<br />

and angles and found that P<strong>on</strong>g should be able to provide a high level <str<strong>on</strong>g>of</str<strong>on</strong>g> sampling uniformity<br />

across a mapped regi<strong>on</strong> provided that care is taken in choosing the angles and spacing.<br />

SCUBA-2 will be an important tool for mapping large areas <str<strong>on</strong>g>of</str<strong>on</strong>g> the submillimetre sky with<br />

high sensitivity in relatively short periods <str<strong>on</strong>g>of</str<strong>on</strong>g> time. In order to exploit this key science goal,<br />

an understanding <str<strong>on</strong>g>of</str<strong>on</strong>g> the sub-array geometry and proposed scanning soluti<strong>on</strong>s is essential.<br />

Once SCUBA-2 is operati<strong>on</strong>al, the full impact <str<strong>on</strong>g>of</str<strong>on</strong>g> these criteria will be better understood.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> layout <str<strong>on</strong>g>of</str<strong>on</strong>g> faulty bolometers at any given time, as well as the ability <str<strong>on</strong>g>of</str<strong>on</strong>g> the telescope to<br />

move in the planned patterns will need to be c<strong>on</strong>sidered when evaluating the success <str<strong>on</strong>g>of</str<strong>on</strong>g> these<br />

proposed scanning strategies. Prior to commisi<strong>on</strong>ing, however, it should be possible to test<br />

explicit scanning patterns at the JCMT to understand whether some <str<strong>on</strong>g>of</str<strong>on</strong>g> the results presented<br />

here might change under real observing c<strong>on</strong>diti<strong>on</strong>s.<br />

A<br />

Boustrophed<strong>on</strong> Scan Descripti<strong>on</strong><br />

This is essentially the existing SCUBA box scan – a rectangular regi<strong>on</strong> is filled in by a series<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> parallel scans.<br />

Pros:<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> orientati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle and the inclinati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan to the rectangle are independent<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>on</strong>e-another. This scan will be the preferred method in cases such as those outlined in<br />

the c<strong>on</strong>s secti<strong>on</strong> for P<strong>on</strong>g. In any instance where the scan directi<strong>on</strong> is highly c<strong>on</strong>strained, this<br />

method is the most efficient for covering rectangular regi<strong>on</strong>s defined in arbitrary coordinate<br />

systems.<br />

C<strong>on</strong>s:<br />

A single map with boustrophed<strong>on</strong> will not naturally c<strong>on</strong>tain any cross-linking (there is no<br />

telescope moti<strong>on</strong> in the directi<strong>on</strong> transverse to the scan apart from the step and turnaround<br />

at the end <str<strong>on</strong>g>of</str<strong>on</strong>g> each row). <str<strong>on</strong>g>The</str<strong>on</strong>g> flexibility to choose arbitrary scan angles makes it easy to<br />

subsequently cover the regi<strong>on</strong> with another boustrophed<strong>on</strong> but with a scan angle orthog<strong>on</strong>al<br />

to the first pass. Even in this case, the cross-linking produced by P<strong>on</strong>g is superior since there<br />

is greater modulati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the time-scales over which each point in the map is visited.<br />

Input parameters:<br />

width<br />

= width <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle in degrees<br />

Page 29 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps B PONG SCAN DESCRIPTION<br />

height<br />

mapangle<br />

scanangle<br />

spacing<br />

velocity<br />

= height <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle in degrees<br />

= positi<strong>on</strong> angle <str<strong>on</strong>g>of</str<strong>on</strong>g> the box in the native coordinate system<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the regi<strong>on</strong> (Nasmyth, AzEl or RADec)<br />

= positi<strong>on</strong> angle <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan in the native coordinate system<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the regi<strong>on</strong> (it may in fact be defined in some other<br />

coordinate system, and then c<strong>on</strong>verted using high-level code).<br />

= space between adjacent (parallel) scan lines<br />

Ideally this should be set to some fracti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

