Andy Fructer (spelling?) is on the stealth phone (the conference phone). Saul is asking him about the future of HST. The proposal coming this September will be for a period beginning between July and December 2001. The period will probably start after the servicing mission; the servicing mission will be sometime between July 2001 and December 2001. (Andy says its hard to get an official launch window.) At this point, because of shuttle problems, they don't even know which astronaut crew they'll use. So, the next period may begin as late as January 2002, but the proposals for them must go in in September. Andy thinks, however, that they're going to have to do something about this; there will probably be more information for users that go out when the call for proposals go out. Greg asks if the application deadline will slip; Andy doesn't think that will happen. Andy does not think that there will be an extra proposal sequence for the 6-month intermediate period. (It's very expensive to go through the TAC process.)
Greg asks how much buffer time there will need to be for the new instruments, after they are installed. For ACS, Andy thinks it will be 1-2 months. However, once science starts again after the servicing mission, ACS will be available for scientific programs. NICMOS II will come on later; they won't stdart futzing with that until after ACS is working.
Saul says for the proposal, we should ask what science could we do if hypothetically someday there was an ACS and a NICMOS II camera on the HST. Andy thinks that if anything, the proposals may be going towards favoring even larger programs, although there is some politics going on. Saul asks Andy what he thinks the right "sweet spot" would be to aim for in terms of number of orbits. Andy says that we should decide what makes good science and go for it. Over 300 is probably too many, but even 150 orbits may not be an unreasonable proposal.
ASC is supposed to be better than WF in I band by "a factor of a few"; probably over 3. In the blue you lose a little bit perhaps because of the smaller pixels; you may still have readnoise problems. In the red, the efficiency should make up for the smaller pixels that you are still sky dominated in short exposures. The pixel scale will be about the same as the PC.
Andy thinks that the CTE is probably going to be a problem. It will be better than WFPC because it's a new chip, but it will die fast, perhaps faster than the WFPC. What's more, the chip is a 2048, so you read out across many more pixels. Greg asks if there is something we could do in terms of positioning the objects in order to get the best possible readout. Hopefully, they will define an aperture at low X and low Y; Andy says he will talk to the ACS people so that they do this.
Saul asks if there is any other thing that we ought to take into account. Peter asks about CR hits; Andy isn't sure what the exact number is, but he doesn't think it will be that different. The chips are probably very similar to STIS, where the cosmic ray rate is higher.
Saul asks if there will be any changes in NICMOS; "will it be the same thing, just a little bit worse?" "Probably." It will be runing at a slightly higher temperature. The quantum efficiency will go up, but there will be higher background. This will hurt the most in the K-band; the H-band is going to be the primary filter, and it shouldn't hurt in the J-band at all. It's conceivable that the higher QE will win in the J-band.
Greg asks what is the photometric accuracy that people are comfortable with given the CTE. E.g., will people believe that you can do any better than 5%. Andy says that the real problem is going from the standards to the supernovae. If you put them in the same spot, it will be very reproducable, although the standards are a lot brighter. In the red, at least, there should be enough background that CTE won't be a big problem. ACS will have 5x the througput of the PC, due to a more sensitive red chip and a reduced number of optics.
Carl has an idea which he's selling before he tells it to us. He says it's practical, can be done with resources in hand, and can help build the educational structure. He wants to be able to churn out 5-6 SNe a night at z=0.5 or so with WFI. Suppose we get half the time on the 2.2m and the WFI, and we get an IFS somewhere. Carl believes that he could bludgeon $200,000 a year or so. The science goal of this is to get lots and lots (hundreds) of z=0.3-0.6 SNe. Carl thinks this is a year or a copule of years away. The science driver would be detailed spectroscopy of some, intermittent J-band photometry. Saul points out that the questions we're trying to answer (dust, evolution) may be better addressed other ways, so we should consider other things. Carl points out that this has a good education angle as well, because it's easy to get all the kids involved. Greg points out that he and others have been fighting to get this telescope for the SN factory, and what Carl proposes here is very different and incompatable. (Presumably more on this later.)
