2

Findings and Conclusions

The Roman Space Telescope will measurably advance knowledge of dark energy and exoplanet demographics. Locally, it will likely enhance understanding of the structure and substructure of the Milky Way and nearby galaxies, including a census of the predicted but elusive ultra-faint dwarf galaxies. At high redshift, it can provide information on the topology of reionization and the abundance of sources like active galactic nuclei and pair-instability supernovae. With a wavelength range of 0.48-2.3 μm, Roman’s Wide Field Imager has the largest etendue of any existing or planned optical/infrared space observatory. The coronagraph technology demonstration instrument will pioneer new capabilities that will be the basis for future instruments capable of directly detecting and characterizing Earth-like planets around nearby stars. If the technology demonstration is successful, observations with the coronagraph could make substantial advances in the study of planetary and disk systems. None of the data collected by Roman is proprietary. The data will become publicly available after being calibrated.

Roman was initiated as the Wide-Field InfraRed Survey Telescope (WFIRST), a 1.5 m telescope conceived by the Astro2010 decadal survey as a large mission at a cost of $1.6 billion. Astro2010 set the following three key science objectives for WFIRST: (1) constrain dark energy using measurements of baryon acoustic oscillations (BAOs), supernovae, and weak lensing; (2) provide exoplanet statistics using gravitational microlensing; and (3) implement a guest investigator program enabling a wide variety of astrophysics survey investigations. Astro2020 found that Roman “remains both powerful and necessary for achieving the scientific goals set by New Worlds, New Horizons (Astro2010). It will carry out cosmological measurements complementing those of Euclid and the Rubin Observatory, and Roman’s microlensing survey will probe planetary occurrence for orbital separations not constrained by Kepler or TESS.”

Roman will achieve its science objectives by conducting CCS and GA surveys. The CCS will be designed to meet the dark energy and exoplanet science requirements. The SRD flows down the science requirements to the following four surveys: the High Latitude Spectroscopic Survey (HLSS), the High Latitude Imaging Survey (HLIS), the Supernova (SN) survey, and the Exoplanet Microlensing (EML) survey. The HLSS and two components of HLIS (shallow and deep) comprise the High Latitude Wide Area Survey (HLWA). The DRM refers to the three CCSs as the HLWA, the High Latitude Time Domain (HLTD or SN), and the Galactic Bulge Time Domain (GBTD or EML) Surveys.

The term general astrophysics refers to a broad range of science objectives beyond those of the CCS. Astro2010 considered “the general investigator program to be an essential element of the mission” and said it would consist of “both key projects and archival studies to address a broad range of astrophysical research topics.” The 2015 Science and Technology Definition Team report lists tens of GA science programs that are uniquely enabled by Roman, and the white papers submitted to Astro2020 further underscore Roman’s broad science reach. Astro2020 notes that “although the most obvious advances will be in cosmology and exoplanets, Roman’s immense discovery potential beyond those areas almost

ensures that its highest impact results will come from other, and possibly unforeseen, directions.”⁷ Much of the information provided to this committee reinforces these comments, and the CAA concurs.

Roman’s GA science can be realized in the following three ways: (1) the data from the CCS can be used for GA science objectives; (2) the CCS can be augmented to enable broader science objectives by, for example, adding filters, modifying observing times or changing scan strategies; and (3) GA survey programs can be selected through an open community-wide, peer-reviewed selection process with principal investigator (PI) teams responding to a broad solicitation. SDT-13 recommended a guest observer program utilizing a minimum of 25 percent of the mission minimum life time. SDT-15 gave an example observing program in which 25 percent of the baseline 6-year mission was for guest observations.

Under NASA’s current plan, Roman’s prime mission duration is 5 years, and no more than 75 percent of its observing time is allocated for the CCS and Coronagraph Technology Demonstration. At least 25 percent of the time will be allocated to the competed GA programs, which will have up to 30 programs. A current estimate of the time allocation provided to the committee shows the CCS and GA surveys taking 58 and 22 percent of 5 years, respectively, and leaving 4.4 percent of margin after accounting for mission operations and calibrations (see Table 2.1). Of the total of 1,478 days currently allocated to astrophysics observations, excluding the 90 days allocated for the coronagraph technology demonstration, the CCS and GA surveys have 72 and 28 percent of the time, respectively. For brevity we will refer to this time split as “75/25,” although it is important to note that it really means “up to 75 percent time” for the CCS and coronagraph technology demonstration and “at least 25 percent time” for the competed GA programs.

