Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Likelihood”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,009 records · Page 56

Search for gamma-ray emission from the Coma cluster with six years of Fermi-LAT data

We present results from γ-ray observations of the Coma cluster incorporating six years of Fermi-LAT data and the newly released “Pass 8” event-level analysis. Our analysis of the region reveals low-significance residual structures within the virial radius of the cluster that are too faint for a detailed investigation with the current data. Using a likelihood approach that is free of assumptions on the spectral shape we derive upper limits on the γ-ray flux that is expected from energetic particle interactions in the cluster. We also consider a benchmark spatial and spectral template motivated by models in which the observed radio halo is mostly emission by secondary electrons. In this case, the median expected and observed upper limits for the flux above 100 MeV are 1.7 × 10{sup −9} ph cm{sup −2} s{sup −1} and 5.2 × 10{sup −9} ph cm{sup −2} s{sup −1} respectively (the latter corresponds to residual emission at the level of 1.8σ). These bounds are comparable to or higher than predicted levels of hadronic gamma-ray emission in cosmic-ray (CR) models with or without reacceleration of secondary electrons, although direct comparisons are sensitive to assumptions regarding the origin and propagation mode of CRs and magnetic field properties. The minimal expected γ-ray flux from radio and star-forming galaxies within the Coma cluster is roughly an order of magnitude below the median sensitivity of our analysis.

79 ASTRONOMY AND ASTROPHYSICS↗

A BAYESIAN ESTIMATE OF THE CMB–LARGE-SCALE STRUCTURE CROSS-CORRELATION

Evidences for late-time acceleration of the universe are provided by multiple probes, such as Type Ia supernovae, the cosmic microwave background (CMB), and large-scale structure (LSS). In this work, we focus on the integrated Sachs–Wolfe (ISW) effect, i.e., secondary CMB fluctuations generated by evolving gravitational potentials due to the transition between, e.g., the matter and dark energy (DE) dominated phases. Therefore, assuming a flat universe, DE properties can be inferred from ISW detections. We present a Bayesian approach to compute the CMB–LSS cross-correlation signal. The method is based on the estimate of the likelihood for measuring a combined set consisting of a CMB temperature and galaxy contrast maps, provided that we have some information on the statistical properties of the fluctuations affecting these maps. The likelihood is estimated by a sampling algorithm, therefore avoiding the computationally demanding techniques of direct evaluation in either pixel or harmonic space. As local tracers of the matter distribution at large scales, we used the Two Micron All Sky Survey galaxy catalog and, for the CMB temperature fluctuations, the ninth-year data release of the Wilkinson Microwave Anisotropy Probe ( WMAP 9). The results show a dominance of cosmic variance over the weak recovered signal, due mainly to the shallowness of the catalog used, with systematics associated with the sampling algorithm playing a secondary role as sources of uncertainty. When combined with other complementary probes, the method presented in this paper is expected to be a useful tool to late-time acceleration studies in cosmology.

79 ASTRONOMY AND ASTROPHYSICS↗

Observation and characterization of a cosmic muon neutrino flux from the northern hemisphere using six years of IceCube data

The IceCube Collaboration has previously discovered a high-energy astrophysical neutrino flux using neutrino events with interaction vertices contained within the instrumented volume of the IceCube detector. We present a complementary measurement using charged current muon neutrino events where the interaction vertex can be outside this volume. As a consequence of the large muon range the effective area is significantly larger but the field of view is restricted to the Northern Hemisphere. IceCube data from 2009 through 2015 have been analyzed using a likelihood approach based on the reconstructed muon energy and zenith angle. At the highest neutrino energies between 194 TeV and 7.8 PeV a significant astrophysical contribution is observed, excluding a purely atmospheric origin of these events at 5.6σ significance. The data are well described by an isotropic, unbroken power-law flux with a normalization at 100 TeV neutrino energy of (0.90{sub −0.27}{sup +0.30})×10{sup −18} GeV{sup −1} cm{sup −2} s{sup −1} sr{sup −1} and a hard spectral index of γ=2.13±0.13. The observed spectrum is harder in comparison to previous IceCube analyses with lower energy thresholds which may indicate a break in the astrophysical neutrino spectrum of unknown origin. The highest-energy event observed has a reconstructed muon energy of (4.5±1.2) PeV which implies a probability of less than 0.005% for this event to be of atmospheric origin. Analyzing the arrival directions of all events with reconstructed muon energies above 200 TeV no correlation with known γ-ray sources was found. Using the high statistics of atmospheric neutrinos we report the current best constraints on a prompt atmospheric muon neutrino flux originating from charmed meson decays which is below 1.06 in units of the flux normalization of the model in Enberg et al.