A pool of z>1 SNIa. Say, 10 SNe observed with HST. This is only worthwhile if done with NICMOS; Peter thinks it's not worthwhile to do z>1 SNe unless you can do it with NICMOS. (U-band lightcurves are uncertain to a half-magnitude.) Peter says that ground based resources would then not be used on z>1 SNe until January 2002. Saul says, that's for this pool.... Carl thinks that this is something that could be searched with Suprime cam. Chris thinks that with Gemini and VLT, using AO, J-band observations of some of these tend to be practical. Conica says J-band at 26 with S/N~3 in 1 hour. If you believe this (it may just be the PR), then we may be able to do it. The problem, however, is convincing people that you can do photometry on them. Peter says that if we want to do ground based AO followup of high-z supernovae, sometime in the next year we have to prove that we can do photometry on these AO systems. There is thought also of doing spectra, as this doesn't require as much proving.
For this coming Spring we have 70 orbits. We said in the proposal that we would get U-band lightcurves, and get an AO point in the IR at Keck for a blue photometry point. Then, of course, we didn't get granted any Keck time.
Andy Fruchter interjects the question as to whether we want to observe once a year or twice a year. Saul says that we'll get to this, after we think about the science we want to do.
So much for keeping a list... the discussion got too messy.
We may also instead consider doing SNe at z~0.8-0.9, since the spectra are easier there. We can also get, from the ground, UBV without NICMOS (though you still need infrared). The question is what, cosmology wise, can you do at z=1.2 that's different from what you can do at z=0.9. Peter says that the U-B color is hard to use, because the absolute peak U is so uncertain. Most of the cosmological leverage, Alex Kim tells us looking at our poster, looks like it's already there at z=0.9.
Andy Fruchter warns us that we should worry about the possibility that the crycooler on NICMOS isn't going to work. Greg says that right now we're trying to figure out what we should do until we have NICMOS, i.e. doing things entirely ground based.
So, it sounds like for the 70 orbits we have to use in the next year, the target should be z=0.85 to 0.9. We seem to agree that the optical lightcurve has to be HST for I-band, since you can't go down below peak far enough on the ground. This would be like 6 SNe. This is probably not unreasonable to do in 5 nights of Keck (which isn't outside the range of reasonability for what we can get over the course of a full year). These SNe would be in the range of VLT as well. To find them, we'd need something like 2 hours of exposure. We'd need a couple nights of searching, in between CTIO and CFHT, to get 8 metafields. We should be careful about designating new search fields; if we have fields that match what we've done before, we have "final ref" backup with HST. Say, 2 nights ref and 2 nights search. Greg worries about this, since we tried to do something very close to this in Spring of 1998, and didn't really get it. Perhaps we need more search time, or perhaps we want to put in some confirmation/followup time.
Another option for the "now" line (i.e. through June 2001) would be to get some z=0.5 supernoave. Saul says that if we do the search for z=0.9 supernovae, we will just get some lower-z supernoave, and what's more we'll probably end up being stuck doing them due to a dearth of high-z supernovae. (Say 3 done on the piggyback.) Peter says that the idea there would be to improve upon supernovae not at peak, but later. Greg asks what is the direct constraint you can put baesd on the light time (e.g. 3-week) spectra. Peter says that you can look at the velocity of the iron lines to place constraints on the amount of Nickel. Peter says that the spectra should be of comparable quality to the one of Beethoven we got at VLT- and, the magnitude 3 weeks after should be (say) 0.5-0.8 fainter than the 8-days early spectrum. Chris points out that this was a 2-hour spectrum under perfect conditions. Greg mentions that we're talking an 8-hour VLT or Keck spectrum for something that much dimmer on one supernova. Perhaps your early spectrum should be longer as well to get a detailed spectrum there.
Saul says he doesn't think that we want to get into the business of finding scads (hundreds) of z=0.2 supernoave.