SCUBA-2 footprint such that fully-sampled maps maybe be<br />

c<strong>on</strong>structed in a single pass.<br />

= target magnitude <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan velocity excluding<br />

turn-arounds<br />

A Boustrophed<strong>on</strong> scan called ‘scanmap’ was implemented for SCUBA. Prestage et al. 10, 9 have<br />

documented and provided pseud-ocode for Boustrophed<strong>on</strong>-like raster mapping strategies.<br />

B<br />

P<strong>on</strong>g Scan Descripti<strong>on</strong><br />

Descripti<strong>on</strong>:<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> telescope covers a rectangular regi<strong>on</strong> by scanning at angles inclined 45 ◦ to the fundamental<br />

axes <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle, and when a side is encountered the angular velocity is reflected. By<br />

choosing the starting positi<strong>on</strong> and spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> nodes (points <str<strong>on</strong>g>of</str<strong>on</strong>g> reflecti<strong>on</strong>) carefully, the telescope<br />

covers the entire map, and the path taken is closed (i.e. eventually repeats itself). Note<br />

that in principle <strong>on</strong>e can generalize this to scanning angles other than 45 ◦ (resulting in ‘cells’<br />

which are diam<strong>on</strong>ds rather than squares), but we do not c<strong>on</strong>sider this further here, since it<br />

would required a different algorithm than given in the pseudo-code below.<br />

Pros:<br />

This method automatically produces the best cross-linking <str<strong>on</strong>g>of</str<strong>on</strong>g> scans currently planned for<br />

implementati<strong>on</strong> <strong>on</strong> the JCMT. Every point <strong>on</strong> the map is visited <strong>on</strong> different time-scales,<br />

with the telescope scanning in orthog<strong>on</strong>al directi<strong>on</strong>s. This scan should generally be the best<br />

method for SCUBA-2.<br />

C<strong>on</strong>s:<br />

Since the scan angle is fixed with respect to the axes <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangular regi<strong>on</strong>, under certain<br />

special circumstances, P<strong>on</strong>g will not be very efficient. If, for example, it is discovered that<br />

the JCMT has a problem with moti<strong>on</strong> in certain directi<strong>on</strong>s (such as induced vibrati<strong>on</strong>s in the<br />

carousel and membrane as was seen with AzTEC while performing azimuthal scans) the scan<br />

angle, and hence box orientati<strong>on</strong> could be severely c<strong>on</strong>strained. In such cases, the rectangles<br />

can not be tiled in such a way as to optimally cover huge areas (such as for SASSy), since<br />

they will have random orientati<strong>on</strong>s <strong>on</strong>ce projected <strong>on</strong>to the sky.<br />

Input parameters:<br />

width<br />

height<br />

angle<br />

= width <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle in degrees<br />

= height <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle in degrees<br />

= positi<strong>on</strong> angle <str<strong>on</strong>g>of</str<strong>on</strong>g> the box in the native coordinate system<br />

Page 30 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps C PSEUDO-CODE FOR ‘STRAIGHT’ PONG SCAN PATTERN<br />

spacing<br />

velocity<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> the regi<strong>on</strong> (Nasmyth, AzEl or RADec - note that Galactic<br />

coordinates and RADec have a static transformati<strong>on</strong> between them,<br />

so a box defined in a high-level descripti<strong>on</strong> in Galactic<br />

coordinates may be transformed to an RADec descripti<strong>on</strong> before<br />

calling the low-level P<strong>on</strong>g scan routine).<br />

= space between adjacent (parallel) scan lines in the P<strong>on</strong>g<br />

pattern. Ideally this should be set to some fracti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

SCUBA-2 footprint, such that fully-sampled maps maybe be<br />

c<strong>on</strong>structed in a single pass.<br />

= target magnitude <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan velocity excluding<br />

turn-arounds<br />

C<br />

Pseudo-code for ‘Straight’ P<strong>on</strong>g Scan <str<strong>on</strong>g>Pattern</str<strong>on</strong>g><br />