79 ASTRONOMY AND ASTROPHYSICS↗

Comoving Stars in Gaia DR1: An Abundance of Very Wide Separation Comoving Pairs

The primary sample of the Gaia Data Release 1 is the Tycho-Gaia Astrometric Solution (TGAS): ≈2 million Tycho-2 sources with improved parallaxes and proper motions relative to the initial catalog. This increased astrometric precision presents an opportunity to find new binary stars and moving groups. We search for high-confidence comoving pairs of stars in TGAS by identifying pairs of stars consistent with having the same 3D velocity using a marginalized likelihood ratio test to discriminate candidate comoving pairs from the field population. Although we perform some visualizations using (bias-corrected) inverse parallax as a point estimate of distance, the likelihood ratio is computed with a probabilistic model that includes the covariances of parallax and proper motions and marginalizes the (unknown) true distances and 3D velocities of the stars. We find 13,085 comoving star pairs among 10,606 unique stars with separations as large as 10 pc (our search limit). Some of these pairs form larger groups through mutual comoving neighbors: many of these pair networks correspond to known open clusters and OB associations, but we also report the discovery of several new comoving groups. Most surprisingly, we find a large number of very wide (>1 pc) separation comoving star pairs, the number of which increases with increasing separation and cannot be explained purely by false-positive contamination. Our key result is a catalog of high-confidence comoving pairs of stars in TGAS. We discuss the utility of this catalog for making dynamical inferences about the Galaxy, testing stellar atmosphere models, and validating chemical abundance measurements.

79 ASTRONOMY AND ASTROPHYSICS↗

Interpreting High-resolution Spectroscopy of Exoplanets using Cross-correlations and Supervised Machine Learning

We present a new method for performing atmospheric retrieval on ground-based, high-resolution data of exoplanets. Our method combines cross-correlation functions with a random forest, a supervised machine-learning technique, to overcome challenges associated with high-resolution data. A series of cross-correlation functions are concatenated to give a “CCF-sequence” for each model atmosphere, which reduces the dimensionality by a factor of ∼100. The random forest, trained on our grid of ∼65,000 models, provides a likelihood-free method of retrieval. The precomputed grid spans 31 values of both temperature and metallicity, and incorporates a realistic noise model. We apply our method to HARPS-N observations of the ultra-hot Jupiter KELT-9b and obtain a metallicity consistent with solar (logM = − 0.2 ± 0.2). Our retrieved transit chord temperature (T=6000{sub −200}{sup +0}K) is unreliable as strong ion lines lie outside of the extent of the training set, which we interpret as being indicative of missing physics in our atmospheric model. We compare our method to traditional nested sampling, as well as other machine-learning techniques, such as Bayesian neural networks. We demonstrate that the likelihood-free aspect of the random forest makes it more robust than nested sampling to different error distributions, and that the Bayesian neural network we tested is unable to reproduce complex posteriors. We also address the claim in Cobb et al. 2019 that our random forest retrieval technique can be overconfident but incorrect. We show that this is an artifact of the training set, rather than of the machine-learning method, and that the posteriors agree with those obtained using nested sampling.

