Uploaded image for project: 'Data Management'
  1. Data Management
  2. DM-9046

Filter characterisation for SNIa cosmology

    Details

    • Team:
      DM Science

      Description

      At the DESC meeting this summer (2016), Nicolas Regnault showed in the Photometric Corrections session that in order for LSST SNIa cosmology not to be systematics limited, the nominal "edge" of the filters, and the evolution thereof, needs to be known to and monitored at the 0.1nm level.

      This is a somewhat surprising result (to some at least), and imposes some very stringent requirements on the photometric calibration system.

      There is currently nothing in the SRD stating that we need knowledge of the filters at this level so this putative requirement is not being flowed down to the design in LDM-151.

      This ticket is for someone to decide if a change to the requirements is necessary as this will likely involve making hardware change requests, given the current plans for filter and photometric calibration hardware.

      A link to the slides can be found here:
      https://confluence.slac.stanford.edu/display/LSSTDESC/Photometric+Corrections

      https://confluence.slac.stanford.edu/download/attachments/211803218/sncalib-desc-oxford.pdf?version=1&modificationDate=1468915736000&api=v2

        Attachments

          Issue Links

            Activity

            mfisherlevine Merlin Fisher-Levine created issue -
            Hide
            swinbank John Swinbank added a comment -

            Merlin Fisher-Levine: Does this block DM-8066 (updates to CPP section of LDM-151)?

            Show
            swinbank John Swinbank added a comment - Merlin Fisher-Levine : Does this block DM-8066 (updates to CPP section of LDM-151)?
            swinbank John Swinbank made changes -
            Field Original Value New Value
            Link This issue relates to DM-8066 [ DM-8066 ]
            Hide
            mfisherlevine Merlin Fisher-Levine added a comment -

            I don't see why it should; LDM-151 is only concerned with the algorithmic implications (right?), whereas the requirements are outlined in the SRD.

            I don't think there will be significant algorithmic changes whatever gets decided here. Should it be decided that we have to do better, hardware configurations (and possibly purchases too) would need to be changed, but I don't think we would do anything (significantly) different in terms of the processing - we will always be doing our best with what we've got - it's just what that ends up being might change.

            Does that seem reasonable?

            Show
            mfisherlevine Merlin Fisher-Levine added a comment - I don't see why it should; LDM-151 is only concerned with the algorithmic implications (right?), whereas the requirements are outlined in the SRD. I don't think there will be significant algorithmic changes whatever gets decided here. Should it be decided that we have to do better, hardware configurations (and possibly purchases too) would need to be changed, but I don't think we would do anything (significantly) different in terms of the processing - we will always be doing our best with what we've got - it's just what that ends up being might change. Does that seem reasonable?
            Hide
            wmwood-vasey Michael Wood-Vasey added a comment -

            Knowledge of the filter edges to 0.1 nm is substantially more stringent than I would have estimated. E.g., if there is a night-to-night random shift of just one edge[*] of a filter by 0.1 nm back and forth with no long-term temporal correlation, that really doesn't matter.

            However, I also can imagine a number of things one could estimate that would naively suggest such a control. Regnault is certainly not naive about these issues, but the specific calculation relies on details of how calibration that is being done. And I agree that we likely do need to know the effective central wavelength of the filter function

            I would suggest that the first task is to more fully capture exactly what Regnault is calculating and suggesting:

            1. What is it that we need to know?

            • [*] What is an "edge": Position at which transmission drops to 50%, 10%?
            • Integrated fractional flux uncertainty of nominal SED integrated over the 10-50 nm of a filter edge transition to better than 10%, 1%, 0.1%?
            • Effective central wavelength of filter multiplied by some given SED? The figures in the referenced PDF show significant deviations over large ranges of redshift. That implies an error in the effective central wavelength of the filter.
              2. Over what timescale do we need to know it:
            • in aggregate
            • per epoch
            • per 3-month time-of-interest for a given supernova
              3. Does the simulations use the fact that the systematic with redshift has significant correlation?
              4. What is the power of any filter edge specification resolution and accuracy vs. expected cosmological signal?
            Show
            wmwood-vasey Michael Wood-Vasey added a comment - Knowledge of the filter edges to 0.1 nm is substantially more stringent than I would have estimated. E.g., if there is a night-to-night random shift of just one edge [*] of a filter by 0.1 nm back and forth with no long-term temporal correlation, that really doesn't matter. However, I also can imagine a number of things one could estimate that would naively suggest such a control. Regnault is certainly not naive about these issues, but the specific calculation relies on details of how calibration that is being done. And I agree that we likely do need to know the effective central wavelength of the filter function I would suggest that the first task is to more fully capture exactly what Regnault is calculating and suggesting: 1. What is it that we need to know? [*] What is an "edge": Position at which transmission drops to 50%, 10%? Integrated fractional flux uncertainty of nominal SED integrated over the 10-50 nm of a filter edge transition to better than 10%, 1%, 0.1%? Effective central wavelength of filter multiplied by some given SED? The figures in the referenced PDF show significant deviations over large ranges of redshift. That implies an error in the effective central wavelength of the filter. 2. Over what timescale do we need to know it: in aggregate per epoch per 3-month time-of-interest for a given supernova 3. Does the simulations use the fact that the systematic with redshift has significant correlation? 4. What is the power of any filter edge specification resolution and accuracy vs. expected cosmological signal?
            mjuric Mario Juric made changes -
            Labels dm-sst
            mjuric Mario Juric made changes -
            Assignee Mario Juric [ mjuric ] Robert Lupton [ rhl ]
            Hide
            mjuric Mario Juric added a comment -

            Robert Lupton, you mentioned back in April that the DESC calibration group is investigating – any updates?

            Show
            mjuric Mario Juric added a comment - Robert Lupton , you mentioned back in April that the DESC calibration group is investigating – any updates?
            swinbank John Swinbank made changes -
            Team Project Science [ 11800 ] DM Science [ 12218 ]
            Hide
            lguy Leanne Guy added a comment -

            Robert Lupton What is the status?

            Show
            lguy Leanne Guy added a comment - Robert Lupton What is the status?
            lguy Leanne Guy made changes -
            Risk Score 0
            lguy Leanne Guy made changes -
            Remote Link This issue links to "Page (Confluence)" [ 18031 ]
            lguy Leanne Guy made changes -
            Rank Ranked higher
            ctslater Colin Slater made changes -
            Remote Link This issue links to "Page (Confluence)" [ 20969 ]
            wmwood-vasey Michael Wood-Vasey made changes -
            Attachment calibration_impact_forecast.pdf [ 39129 ]
            Hide
            wmwood-vasey Michael Wood-Vasey added a comment - - edited

            I've attached the relevant DESC Note by Hazenberg, Betoule, Bongard, Le Guillou, Regnault, Gris for DESC
            calibration_impact_forecast.pdf

            Show
            wmwood-vasey Michael Wood-Vasey added a comment - - edited I've attached the relevant DESC Note by Hazenberg, Betoule, Bongard, Le Guillou, Regnault, Gris for DESC calibration_impact_forecast.pdf
            Hide
            ctslater Colin Slater added a comment -

            The SST discussed this and would like to understand what these constraints would translate to if they were framed in terms of DM's requirements, e.g. on single-epoch measurement of source color. Without that, it's hard to know what DM can do about this (or whether in practice this is the dominant systematic uncertainty).

            Show
            ctslater Colin Slater added a comment - The SST discussed this and would like to understand what these constraints would translate to if they were framed in terms of DM's requirements, e.g. on single-epoch measurement of source color. Without that, it's hard to know what DM can do about this (or whether in practice this is the dominant systematic uncertainty).
            Hide
            wmwood-vasey Michael Wood-Vasey added a comment -

            I read the document (finally). It makes two key statements

            • Photometric calibration (in relative color) must be know to sub mmag level.
            • The average effective filter mean wavelength needs to be known to better than 0.1 nm.

            I agree with the conclusion that DESC SN Cosmology requires sub mmag precision for SN Ia lightcurves to reach the full potential of LSST SNeIa. Figuring out how to do that is within DESC's remit and not the Project or Operations. It's also possible, although challenging.