; Determine number <str<strong>on</strong>g>of</str<strong>on</strong>g> vertices (reflecti<strong>on</strong> points) al<strong>on</strong>g each side <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

; box which satisfies the comm<strong>on</strong>-factors criteri<strong>on</strong> and the requested<br />

; size / spacing<br />

; Find out how far apart the vertices are al<strong>on</strong>g the axes<br />

vert_spacing = ( 2.0 * spacing ) / sqrt ( 2.0 )<br />

; Determine how many vertices (minimum) there must be al<strong>on</strong>g<br />

; each axis to cover the required area<br />

x_numvert = ceil ( width / vert_spacing )<br />

y_numvert = ceil ( height / vert_spacing )<br />

; Determine which is lower and check to make sure that <strong>on</strong>e is<br />

; even while the other is odd<br />

if ( x_numvert >= y_numvert ) begin<br />

most = x_numvert<br />

least = y_numvert<br />

else begin<br />

most = y_numvert<br />

least = x_numvert<br />

end<br />

; If both are odd or both are even, increment the lesser <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

; the two, and update which is least<br />

if ( ( x_numvert % 2 ) == (y_numvert % 2 ) ) begin<br />

least += 1;<br />

if ( x_numvert >= y_numvert ) begin<br />

most = x_numvert<br />

least = y_numvert<br />

else begin<br />

most = y_numvert<br />

least = x_numvert<br />

end<br />

end<br />

Page 31 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps C PSEUDO-CODE FOR ‘STRAIGHT’ PONG SCAN PATTERN<br />

; Check for comm<strong>on</strong> factors between the two, and adjust as<br />

; necessary until x_numvert and y_numvert do not share any<br />

; comm<strong>on</strong> factors<br />

for ( i = 3; i


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps C PSEUDO-CODE FOR ‘STRAIGHT’ PONG SCAN PATTERN<br />

ygrid[i] = (i - y_ngridseg/2)*grid_space<br />

end<br />

; Initializati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan pattern<br />

x_init = 0<br />

y_init = 0<br />

x_<str<strong>on</strong>g>of</str<strong>on</strong>g>f = x_init<br />

y_<str<strong>on</strong>g>of</str<strong>on</strong>g>f = y_init<br />

; starting grid coordinates<br />

; current grid <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets<br />

grid[0,0] = xgrid[x_init] ; starting tanplane <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan<br />

grid[0,1] = ygrid[y_init]<br />

steps = 0<br />

; count number <str<strong>on</strong>g>of</str<strong>on</strong>g> steps taken in the pattern<br />

; Loop over line segments and calculate list <str<strong>on</strong>g>of</str<strong>on</strong>g> endpoints for each<br />

; reflecti<strong>on</strong> in order<br />

for i=0 to n_seg begin<br />

; increment steps to the next boundary reflecti<strong>on</strong> (whichever comes<br />

; first - al<strong>on</strong>g the sides or top/bottom <str<strong>on</strong>g>of</str<strong>on</strong>g> the rectangle)<br />

if( (x_ngridseg-x_<str<strong>on</strong>g>of</str<strong>on</strong>g>f)


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps D PSEUDO-CODE FOR ‘CURVY’ PONG SCAN PATTERN<br />

if( even(y_refl) ) begin<br />

grid[i,0] = ygrid[y_<str<strong>on</strong>g>of</str<strong>on</strong>g>f]<br />

end<br />

else begin<br />

grid[i,0] = ygrid[y_ngridseg-y_<str<strong>on</strong>g>of</str<strong>on</strong>g>f]<br />

end<br />

end<br />

; even reflecti<strong>on</strong>s<br />

; odd reflecti<strong>on</strong>s<br />

; On exit, "grid" c<strong>on</strong>tains the ordered list <str<strong>on</strong>g>of</str<strong>on</strong>g> tanplane <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets for the<br />