The SRD and DRM describe fiducial designs for the CCS. They present reference and baseline surveys that could meet Roman’s science objectives. The requirements for the cosmology surveys are quantified in terms of achieving certain values for the standard figure of merit (FoM) used to quantify uncertainties on dark energy parameters. The science requirements of the EML survey are quantified in terms of the expected number of planets detected and the report gives a table of the expected yield of bound planets as a function of their mass. Table 2.1 gives a summary of the duration of the reference surveys, the required performance as listed in the SRD, and the current best estimate for the expected performance of Roman. The reference surveys include margins relative to the baseline version. Experts provided information to the CAA indicating that the 75/25 split can achieve the current CCS science requirements.

___________________

⁷ National Academies of Sciences, Engineering, and Medicine, 2021, Pathways to Discovery in Astronomy and Astrophysics for the 2020s, prepublication release, Washington, DC: The National Academies Press, p. 7-34.

TABLE 2.1 Reference Survey Parameters

^a “Current Best Estimate” gives the expected performance of the Reference survey.

^b Margin is the ratio of the current best estimate FoM to the requirement FoM.

^c Values given by SRD as representing Astro2010 requirements.

^d Systematics-limited to statistics-limited.

NOTE: BAO = baryon acoustic oscillations; FoM = figure of merit; GBTD = Galactic Bulge Time Domain; HLTD = High Latitude Time Domain; HLWA = High Latitude Wide Area Survey; RSD = red-shift space distortions; WL = weak lensing.

SOURCES: Data from SRD (“Roman Space Telescope Science Requirements Document,” RST-SYS-REQ-0020, Revision C); DRM (“Roman Observations and Design Reference Mission,” February 24 slide presentation); and “Roman Observations and Design Reference Mission,” presentation to the CAA. If the SRD and DRM are different, DRM values are used.

The reference surveys outlined in the DRM are a demonstration that Roman can achieve the science goals set by SDT-15 with the 75/25 time split. Many community members in the survey teams contributed substantially to forming the current surveys, and significant progress has been made in quantifying the detailed trade-offs in their design. But it has always been clear that the reference surveys are not necessarily Roman’s flight surveys. An additional process, planned to begin in 2022, is supposed to lead to the final design of the flight surveys. Considering the information it received, the committee notes that very little, if any, element of observing time competition was introduced during the design of the reference surveys. This was an appropriate choice for that level of project development.

NASA has outlined a process of community-based design and committee-driven selection that will take place between now and 18 months prior to launch during which Roman’s CCS observing plans will be finalized. At the time of the CAA deliberations, the details of the process were not finalized, but a rough outline follows: the project was planning to conduct workshops, solicit white papers, and fund teams that will be coordinated by three CCS subcommittees. The three committees will be charged to assemble community input and converge on design options for each of the surveys. A steering committee,

organizationally above the CCS subcommittees, will make final selections and resolve conflicts. A separate competitive process involving a standard TAC would be instituted for selection of the up to 30 GA proposals in three cycles.

Having heard from community members, some of whom lead the survey design teams, the committee concluded that more work needs to be invested in designing Roman’s flight surveys. While the reference surveys are an appropriate demonstration that Roman’s science objectives can be reached within the 5-year prime mission, committee members noted cases for which identical or similar science objectives could potentially be reached with significantly less time. The current reference surveys also do not provide quantitative analyses of the relation between observing time and scientific metrics, making it difficult to balance overall observing time allocations among competing science objectives. The CCS design teams acknowledged these shortcomings. Examples of areas that require deeper analyses include the effect of using fewer than four filters for the HLIS and the impact of a slower cadence in the EML survey on detecting Earth-sized (as opposed to Mars-sized) planets. The teams have also acknowledged that some assumptions about the state of knowledge and the availability of supporting data, such as the number of low-redshift supernovae for the HLSS, will need to be revisited closer to launch. A quantitative analysis of the impact of such choices on the CCS may well result in significant savings in observing time. As noted earlier, little competition has been introduced so far either among the CCS or between any of the CCS and the potential GA science that can be done with Roman. Recognizing that the refined quantitative studies necessary to define the CCS flight surveys are likely to identify both observing time savings and opportunities to address GA objectives, the question is how to optimize overall time allocation across diverse science objectives.