            The 0.1 nm specification from the document is a requirement on the knowledge of the mean system bandpass for that filter (i.e., the "effective wavelength of the filter"). Conceptually this is about the issue of when we say that the precision in "g" band is 3~mmag, what did we mean by "g"?

            It is not a requirement on determining the edges of the filter response to 0.1 nm on a per-visit basis, which would lie somewhere between "not possible" and "crazy".

            While the paper presents its analysis as being optimistic, I think they're missing an opportunity for further self-calibration. In particular, the analysis presented in the paper ignores the fact that SNeIa at a range of redshifts will be being observed simultaneously in each image. So an effective wavelength variation would affect a set of supernovae, and you would know which ones those were.

            Show
            wmwood-vasey Michael Wood-Vasey added a comment - I read the document (finally). It makes two key statements Photometric calibration (in relative color) must be know to sub mmag level. The average effective filter mean wavelength needs to be known to better than 0.1 nm. I agree with the conclusion that DESC SN Cosmology requires sub mmag precision for SN Ia lightcurves to reach the full potential of LSST SNeIa. Figuring out how to do that is within DESC's remit and not the Project or Operations. It's also possible, although challenging. The 0.1 nm specification from the document is a requirement on the knowledge of the mean system bandpass for that filter (i.e., the "effective wavelength of the filter"). Conceptually this is about the issue of when we say that the precision in "g" band is 3~mmag, what did we mean by "g"? It is not a requirement on determining the edges of the filter response to 0.1 nm on a per-visit basis, which would lie somewhere between "not possible" and "crazy". While the paper presents its analysis as being optimistic, I think they're missing an opportunity for further self-calibration. In particular, the analysis presented in the paper ignores the fact that SNeIa at a range of redshifts will be being observed simultaneously in each image. So an effective wavelength variation would affect a set of supernovae, and you would know which ones those were.
            Hide
            wmwood-vasey Michael Wood-Vasey added a comment -

            I think the root of the concern from the SN cosmology community is that it seems possible that LSST could produce standardized fluxes of stars to better precision that LSST could tell you the per-wavelength transmission function.

            This can be the case merely implicitly because one never quoted a transmission function, e.g., in the traditional ("legacy") model of calculating color terms between standard systems and your observation, one doesn't explicit calculate the transmission function, even though the terms one applies could be inverted to do so. But it also is likely to be the case from an available information perspective, as the range of color-color-color-color space occupied by main-sequence stars is a subset of the overall color space, so perhaps there are degeneracies possible in the effective transmission function that are not resolved solely by calibrating stars.

            One potential action would be to clarify if we have an explicit per-epoch requirement on the knowledge of the transmission function. I think the answer is no, and I'm leery of trying to writing one down (the easy thing to write down isn't constraining enough, and the naive desirement one would like to write down is too constraining).

            Show
            wmwood-vasey Michael Wood-Vasey added a comment - I think the root of the concern from the SN cosmology community is that it seems possible that LSST could produce standardized fluxes of stars to better precision that LSST could tell you the per-wavelength transmission function. This can be the case merely implicitly because one never quoted a transmission function, e.g., in the traditional ("legacy") model of calculating color terms between standard systems and your observation, one doesn't explicit calculate the transmission function, even though the terms one applies could be inverted to do so. But it also is likely to be the case from an available information perspective, as the range of color-color-color-color space occupied by main-sequence stars is a subset of the overall color space, so perhaps there are degeneracies possible in the effective transmission function that are not resolved solely by calibrating stars. One potential action would be to clarify if we have an explicit per-epoch requirement on the knowledge of the transmission function. I think the answer is no, and I'm leery of trying to writing one down (the easy thing to write down isn't constraining enough, and the naive desirement one would like to write down is too constraining).

              People

              • Assignee:
                rhl Robert Lupton
                Reporter:
                mfisherlevine Merlin Fisher-Levine
                Watchers:
                Colin Slater, Johann Cohen-Tanugi, John Swinbank, Leanne Guy, Merlin Fisher-Levine, Michael Wood-Vasey
              • Votes:
                0 Vote for this issue
                Watchers:
                6 Start watching this issue

                Dates

                • Created:
                  Updated:

                  Summary Panel