; individual line segments for the P<strong>on</strong>g pattern. At this stage, the rotati<strong>on</strong><br />

; <str<strong>on</strong>g>of</str<strong>on</strong>g> the box through "angle" may be applied to grid.<br />

for ( j=0; j< numvertices; j++ ) begin<br />

tx = grid[j][0] * cos(angle) - grid[j][1] * sin(angle);<br />

ty = grid[j][0] * sin(angle) + grid[j][1] * cos(angle);<br />

grid[j][0] = tx;<br />

grid[j][1] = ty;<br />

end<br />

Since this first rotati<strong>on</strong> occurs in the native coordinate system <str<strong>on</strong>g>of</str<strong>on</strong>g> the box (Nasmyth/AzEl/RADec),<br />

a further rotati<strong>on</strong> must be applied such that the scan end points are defined in a coordinate<br />

system useful for the TCS (probably AzEL tangent-plane <str<strong>on</strong>g>of</str<strong>on</strong>g>fsets). As the scan is executed,<br />

the telescope slew speed in the tangent-plane should be ‘velocity’.<br />

D<br />

Pseudo-code for ‘Curvy’ P<strong>on</strong>g Scan <str<strong>on</strong>g>Pattern</str<strong>on</strong>g><br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> ‘Curvy’ P<strong>on</strong>g pattern is based <strong>on</strong> the ‘boxscan’ used by SHARC-II and allows for an<br />

approximati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> a P<strong>on</strong>g pattern while avoiding sharp turnarounds at the vertices. <str<strong>on</strong>g>The</str<strong>on</strong>g> path<br />

<str<strong>on</strong>g>of</str<strong>on</strong>g> this ‘Curvy’ P<strong>on</strong>g pattern is described by two triangle wave approximati<strong>on</strong>s. With x(t) and<br />

y(t) driven by Fourier expansi<strong>on</strong>s (approximately 5 terms) <str<strong>on</strong>g>of</str<strong>on</strong>g> a triangle wave, the pattern<br />

approaches straight lines across the central regi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the scan while rounding <str<strong>on</strong>g>of</str<strong>on</strong>g>f the corners.<br />

This approach may be necessary if it is found that abrupt accelerati<strong>on</strong> changes result in<br />

pointing problems. Note, with <strong>on</strong>ly <strong>on</strong>e term, this pattern describes a Lissajous pattern,<br />

while with an infinite number <str<strong>on</strong>g>of</str<strong>on</strong>g> terms it recreates the ‘Straight’ P<strong>on</strong>g pattern as above.<br />

; Determine number <str<strong>on</strong>g>of</str<strong>on</strong>g> vertices (reflecti<strong>on</strong> points) al<strong>on</strong>g each side <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />

; box which satisfies the comm<strong>on</strong>-factors criteri<strong>on</strong> and the requested<br />

; size / spacing<br />

; Find out how far apart the vertices are al<strong>on</strong>g the axes<br />

vert_spacing = ( 2.0 * spacing ) / sqrt ( 2.0 )<br />

; Determine how many vertices (minimum) there must be al<strong>on</strong>g<br />

; each axis to cover the required area<br />

x_numvert = ceil ( width / vert_spacing )<br />

y_numvert = ceil ( height / vert_spacing )<br />

Page 34 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps D PSEUDO-CODE FOR ‘CURVY’ PONG SCAN PATTERN<br />

; Determine which is lower and check to make sure that <strong>on</strong>e is<br />

; even while the other is odd<br />

if ( x_numvert >= y_numvert ) begin<br />

most = x_numvert<br />

least = y_numvert<br />

else begin<br />

most = y_numvert<br />

least = x_numvert<br />

end<br />

; If both are odd or both are even, increment the lesser <str<strong>on</strong>g>of</str<strong>on</strong>g><br />