The committee endorses the following principles to guide the process of survey design and time allocations:

1. Roman is first and foremost a wide-field survey instrument. Any observation will be a survey, be it small or large. Each of its surveys will yield rich data sets that can be mined for diverse science goals. As such, it is advantageous to plan the surveys such that they are useful for a broad range of science objectives. With proper pre-planning, adjustments in survey parameters, some of them potentially minor, may yield scientific windfalls for multiple science areas. To date there has been little GA input into the design of the reference surveys. The planned community process can be designed to remedy the situation. Chapter 3 of this report gives an example of a process through which GA science could be more effectively infused into the CCS.

Conclusion: Roman’s overall science output could be increased by having the design of the CCS be informed by GA science objectives.

2. There are a number of ways to address GA science goals with Roman. While some GA science goals may be achieved through CCS observations, and some with enhancements to the CCS, there are GA objectives that will require their own surveys, including potential use of the coronagraph. To maximize Roman’s science return, GA time allocations of these three types cannot be considered independently. It would be valuable to evaluate and contrast all GA science, and the observing time various objectives require, together.

Conclusion: Roman’s overall science output could be increased by establishing a combined evaluation of all observing time requirements, including CCS, their GA extensions, and GA-only surveys.

3. NASA is considering a collaborative process to facilitate convergence on Roman’s observing plan. The CAA endorses the collaborative process and encourages the project to continue with its plans. However, the CAA notes that competition is a good way to generate proposals that are as

1. efficient as possible at achieving their science goals. Competition can also be effective whenever it is necessary to balance diverse science objectives. The committee endorses the inclusion of competition into the final survey selection process, and thus promotes a combination of collaborative and competitive processes.

   CCS design teams working collaboratively, potentially together with GA scientists, can be charged with proposing several alternative core surveys. GA teams that are not directly associated with CCS teams would also be encouraged to submit survey proposals. A competitive process could be implemented to choose among the proposed alternatives. Competition would create incentives for teams to justify observing choices and would ensure that observing time selections favor teams that have included more quantitative arguments and have better justified the scientific returns. A competitive process will likely motivate CCS teams to collaborate with GA scientists to ensure that surveys are as impactful as possible. It would also be beneficial for making choices between GA projects that can be efficiently accomplished within the core surveys and those that can only be done separately. Chapter 3 of this report gives an example of how such competition could be implemented together with a collaborative community process.

Conclusion: The process of selecting Roman’s observing plan would benefit from including both community collaboration and competition.

4. For the competitive process envisioned above to be effective, it must be arbitrated by an independent committee with the breadth and depth of expertise needed to evaluate competing considerations. NASA has long used such TACs and executive TACs for its other flagship missions (Hubble Space Telescope, Chandra X-ray Observatory, Spitzer Space Telescope), and the CAA suggests the establishment of a similar STAC to decide the final Roman observing time allocations before launch. The STAC would make choices among several proposals submitted by each CCS team, which may include GA science, and proposals submitted by teams advocating only GA science. Chapter 3 of this report provides more detail about the suggested nature of the STAC.

Conclusion: Roman’s science output would benefit from an observing plan selected by an independent STAC.

5. To make its decisions in the most informed way, the STAC must be able to assess the trade-offs between observing time and science deliverables for the survey options it receives, including both the CCS options and the GA-only surveys. The committee suggests that proposals include a quantitative sensitivity analysis relating observing time options with science deliverables. Smaller exploratory projects might be exempt.

Conclusion: To make optimal use of Roman observing time, proposing teams would provide the STAC with quantitative sensitivity analyses relating observing time options with science deliverables. Smaller exploratory projects might be exempt.

6. As noted earlier, the CAA has identified areas in which CCS might need significantly less time than planned in the current reference surveys to achieve their science goals, and the committee advocated that each CCS submit several survey options, some of which could include GA science objectives. It is conceivable that the best choice of overall time allocation is one in which the time allocated for each CCS is the absolute minimum needed to achieve the required science objectives. The CAA therefore advocates that among the options each CCS will propose, one will be for the minimum time needed to achieve the science objectives as prescribed by Astro2010. As discussed above, this option should also include quantitative explanations of the trade-offs

1. between observing time and science deliverables. For example, teams need to provide the STAC with quantitative discussion of the margins needed to control systematics. Given the existence of the current reference surveys, and absent discoveries of new systematic uncertainties, the committee anticipates that the combined minimum time to achieve the science objectives would be below 75 percent of net observing time, and might be well below that level, depending on the level of margins the STAC decides to adopt.