79 ASTRONOMY AND ASTROPHYSICS↗

Prospects for Characterizing the Haziest Sub-Neptune Exoplanets with High-resolution Spectroscopy

Observations to characterize planets larger than Earth but smaller than Neptune have led to largely inconclusive interpretations at low spectral resolution due to hazes or clouds that obscure molecular features in their spectra. However, here we show that high-resolution spectroscopy (R ∼ 25,000–100,000) enables one to probe the regions in these atmospheres above the clouds where the cores of the strongest spectral lines are formed. We present models of transmission spectra for a suite of GJ 1214b–like planets with thick photochemical hazes covering 1–5 μm at a range of resolutions relevant to current and future ground-based spectrographs. Furthermore, we compare the utility of the cross-correlation function that is typically used with a more formal likelihood-based approach, finding that only the likelihood-based method is sensitive to the presence of haze opacity. We calculate the signal-to-noise ratio (S/N) of these spectra, including telluric contamination, Required to robustly detect a host of molecules such as CO, CO{sub 2}, H{sub 2}O, and CH{sub 4} and photochemical products like HCN as a function of wavelength range and spectral resolution. Spectra in the M band require the lowest S/N{sub res} to detect multiple molecules simultaneously. CH{sub 4} is only observable for the coolest models (T {sub eff} = 412 K) and only in the L band. We quantitatively assess how these requirements compare to what is achievable with current and future instruments, demonstrating that characterization of small cool worlds with ground-based high-resolution spectroscopy is well within reach.

79 ASTRONOMY AND ASTROPHYSICS↗

THE CONTRIBUTION OF FERMI -2LAC BLAZARS TO DIFFUSE TEV–PEV NEUTRINO FLUX

The recent discovery of a diffuse cosmic neutrino flux extending up to PeV energies raises the question of which astrophysical sources generate this signal. Blazars are one class of extragalactic sources which may produce such high-energy neutrinos. We present a likelihood analysis searching for cumulative neutrino emission from blazars in the 2nd Fermi -LAT AGN catalog (2LAC) using IceCube neutrino data set 2009-12, which was optimized for the detection of individual sources. In contrast to those in previous searches with IceCube, the populations investigated contain up to hundreds of sources, the largest one being the entire blazar sample in the 2LAC catalog. No significant excess is observed, and upper limits for the cumulative flux from these populations are obtained. These constrain the maximum contribution of 2LAC blazars to the observed astrophysical neutrino flux to 27% or less between around 10 TeV and 2 PeV, assuming the equipartition of flavors on Earth and a single power-law spectrum with a spectral index of −2.5. We can still exclude the fact that 2LAC blazars (and their subpopulations) emit more than 50% of the observed neutrinos up to a spectral index as hard as −2.2 in the same energy range. Our result takes into account the fact that the neutrino source count distribution is unknown, and it does not assume strict proportionality of the neutrino flux to the measured 2LAC γ -ray signal for each source. Additionally, we constrain recent models for neutrino emission by blazars.

79 ASTRONOMY AND ASTROPHYSICS↗

Toward Rapid Transient Identification and Characterization of Kilonovae

With the increasing sensitivity of advanced gravitational-wave (GW) detectors, the first joint detection of an electromagnetic and GW signal from a compact binary merger will hopefully happen within this decade. However, current GW likelihood sky areas span ∼100--1000 deg{sup 2}, and thus it is a challenging task to identify which, if any, transient corresponds to the GW event. In this study, we make a comparison between recent kilonova/macronova light-curve models for the purpose of assessing potential light-curve templates for counterpart identification. We show that recent analytical and parameterized models for these counterparts result in qualitative agreement with more complicated radiative transfer simulations. Our analysis suggests that with improved light-curve models with smaller uncertainties it will become possible to extract information about ejecta properties and binary parameters directly from the light-curve measurement. Even tighter constraints are obtained in cases for which GW and kilonova parameter estimation results are combined. It will therefore be important to make comparisons and potentially combine parameter estimation with the kilonova and GW results. However, to be prepared for upcoming detections, more realistic kilonova models are needed. These will require numerical relativity with more detailed microphysics, better radiative transfer simulations, and a better understanding of the underlying nuclear physics.