; the two, and update which is least<br />

if ( ( x_numvert % 2 ) == (y_numvert % 2 ) ) begin<br />

least += 1;<br />

if ( x_numvert >= y_numvert ) begin<br />

most = x_numvert<br />

least = y_numvert<br />

else begin<br />

most = y_numvert<br />

least = x_numvert<br />

end<br />

end<br />

; Check for comm<strong>on</strong> factors between the two, and adjust as<br />

; necessary until x_numvert and y_numvert do not share any<br />

; comm<strong>on</strong> factors<br />

for ( i = 3; i


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps D PSEUDO-CODE FOR ‘CURVY’ PONG SCAN PATTERN<br />

period = x_numvert * y_numvert * vert_spacing * 2 / vavg<br />

; Determine the total number <str<strong>on</strong>g>of</str<strong>on</strong>g> positi<strong>on</strong>s required to be<br />

; recorded<br />

p<strong>on</strong>gcount = period / sample_interval<br />

; Calculate the amplitudes <str<strong>on</strong>g>of</str<strong>on</strong>g> x(t) and y(t)<br />

amp_x = x_numvert * vert_spacing / 2<br />

amp_y = y_numvert * vert_spacing / 2<br />

; Get the positi<strong>on</strong>s at each time<br />

t_count = 0<br />

for ( i = 0 to p<strong>on</strong>gcount ) begin<br />

end<br />

; Calculate the grid positi<strong>on</strong>s from 5 terms<br />

tx = ( ( 8.0 * amp_x ) / ( pi * pi ) ) *<br />

( sin ( 2.0 * pi * t_count / peri_x ) -<br />

( 1.0/9.0 * sin ( 6.0 * pi * t_count / peri_x ) ) +<br />

( 1.0/25.0 * sin ( 10.0 * pi * t_count / peri_x ) ) -<br />

( 1.0/49.0 * sin ( 14.0 * pi * t_count / peri_x ) ) +<br />

( 1.0/81.0 * sin ( 18.0 * pi * t_count / peri_x ) ) )<br />

ty = ( ( 8.0 * amp_y ) / ( pi * pi ) ) *<br />

( sin ( 2.0 * pi * t_count / peri_y ) -<br />

( 1.0/9.0 * sin ( 6.0 * pi * t_count / peri_y ) ) + \<br />

( 1.0/25.0 * sin ( 10.0 * pi * t_count / peri_y ) ) -<br />

( 1.0/49.0 * sin ( 14.0 * pi * t_count / peri_y ) ) +<br />

( 1.0/81.0 * sin ( 18.0 * pi * t_count / peri_y ) ) )<br />

; Apply the rotati<strong>on</strong> angle<br />

grid[i][0] = tx * cos(angle) - ty * sin(angle)<br />

grid[i][1] = tx * sin(angle) + ty * cos(angle)<br />

t_count++<br />

Page 36 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps E VELOCITIES AND ACCELERATIONS OF ‘CURVY’ PONG<br />

E<br />

Velocities and Accelerati<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> ‘Curvy’ P<strong>on</strong>g<br />

Here we give an explicit example <str<strong>on</strong>g>of</str<strong>on</strong>g> a ‘Curvy’ P<strong>on</strong>g pattern and examine the velocity and<br />

accelerati<strong>on</strong> comp<strong>on</strong>ents.<br />

Figure 72: ‘Curvy’ P<strong>on</strong>g used to examine velocities and accelerati<strong>on</strong>s.<br />

We choose the following parameters, based <strong>on</strong> values determined from a simulated P<strong>on</strong>g<br />

observati<strong>on</strong>. α x and α y are the amplitudes <str<strong>on</strong>g>of</str<strong>on</strong>g> the x and y moti<strong>on</strong>, respectively, and β x and<br />

