Not too close! Evaluating the impact of the baseline on the localization of binary black holes by next-generation gravitational-wave detectors

This paper evaluates the impact of detector baseline on binary black hole localization for next-generation gravitational-wave observatories, finding that baselines of 2300–3300 km offer an optimal compromise for a two-detector network while demonstrating that adding a third detector (such as LIGO-India or the Einstein Telescope) is crucial for eliminating localization multimodality and ensuring robust sky localization.

The Gist

Imagine the universe is a giant, dark room, and black holes are two heavy dancers spinning toward each other. When they finally crash, they send out ripples in space-time called gravitational waves. Our goal is to find out exactly where in the room these dancers are so we can point our telescopes at them and see if there's any light (like a flash of gamma rays) accompanying the crash.

This paper is about building the best possible "ears" to listen to these ripples and figuring out how far apart we should place those ears to hear the location most clearly.

The Problem: The "Two-Ear" Dilemma

Right now, we have a few detectors (like LIGO and Virgo). When two detectors hear a sound, they can guess where it came from by comparing when the sound hit each ear. This is called triangulation.

Think of it like this:

• If you and a friend stand far apart and hear a clap, you can easily tell if the sound came from the left, right, front, or back based on the tiny split-second difference in when you heard it.
• If you and your friend stand right next to each other, the sound hits both of you at almost the exact same time. You have no idea where the clap came from; it could be anywhere in a giant circle around you.

The paper asks: How far apart should we place our next-generation gravitational wave detectors (called Cosmic Explorer or CE) to get the best "map" of the sky?

The Experiment: Testing the Distance

The authors simulated a network of two massive new detectors in the United States. They tested placing them at different distances apart, ranging from a short hop (about 600 km) to a long stretch across the country (about 4,500 km).

They looked at two main things:

1. The "Blur" (Localization Area): How big is the patch of sky where the black hole might be? We want this patch to be small, like a bullseye, so telescopes can find it.
2. The "Ghost Images" (Multimodality): Sometimes, with only two detectors, the math gets confused. Instead of one clear spot, the map shows two or more possible locations (like a "ghost image" in a mirror). This is bad because telescopes don't know which ghost to look at.

The Findings: "Not Too Close, Not Too Far"

The researchers found a "Goldilocks" zone for the distance between the detectors:

• Too Close (The "Huddle"): If the detectors are too close (less than 2,000 km), the "time difference" in hearing the wave is too small to measure accurately. The result? The map becomes a giant, blurry mess with multiple ghost images. It's like trying to find a whispering person when you are standing shoulder-to-shoulder with your friend; you can't tell which way the voice is coming from.
• Just Right (The "Sweet Spot"): A distance of about 2,300 to 3,300 km (roughly the distance from New York to Los Angeles) works best. This creates a clear time difference, shrinking the sky map to a manageable size and usually leaving just one or two possible locations.
• The "Three-Ear" Solution: The paper reveals a magic trick: Add a third detector.
  • Even if the two main detectors are close together, adding a third one (like the future LIGO-India or the Einstein Telescope in Europe) acts like a third ear.
  • This third ear instantly breaks the confusion. It eliminates the "ghost images" and gives a single, clear location for almost every event. It's the difference between guessing a location with two people and knowing it for sure with three.

Why Does This Matter?

If we can pinpoint the location of a black hole crash quickly and accurately:

1. We can catch the light: Telescopes can swing around and look at the exact spot to see if there is a flash of light, helping us understand the physics of these crashes.
2. We can measure the universe: By knowing exactly where the crash happened and how far away it is, we can measure how fast the universe is expanding (like using a "standard ruler" for the cosmos).
3. We save time: If the map is too big or has too many ghost locations, telescopes waste hours scanning empty sky. A precise map saves precious observing time.