79 ASTRONOMY AND ASTROPHYSICS↗

Constraining Temporal Oscillations of Cosmological Parameters Using SNe Ia

The existing set of SNe Ia is now sufficient to detect oscillatory deviations from the canonical ΛCDM cosmology. In this work we determine that the Fourier spectrum of the Pantheon data set of spectroscopically well-observed SNe Ia is consistent with the predictions of ΛCDM. We also develop and describe two complementary techniques for using SNe Ia to constrain those alternate cosmological models that predict deviations from ΛCDM that are oscillatory in conformal time. The first technique uses the reduced $χ^2$ statistic to determine the likelihood that the observed data would result from a given model. The second technique uses bootstrap analysis to determine the likelihood that the Fourier spectrum of a proposed model could result from statistical fluctuations around ΛCDM. We constrain three oscillatory alternate cosmological models: one in which the dark energy equation of state parameter oscillates around the canonical value of $w_Λ = -1$, one in which the energy density of dark energy oscillates around its ΛCDM value, and one in which gravity derives from a scalar field evolving under an oscillatory coupling. We further determine that any alternate cosmological model that produces distance modulus residuals with a Fourier amplitude of $\simeq$36 millimags is strongly ruled out, given the existing data, for frequencies between $\simeq$0.08 Gyr$^{-1} h_{100}$ and $\simeq$80 Gyr$^{-1} h_{100}$.

79 ASTRONOMY AND ASTROPHYSICS↗

Compressing the Cosmological Information in One-dimensional Correlations of the Lyman-$\alpha$ Forest

Observations of the Lyman-α forest from spectroscopic surveys such as the Baryon Oscillation Spectroscopic Survey or its extension, eBOSS, or the ongoing Dark Energy Spectroscopic Instrument (DESI) survey offer a unique window to study the growth of structure on megaparsec scales. Interpretation of these measurements is a complicated task, requiring hydrodynamical simulations to model and marginalize over the thermal and ionization state of the intergalactic medium. This complexity has limited the use of Lyα clustering measurements in joint cosmological analyses. In this work we show that the cosmological information content of the one-dimensional power spectrum (P 1D ) of the Lyman-α forest can be compressed into a simple two-parameter likelihood without any significant loss of constraining power. We simulate P 1D measurements from DESI using hydrodynamical simulations and show that the compressed likelihood is model independent and lossless, recovering unbiased results even in the presence of massive neutrinos or running of the primordial power spectrum.

79 ASTRONOMY AND ASTROPHYSICS↗

Multimodality in the Search for New Physics in Pulsar Timing Data and the Case of Kination-amplified Gravitational-wave Background from Inflation

We investigate the kination-amplified inflationary gravitational-wave background (GWB) interpretation of the signal recently reported by various pulsar timing array (PTA) experiments. Kination is a post-inflationary phase in the expansion history dominated by the kinetic energy of some scalar field, characterized by a stiff equation of state w = 1. Within the inflationary GWB model, we identify two modes that can fit the current data sets (NANOGrav and EPTA) with equal likelihood: the kination-amplification (KA) mode and the ordinary, no-kination-amplification (no-KA) mode. The multimodality of the likelihood motivates a Bayesian analysis with nested sampling. We analyze the free spectra of current PTA data and mock free spectra constructed with higher signal-to-noise ratios using nested sampling. The analysis of the mock spectrum designed to be consistent with the best fit to the NANOGrav 15 yr (NG15) data successfully reveals the expected bimodal posterior for the first time while excluding the reheating mode that appears in the fit to the current NG15 data, making a case for our correct and comprehensive treatment of potential multimodal posteriors arising from future PTA data sets. The resultant Bayes factor is $\mathcal{B}$ $\equiv$ Z no–KA /Z KA = 2.9 ± 1.9, indicating comparable statistical significance between the two modes. Given the theoretical model-building challenges of producing highly blue-tilted primordial tensor spectra, the KA mode has the advantage of requiring less blue primordial spectra, compared with the no-KA mode. The synergy between future cosmic microwave background polarization, pulsar timing, and laser interferometer measurements of gravitational waves will help resolve the ambiguity implied by the multimodal posterior in PTA-only searches.