Conclusion: Among the options each CCS proposes, it will be beneficial to include one with the minimum time required for achieving the science objectives laid out by Astro2010.

7. The DRM demonstrates that the CCS can achieve their science goals in 75 percent (or less) of the observing time. The CAA has also noted that those science goals may be achieved in significantly less observing time, although a full quantitative analysis is not possible given the available information. The inclusion of GA may increase the time each CCS takes beyond their baseline science goals. At the same time, it is possible that community-initiated GA surveys could be very large, even comparable in time to the CCS. In principles 1 to 6, above, the committee has advocated for a combination of collaborative and competitive process that would lead to a final combined evaluation of all of Roman’s science output. This process may lead to having the CCS, including their GA expansions, extending beyond 75 percent of observing time. Alternatively, the STAC might choose one or more minimal CCS plans, resulting in a CCS observing time substantially lower than 75 percent of observing time and several large GA surveys. All of these outcomes are acceptable as long as Roman’s science objectives, as defined by Astro2010, can be achieved.

Conclusion: An independent STAC would optimize Roman’s scientific return without constraints of pre-set observing time allocations.

8. Community members have ideas for Roman GA surveys that could be transformational. It is not yet known which surveys will produce the most exciting science. They may range from quite small surveys to surveys that far exceed the average time of 14 days, which is currently envisioned for a typical GA survey. To maximize Roman’s science return, the CAA suggests that the competitive process described above would also consider a broad spectrum of sizes for GA surveys, from small to very large. To optimize observing efficiency, the Roman project may consider implementing a two-step proposal process, with the first phase providing opportunities to find observing synergies and to consolidate overlapping observational programs.

Conclusion: For a survey instrument, it is beneficial to establish a process of consolidating observational programs. However, it is conceivable that some programs cannot be consolidated, and the CAA encourages a selection process that allows a broad range of program sizes.

9. Due to cost considerations, NASA is currently planning up to 30 competitively selected GA observational projects during the 5-year prime mission. The committee received briefings indicating that with an additional cost of approximately $10 million, two to three times as many observational projects could be ingested, handled, and scheduled by IPAC and the Space Telescope Science Institute—Roman’s science support and operations centers, respectively. This estimate does not include other sources of support needed to carry out additional approved programs. It therefore finds that the marginal cost of receiving more GA proposals is relatively low. While the committee recognizes the benefits of merging smaller programs into larger surveys to create homogeneous data sets, it is also advantageous to have flexibility in the number

1. of observational projects. The optimal number of observational programs will likely become clearer as the mission develops, and both cost constraints and science drivers become clearer.

Conclusion: Roman’s science output may benefit from increasing the number of GA competed programs above 30. The final number of programs may be best determined by the STAC, or subsequent regular TAC reviews, so as to maximize the scientific return consistent with programmatic constraints at the time of the review.

10. Roman lies in a distinctly challenging regime as a large NASA mission where a substantial fraction of the observations are pre-planned. There are compelling science-driven reasons for this approach, as the committee heard in many presentations. At the same time, the high degree of pre-planning makes Roman susceptible to unexpected events such as mission-performance problems or science discoveries made close to the time of launch. Performance problems have occurred on other large NASA missions, such as the Chandra X-ray Observatory, the Compton Gamma-Ray Observatory, and the Hubble Space Telescope. NASA and the astronomical community adapted to Chandra’s and Hubble’s mission-performance anomalies through competed proposals that maximized the science return for the post-launch mission capabilities. Eventually Hubble, which was at low Earth orbit, was repaired through a servicing mission, but this option is not presently available for Roman at L2. Adaptability proved key for the scientific success of those missions. Flexibility of the observing program is also important for responding to scientific opportunities that appear after the pre-launch planning phase, or if the instrument performs better than predicted.

Conclusion: It would be beneficial to plan for post-launch flexibility should updates to the observing plan be warranted.

Chapter 4 of this report gives specific suggestions for how to plan for such additional flexibility.