The Bottom Line

To build the ultimate gravitational wave network, we shouldn't just put our new detectors anywhere. We need to space them out enough (thousands of kilometers) to get a good "time delay" signal. However, the most important lesson is that numbers matter. A network of just two detectors will often leave us guessing with multiple possible locations. But a network of three or more detectors (like two in the US, one in India, and one in Europe) will give us a crystal-clear view of the universe, eliminating the confusion and letting us see the cosmos in high definition.

Technical Summary

1. Problem Statement

Next-generation (XG) gravitational-wave (GW) detectors, specifically the Cosmic Explorer (CE) in the US and the Einstein Telescope (ET) in Europe, are expected to detect binary black hole (BBH) mergers at unprecedented rates and signal-to-noise ratios (SNRs). While high SNRs allow for precise reconstruction of intrinsic parameters (masses, spins), sky localization remains a critical challenge, particularly for short-duration signals (most BBHs) that spend only minutes in the detector band.

For short signals, localization relies primarily on timing triangulation across a network of detectors. A key design variable is the baseline (distance) between detectors.

• The Issue: If detectors are too close, the time-delay between signal arrivals is small relative to timing uncertainty, leading to poor localization and multimodal posteriors (multiple distinct sky regions with high probability). Multimodality hinders electromagnetic (EM) follow-up and statistical host galaxy identification ("dark sirens").
• The Question: What is the optimal baseline separation for a network of two CE detectors to balance localization accuracy, computational feasibility, and scientific utility (e.g., EM follow-up), and how does adding a third detector (LIGO-India or ET) mitigate multimodality?

2. Methodology

The authors employed a comprehensive simulation and analysis framework to evaluate 3D sky localization capabilities:

• Detector Networks:
  • Primary Focus: A network of two L-shaped CE detectors (40 km and 20 km arms) located on the US mainland.
  • Baseline Variation: The 20 km detector was placed at various distances from the 40 km detector, corresponding to light travel times of 2 ms to 15 ms (distances: 595 km to 4465 km).
  • Global Networks: Simulations included adding a third detector: LIGO-India (A+ and A# sensitivity), a single L-shaped ET, a triangular ET, or two misaligned L-shaped ET detectors.
• Signal Injection & Population:
  • Fixed Parameters: Equal-mass, non-spinning, face-on BBHs with total detector-frame masses ($M_{tot}$) ranging from 20 to 1000 $M_\odot$.
  • Realistic Population: A 1-year catalog of BBHs consistent with the latest GWTC-4 (LVK) results, including mass distributions, spins, and redshifts up to $z=10$.
  • SNR Scenarios: Events were scaled to network SNRs of 30, 60, and 120 (representing median, 68th, and 95th percentiles).
• Analysis Tools:
  • BAYESTAR: Used for rapid sky localization to capture multimodality (which Fisher matrix approximations miss).
  • Waveforms: IMRPhenomXAS (dominant mode only) for the main analysis; IMRPhenomXHM (including higher-order modes) for a subset to test robustness.
  • Metrics:
    • Fraction of sky with unimodal/bimodal posteriors ($f_{sky}$).
    • 90% credible localization area ($\Delta\Omega_{90\%}$).
    • Fraction of events where the entire posterior fits within the Field of View (FoV) of ZTF and LSST.
    • Fraction of events localized within a single hemisphere (observable by one ground-based instrument).
    • Multimodality in luminosity distance ($d_L$) posteriors.

3. Key Contributions

1. Quantification of Baseline Impact: The study provides a rigorous mapping of how detector separation affects the transition from unimodal to multimodal localization for XG detectors.
2. Identification of a "Sweet Spot": The authors identify a specific baseline range that offers a reasonable compromise between construction feasibility and scientific performance.
3. Third-Detector Necessity: The paper demonstrates that while baseline optimization helps, adding a third detector is the most effective method to eliminate multimodality for the vast majority of sources.
4. Realistic Population Forecasts: Unlike previous studies relying solely on fixed parameters, this work includes a realistic BBH population consistent with current LVK observations, providing actionable forecasts for the next decade of GW astronomy.
5. Dark Siren Forecasts: The paper estimates the number of "silver" ($\Delta\Omega \le 1$ deg$^2$) and "golden" ($\Delta\Omega \le 0.1$ deg$^2$) dark sirens achievable with different network configurations.