Cosmology↗

CMB-S4: Foreground-cleaning Pipeline Comparison for Measuring Primordial Gravitational Waves

We compare multiple foreground-cleaning pipelines for estimating the tensor-to-scalar ratio, r, using simulated maps of the planned CMB-S4 experiment within the context of the South Pole Deep Patch. To evaluate robustness, we analyze bias and uncertainty on r across various foreground suites using map-based simulations. The foreground-cleaning methods include: a parametric maximum likelihood approach applied to auto- and cross-power spectra between frequency maps; a map-based parametric maximum-likelihood method; and a harmonic-space internal linear combination using frequency maps. We summarize the conceptual basis of each method to highlight their similarities and differences. To better probe the impact of foreground residuals, we implement an iterative internal delensing step, leveraging a map-based pipeline to generate a lensing B-mode template from the large aperture telescope frequency maps. Our results show that the performance of the three approaches is comparable for simple and intermediate-complexity foregrounds, with σ(r) ranging from 3–5 ×10 −4 . However, biases at the 1σ–2σ level appear when analyzing more complex forms of foreground emission. By extending the baseline pipelines to marginalize over foreground residuals, we demonstrate that contamination can be reduced to within statistical uncertainties, albeit with a pipeline-dependent impact on σ(r), which translates to a detection significance between 2σ and 4σ for an input value of r = 0.003. These findings suggest varying levels of maturity among the tested pipelines, with the auto- and cross-spectra-based approach demonstrating the best stability and overall performance. Moreover, given the extremely low noise levels, mutual validation of independent foreground-cleaning pipelines is essential to ensure the robustness of any potential detection.

astronomy data analysis↗

A Nonparametric Method for the Inference of Halo Occupation Distributions

The galaxy–halo connection traces processes by which galaxies form and evolve. The halo occupation distribution (HOD) describes the relationship between galaxies and their host dark matter haloes. Measurements of the galaxy two-point correlation function (2PCF) allow us to extract information about the HODs of observed galaxy samples. Several parametric HOD models have been proposed in the literature, but the choice of parameterization restricts the space of possible HODs. To resolve this issue, we introduce a nonparametric HOD fitting method in which we train an emulator to learn the mappings among the galaxy 2PCF, physical properties used to select galaxy samples, and the HOD, all obtained from simulated past light cones constructed with the Santa Cruz semianalytic model. Implementing this emulator within a likelihood analysis framework, we derive constraints on the HOD of a galaxy sample when provided with a measurement of its 2PCF. Using the emulator to accelerate likelihood evaluations, we test the nonparametric HOD approach on a set of 2PCFs for mock galaxy samples drawn from the TNG100-1 simulation and selected above threshold values of stellar mass and star formation rate. Our framework is able to recover TNG100-1 HODs within 0.2 dex. We use the TNG100-1 mocks to tune the reported uncertainties to estimate those expected in the analysis of observations. Comparing to parametric HOD modelling routines applied to the same mock galaxy samples, our approach consistently infers the HOD with comparable or greater precision and accuracy.

Kennedy, Jacob [Rutgers Univ., Piscataway, NJ (Uni↗

A Compound Poisson Generator Approach to Point-source Inference in Astrophysics

Abstract The identification and description of point sources is one of the oldest problems in astronomy, yet even today the correct statistical treatment for point sources remains one of the field’s hardest problems. For dim or crowded sources, likelihood-based inference methods are required to estimate the uncertainty on the characteristics of the source population. In this work, a new parametric likelihood is constructed for this problem using compound Poisson generator (CPG) functionals that incorporate instrumental effects from first principles. We demonstrate that the CPG approach exhibits a number of advantages over non-Poissonian template fitting (NPTF)—an existing method—in a series of test scenarios in the context of X-ray astronomy. These demonstrations show that the effect of the point-spread function, effective area, and choice of point-source spatial distribution cannot, generally, be factorized as they are in NPTF, while the new CPG construction is validated in these scenarios. Separately, an examination of the diffuse-flux emission limit is used to show that most simple choices of priors on the standard parameterization of the population model can result in unexpected biases: when a model comprising both a point-source population and diffuse component is applied to this limit, nearly all observed flux will be assigned to either the population or to the diffuse component. A new parameterization is presented for these priors that properly estimates the uncertainties in this limit. In this choice of priors, CPG correctly identifies that the fraction of flux assigned to the population model cannot be constrained by the data.