β y are the periods <str<strong>on</strong>g>of</str<strong>on</strong>g> the x and y moti<strong>on</strong>, respectively. t ranges from 0 to 17 (just over <strong>on</strong>e<br />

full period <str<strong>on</strong>g>of</str<strong>on</strong>g> the complete pattern). <str<strong>on</strong>g>The</str<strong>on</strong>g> original P<strong>on</strong>g simulati<strong>on</strong> had width ≃2400 arcsec,<br />

height ≃1600 arcsec, and sweep spacing <str<strong>on</strong>g>of</str<strong>on</strong>g> 600 arcsec.<br />

α x = 1272.792206; α y = 848.528137; β x = 8.475; β y = 5.65; (3)<br />

a = 8α x<br />

π 2 ;<br />

b = 2π<br />

β x<br />

;<br />

c = 8α y<br />

π 2 ;<br />

d = 2π<br />

β y<br />

; (4)<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g>n we have the triangle wave expansi<strong>on</strong>s (plotted in Fig. 73) :<br />

[<br />

x(t) = a sin(bt) − 1 9 sin(3bt) + 1<br />

25 sin(5bt) − 1<br />

49 sin(7bt) + 1 ]<br />

81 sin(9bt) ; (5)<br />

[<br />

y(t) = c sin(dt) − 1 9 sin(3dt) + 1<br />

25 sin(5dt) − 1 49 sin(7dt) + 1 ]<br />

81 sin(9dt) . (6)<br />

Page 37 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps E VELOCITIES AND ACCELERATIONS OF ‘CURVY’ PONG<br />

Figure 73: x(t) (bottom) and y(t) (top) <str<strong>on</strong>g>of</str<strong>on</strong>g>fset by 7000 for clarity.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> velocity and accelerati<strong>on</strong> comp<strong>on</strong>ents are :<br />

ẋ(t) = ab<br />

[cos(bt) − 1 3 cos(3bt) + 1 5 cos(5bt) − 1 7 cos(7bt) − 1 ]<br />

9 cos(9bt) ; (7)<br />

ẏ(t) = cd<br />

[cos(dt) − 1 3 cos(3dt) + 1 5 cos(5dt) − 1 7 cos(7dt) − 1 ]<br />

9 cos(9dt) ; (8)<br />

ẍ(t) = ab 2 [− sin(bt) + sin(3bt) − sin(5bt) + sin(7bt) − sin(9bt)] ; (9)<br />

ÿ(t) = cd 2 [− sin(dt) + sin(3dt) − sin(5dt) + sin(7dt) − sin(9dt)] . (10)<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g>se are plotted in Figs. 74 and 75.<br />

We can also define the magnitude <str<strong>on</strong>g>of</str<strong>on</strong>g> the velocity as<br />

v(t) = √ [ẋ 2 (t) + ẏ 2 (t)] (11)<br />

and similarly for the magnitude <str<strong>on</strong>g>of</str<strong>on</strong>g> the accelerati<strong>on</strong> |a(t)| as plotted in Figs. 76 and 77.<br />

<str<strong>on</strong>g>The</str<strong>on</strong>g> accelerati<strong>on</strong> comp<strong>on</strong>ents are mildy oscillating, with spikes <str<strong>on</strong>g>of</str<strong>on</strong>g> up to about 4000 arcsec<br />

s −2 . <str<strong>on</strong>g>The</str<strong>on</strong>g> magnitudes <str<strong>on</strong>g>of</str<strong>on</strong>g> the total velocity and total accelerati<strong>on</strong> vary widely depending <strong>on</strong> the<br />

moti<strong>on</strong> in the x and y directi<strong>on</strong>s at any given time. Clearly, the final result will depend <strong>on</strong> the<br />

individual x and y velocities, as well as the number <str<strong>on</strong>g>of</str<strong>on</strong>g> terms used in the Fourier expansi<strong>on</strong>.<br />

Page 38 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps E VELOCITIES AND ACCELERATIONS OF ‘CURVY’ PONG<br />

Figure 74: ẋ(t) (bottom) and ẏ(t) (top) <str<strong>on</strong>g>of</str<strong>on</strong>g>fset by 7000 for clarity.<br />