4. Key Results

A. Impact of Baseline (Two-CE Network)

• Optimal Baseline: A baseline corresponding to a light travel time of 8–11 ms (approx. 2300–3300 km) is identified as a reasonable compromise.
  • For $M_{tot} \lesssim 100 M_\odot$, this baseline yields predominantly unimodal or bimodal posteriors.
  • Shorter baselines (e.g., 595 km) significantly degrade localization, especially for high-SNR events, leading to highly multimodal posteriors where modes are widely separated.
• Multimodality:
  • At the shortest baseline (595 km) and high SNR (120), only ~25% of systems have at most two modes.
  • At the longest baseline (4465 km), >70% of systems up to 100 $M_\odot$ have at most two modes.
• EM Follow-up Feasibility:
  • ZTF/LSST Coverage: For the longest baseline, ~15% of events (at SNR=30) can be covered by a single telescope pointing. This drops to ~1% for the shortest baseline.
  • Hemisphere Constraint: For the shortest baseline, ~94% of events at SNR=30 have modes in different hemispheres, requiring multiple instruments for coverage. This improves to ~28% for the longest baseline.
• Luminosity Distance ($d_L$): Multimodal $d_L$ posteriors correlate with multimodal sky localization. While increasing SNR improves area, it can increase multimodality due to the separation of modes.

B. Impact of Mass Ratio and Inclination

• Mass Ratio: Varying the mass ratio ($q$) has a marginal effect on localization compared to the baseline. Higher mass ratios (more asymmetric) slightly improve localization for massive systems due to longer signal duration in band.
• Inclination: Edge-on systems ($\iota = 90^\circ$) show slightly worse sky localization but improved $d_L$ inference due to reduced degeneracy between inclination and distance.

C. Impact of Adding a Third Detector

• LIGO-India: Adding LIGO-India (even at A+ sensitivity) drastically reduces multimodality.
  • For the shortest baseline (595 km) with SNR=120, adding LIGO-India increases the fraction of unimodal events from <10% to >60%.
  • At A# sensitivity, this fraction rises further, effectively making the CE baseline less critical for events up to $M_{tot} \sim 300 M_\odot$.
• Einstein Telescope (ET):
  • A network of 2 CEs + ET (either triangular or 2L) results in unimodal posteriors for essentially 100% of events, regardless of the CE baseline.
  • Even a single L-shaped ET combined with 2 CEs achieves unimodality for ~94% of events.

D. Dark Sirens

• Silver Dark Sirens ($\Delta\Omega \le 1$ deg$^2$):
  • Longest baseline (4465 km): ~106 events/year.
  • Shortest baseline (595 km): ~28 events/year.
• Golden Dark Sirens ($\Delta\Omega \le 0.1$ deg$^2$): Very rare (2–4 events/year), but their localization areas are small enough for deep follow-up by facilities like ngVLA or Roman Space Telescope.

5. Significance and Conclusions

• Design Guidance: The paper argues that while maximizing baseline is beneficial, a separation of ~2300–3300 km (comparable to current LIGO Hanford-Livingston separation) is sufficient to ensure robust localization for the majority of BBHs without requiring the extreme separation of 4465 km.
• Network Synergy: The most critical finding is that baseline optimization alone is insufficient to eliminate multimodality for all sources. A global network including a third detector (LIGO-India or ET) is essential to guarantee unimodal posteriors, which is a prerequisite for efficient EM follow-up and precision cosmology (dark sirens).
• Conservative Estimates: The authors note their results are conservative because they focus on short-duration signals and use only the dominant waveform mode. Including higher-order modes and longer inspirals would likely improve localization further.
• Scientific Impact: These findings directly inform the siting and configuration of the Cosmic Explorer and the global XG detector network, ensuring that the next generation of GW astronomy can fully realize its potential for multi-messenger science and cosmology.