79 ASTRONOMY AND ASTROPHYSICS↗

Photometry of Outer Solar System Objects from the Dark Energy Survey. I. Photometric Methods, Light-curve Distributions, and Trans-Neptunian Binaries

We report the methods of and initial scientific inferences from the extraction of precision photometric information for the >800 trans-Neptunian objects (TNOs) discovered in the images of the Dark Energy Survey (DES). Scene-modeling photometry is used to obtain shot-noise-limited flux measures for each exposure of each TNO, with background sources subtracted. Comparison of double-source fits to the pixel data with single-source fits are used to identify and characterize two binary TNO systems. A Markov Chain Monte Carlo method samples the joint likelihood of the intrinsic colors of each source as well as the amplitude of its flux variation, given the time series of multiband flux measurements and their uncertainties. A catalog of these colors and light-curve amplitudes A is included with this publication. We show how to assign a likelihood to the distribution q(A) of light-curve amplitudes in any subpopulation. Using this method, we find decisive evidence (i.e., evidence ratio <0.01) that cold classical (CC) TNOs with absolute magnitude 6 < H$_{r}$ < 8.2 are more variable than the hot classical (HC) population of the same H$_{r}$, reinforcing theories that the former form in situ and the latter arise from a different physical population. Resonant and scattering TNOs in this H$_{r}$ range have variability consistent with either the HCs or CCs. DES TNOs with H$_{r}$ < 6 are seen to be decisively less variable than higher-H$_{r}$ members of any dynamical group, as expected. More surprising is that detached TNOs are decisively less variable than scattering TNOs, which requires them to have distinct source regions or some subsequent differential processing.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A NICER View of the Massive Pulsar PSR J0740+6620 Informed by Radio Timing and XMM-Newton Spectroscopy

We report on Bayesian estimation of the radius, mass, and hot surface regions of the massive millisecond pulsar PSR J0740+6620, conditional on pulse-profile modeling of Neutron Star Interior Composition Explorer X-ray Timing Instrument event data. We condition on informative pulsar mass, distance, and orbital inclination priors derived from the joint North American Nanohertz Observatory for Gravitational Waves and Canadian Hydrogen Intensity Mapping Experiment/Pulsar wideband radio timing measurements of Fonseca et al. We use XMM-Newton European Photon Imaging Camera spectroscopic event data to inform our X-ray likelihood function. The prior support of the pulsar radius is truncated at 16 km to ensure coverage of current dense matter models. We assume conservative priors on instrument calibration uncertainty. We constrain the equatorial radius and mass of PSR J0740+6620 to be ${12.39}_{-0.98}^{+1.30}$ km and ${2.072}_{-0.066}^{+0.067}$ M ⊙ respectively, each reported as the posterior credible interval bounded by the 16% and 84% quantiles, conditional on surface hot regions that are non-overlapping spherical caps of fully ionized hydrogen atmosphere with uniform effective temperature; a posteriori, the temperature is ${\mathrm{log}}_{10}(T\,[{\rm{K}}])={5.99}_{-0.06}^{+0.05}$ for each hot region. All software for the X-ray modeling framework is open-source and all data, model, and sample information is publicly available, including analysis notebooks and model modules in the Python language. Furthermore, our marginal likelihood function of mass and equatorial radius is proportional to the marginal joint posterior density of those parameters (within the prior support) and can thus be computed from the posterior samples.