Saul also brings up the idea of doing a SN II run. Greg points out that if you do a ref, a search 3 weeks later, and a "confirm" run a week after that, you can use the 1-week gap to to search for SN II. A search that gets Ias z=0.85-0.9 should also get SN II at 0.2-0.3. The point of doing SN II would be to show that you can do this kind of cosmological measurement at all with them. (Plus, if you succeed, you get a strong check of systematics, since this is a different type of object.) Peter says that the followup photometry of these SN II would be analogous to the z=0.85 SN... but that, even though the magnitudes are comparable, it's easier to do the II's from the ground. The reason is that there's a bright plateau (0.5, maybe as much as 1.0, magnitudes down from a Ia; the peak is maybe 1.7 down from a Ia at the same z) on the IIp supernoave (the ones you find half the time). This makes getting later epoch points on a II easier than on Ias. You need spectra at peak and then again 30 days later. Peter says you want to measure these things to a few hundred kilometers per second. He says that the typical features are 100-150 angstroms wide. You could in principle of this figure out how long you have to go to get sufficient S/N. He says it's basically the same amount of time you need to get a decent 0.8 Ia. So, we're talking about 4 hours, twice.
For these II's, you might actually want to do a B-band or V-band search, since the II's are brighter then.
Peter notes that the SN II can actually do a good job of saying something about the grey dust. You have to get the color very well, you have to hammer it, you have to subtract the host galaxy very well. The grey dust would affect the shape of the (blackbody-like) specturm different from how host galaxy extinction. You can do this just with multicolor (say 5 colors) photometry (easier than fluxing a spectrum over a long range). The spectrum just tells you that it's early and it's hot.
Ariel wonders why you don't just start this II stuff with the SN factory, where you can find them at lower redshift. Peter says, though, that almost certainly if we don't do this first, the other group's expertise in II's will mean that we'll have almost no say in this game in the future. So perhaps we need to shove our foot in the door.
Peter says to do this, you need a 1-week basline one the search. He says he'd make it a separate proposal, and wouldn't tie it in to anything else we're doing.
Saul says it's not clear how we word the proposal. Perhaps for advertising purposes, we say we'll do 0.8-1.1 or 1.2, and not say how much we're really hoping to do below a redshift of 0.9. Next question, do we do this once during this coming year, and what time of year would it be? First question is, do we have to do it equatorial? If we hope to use VLT, then yes.THE PLAN:
In terms of next spring, the Diemos Spring field is something like 16h, which is late. Given that, it may make sense for us to try to go early, say beginning/mid February. We're trying to get 6 z=0.85-0.9 supernoave, with maybe a couple at z=0.5. We'd have to choose the fields carefully; some of the old ones show dust clumps on the higher resoultion maps. We'd probably have to do two equatorial fields in order to fill out the night (avoiding the logisitical problem of scheduling half-nights everywhere). Hopefully we will be able to do overkill on time to make sure we get enough in each field.
We need to figure out when the new moon is, but if it's something like Feb 20, then we'll need to do a Feb/March search, with HST followup in March/April/May. Spectra followup will be in March, then, with perhaps VLT time at the very beginning of April.
Peter notes, as we think about the future HST proposal, that Adam Reiss is trying to raise some groundswell for doing a *search* for Ia's at z>1 with HST. Peter thinks this sounds half-cocked; most people tend to think that you can search just as well on the ground. This might be the next generation HDF, and the whole SN search may actually be a piggyback on the HDFng- just get the HDFng observations timed properly. In the proposal, we may need to address why we thinks it makes more sense to search from the ground.
Greg says that along the same lines is the whole issue of Tyson posting supernovae for free on the web. As we apply at CTIO, they might wonder why we don't just arrange with Tyson for him to find supernova, and plan our followup accordingly. Apparently he's finding a couple a night, at something like 24th magnitude. (He's doing a large weak lensing survey program.)Coming fall
The coming fall, Reynald says that they already have 2x1.5 nights on the CFHT in September and October. We'll discuss later what to do with this time once we find out what we have with VLT. We won't be able to use HST with any of this, because it's already set on the schedule for the spring.
Tomorrow, big meeting starts here at 10:30.