Figure 75: ẍ(t) (bottom) and ÿ(t) (top) <str<strong>on</strong>g>of</str<strong>on</strong>g>fset by 7000 for clarity.<br />

Page 39 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps E VELOCITIES AND ACCELERATIONS OF ‘CURVY’ PONG<br />

Figure 76: |v(t)|.<br />

Figure 77: |a(t)|.<br />

Page 40 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41


SCUBA-2: <str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Impact</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Scanning</str<strong>on</strong>g> <str<strong>on</strong>g>Pattern</str<strong>on</strong>g> <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> <strong>on</strong> <strong>Uniform</strong> <strong>Sky</strong> Coverage <str<strong>on</strong>g>of</str<strong>on</strong>g> Large<br />

Maps<br />

REFERENCES<br />

References<br />

[1] C. Borys. Box Scan for SHARC-II. CSO telescope notes, URL:<br />

http://www.submm.caltech.edu/∼sharc/operating/boxscan.htm, 2003.<br />

[2] B. G. Gom. SCUBA-2 FTS Science Case. SCUBA-2 Project SC2/FTS/SCE/001, 2003.<br />

[3] W. S. Holland, E. I. Robs<strong>on</strong>, W. K. Gear, C. R. Cunningham, J. F. Lightfoot, T. Jenness,<br />

R. J. Ivis<strong>on</strong>, J. A. Stevens, P. A. R. Ade, M. J. Griffin, W. D. Duncan, J. A. Murphy, and<br />

D. A. Naylor. SCUBA: a comm<strong>on</strong>-user submillimetre camera operating <strong>on</strong> the James<br />

Clerk Maxwell Telescope. M<strong>on</strong>thly Notices <str<strong>on</strong>g>of</str<strong>on</strong>g> the Royal Astr<strong>on</strong>omical Society, 303:659–<br />

672, March 1999.<br />

[4] Wayne S. Holland. Functi<strong>on</strong>al and performance requirements for SCUBA-2. SCUBA-2<br />

Project SC2/SRE/SC200/002.<br />

[5] Wayne S. Holland. Science requirements. SCUBA-2 Project SC2/SRE/SC200/001.<br />

[6] B. D. Kelly. SCUBA-2 modes and data processing. SCUBA-2 Project SC2/ANA/S100/028,<br />

2001.<br />

[7] B. D. Kelly and W. D. Duncan. Applicati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> DREAM Observing to SCUBA-2. SCUBA-2<br />

Project SC2/ANA/S100/006, 2001.<br />

[8] P.Bastien, T.Jenness, and J.Molnar. A Polarimeter for SCUBA-2. Astr<strong>on</strong>omical Polarimetry:<br />

Current Status and Future Directi<strong>on</strong>s ASP C<strong>on</strong>gerence Series, 343, 69, 2005.<br />

[9] R. Prestage, F. J. Oliveira, J. A. Bailey, P. Friberg, and S. Kenderdine. Route-finding<br />

algorithm for the optimal c<strong>on</strong>trol <str<strong>on</strong>g>of</str<strong>on</strong>g> telescope slews. In H. Lewis, editor, Proc. SPIE Vol.<br />

3112, p. 76-87, Telescope C<strong>on</strong>trol Systems II, Hilt<strong>on</strong> Lewis; Ed., pages 76–87, September<br />

1997.<br />

[10] Richard Prestage. Discussi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the JCMT Cell and Rastering Facilities. Joint Astr<strong>on</strong>omy<br />

Centre James Clerk Maxwell Telescope TCS/DN/001, 1996.<br />

[11] Alex van Engelen and Douglas Scott. SCAN Mode <str<strong>on</strong>g>Strategies</str<strong>on</strong>g> for SCUBA-2. SCUBA-2<br />

Project SC2/ANA/S210/005, 2005.<br />

Page 41 <str<strong>on</strong>g>of</str<strong>on</strong>g> 41

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!