79 ASTRONOMY AND ASTROPHYSICS↗

Maximizing ozone signals among chemical, meteorological, and climatological variability

The detection of meteorological, chemical, or other signals in modeled or observed air quality data – such as an estimate of a temporal trend in surface ozone data, or an estimate of the mean ozone of a particular region during a particular season – is a critical component of modern atmospheric chemistry. However, the magnitude of a surface air quality signal is generally small compared to the magnitude of the underlying chemical, meteorological, and climatological variabilities (and their interactions) that exist both in space and in time, and which include variability in emissions and surface processes. This can present difficulties for both policymakers and researchers as they attempt to identify the influence or signal of climate trends (e.g., any pauses in warming trends), the impact of enacted emission reductions policies (e.g., United States NO x State Implementation Plans), or an estimate of the mean state of highly variable data (e.g., summertime ozone over the northeastern United States). Here we examine the scale dependence of the variability of simulated and observed surface ozone data within the United States and the likelihood that a particular choice of temporal or spatial averaging scales produce a misleading estimate of a particular ozone signal. Our main objective is to develop strategies that reduce the likelihood of overconfidence in simulated ozone estimates. We find that while increasing the extent of both temporal and spatial averaging can enhance signal detection capabilities by reducing the noise from variability, a strategic combination of particular temporal and spatial averaging scales can maximize signal detection capabilities over much of the continental US. For signals that are large compared to the meteorological variability (e.g., strong emissions reductions), shorter averaging periods and smaller spatial averaging regions may be sufficient, but for many signals that are smaller than or comparable in magnitude to the underlying meteorological variability, we recommend temporal averaging of 10–15 years combined with some level of spatial averaging (up to several hundred kilometers). If this level of averaging is not practical (e.g., the signal being examined is at a local scale), we recommend some exploration of the spatial and temporal variability to provide context and confidence in the robustness of the result. These results are consistent between simulated and observed data, as well as within a single model with different sets of parameters. The strategies selected in this study are not limited to surface ozone data and could potentially maximize signal detection capabilities within a broad array of climate and chemical observations or model output.

54 ENVIRONMENTAL SCIENCES↗

Huge ensembles – Part 1: Design of ensemble weather forecasts using spherical Fourier neural operators

Abstract. Simulating low-likelihood high-impact extreme weather events in a warming world is a significant and challenging task for current ensemble forecasting systems. While these systems presently use up to 100 members, larger ensembles could enrich the sampling of internal variability. They may capture the long tails associated with climate hazards better than traditional ensemble sizes. Due to computational constraints, it is infeasible to generate huge ensembles (comprised of 1000–10 000 members) with traditional, physics-based numerical models. In this two-part paper, we replace traditional numerical simulations with machine learning (ML) to generate hindcasts of huge ensembles. In Part 1, we construct an ensemble weather forecasting system based on spherical Fourier neural operators (SFNOs), and we discuss important design decisions for constructing such an ensemble. The ensemble represents model uncertainty through perturbed-parameter techniques, and it represents initial condition uncertainty through bred vectors, which sample the fastest-growing modes of the forecast. Using the European Centre for Medium-Range Weather Forecasts Integrated Forecasting System (IFS) as a baseline, we develop an evaluation pipeline composed of mean, spectral, and extreme diagnostics. With large-scale, distributed SFNOs with 1.1 billion learned parameters, we achieve calibrated probabilistic forecasts. As the trajectories of the individual members diverge, the ML ensemble mean spectra degrade with lead time, consistent with physical expectations. However, the individual ensemble members' spectra stay constant with lead time. Therefore, these members simulate realistic weather states during the rollout, and the ML ensemble passes a crucial spectral test in the literature. The IFS and ML ensembles have similar extreme forecast indices, and we show that the ML extreme weather forecasts are reliable and discriminating. These diagnostics ensure that the ensemble can reliably simulate the time evolution of the atmosphere, including low-likelihood high-impact extremes. In Part 2, we generate a huge ensemble initialized each day in summer 2023, and we characterize the simulations of extremes.

Mahesh, Ankur↗