Search NASA⌕ Search

SEARCH · Search NASA

Results for “variability analysis”

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 973 records · Page 54

Development of techniques for measuring pilot workload

An objective method of assessing information workload based on physiological measurements was developed. Information workload, or reserve capacity, was measured using a visual discrimination secondary task and subjective rating of task difficulty. The primary task was two axis (pitch and roll) tracking, and the independent variables in this study were aircraft pitch dynamics and wind gust disturbances. The study was structured to provide: (1) a sensitive, nonloading measure of reserve capacity, and (2) an unencumbering reliable measurement of the psychophysiological state. From these, a measured workload index (MWI) and physiological workload index (PWI) were extracted. An important measure of the success of this study was the degree to which the MWI and PWI agreed across the 243 randomly-presented, four-minute trials (9 subjects X 9 tasks X 3 replications). The electrophysiological data collected included vectorcardiogaram, respiration, electromyogram, skin impedance, and electroencephalogram. Special computer programs were created for the analysis of each physiological variable. The digital data base then consisted of 82 physiological features for each of the 243 trials. A prediction of workload based on physiological observations was formulated as a simultaneous least-squares prediction problem. A best subset of 10 features was chosen to predict the three measures of reserve capacity. The cannonical correlation coefficient was .754 with a chi squared value of 91.3 which allows rejection of the null hypothesis with p of .995.

Spyker, D. A.↗

SU-E-T-547: A Method to Correlate Treatment Planning Issue with Clinical Analysis for Prostate Stereotactic Body Radiotherapy (SBRT)

Purpose: In this study, the algorithms and calculation setting effect and contribution weighing on prostate Volumetric Modulated Arc Therapy (VMAT) based SBRT were evaluated for clinical analysis. Methods: A low risk prostate patient under SBRT was selected for the treatment planning evaluation. The treatment target was divided into low dose prescription target volume (PTV) and high Dose PTV. Normal tissue constraints include urethra and femur head, and rectum was separated into anterior, lateral and posterior parts. By varying the constraint limit of treatment plan calculation setting and algorithms, the effect on dose coverage and normal tissue dose constraint parameter carried effective comparison for the nominal prescription and constraint. For each setting, their percentage differences to the nominal value were calculated with geometric mean and harmonic mean. Results: In the arbitrary prostate SBRT case, 14 variables were selected for this evaluation by using nominal prescription and constraint. Six VMAT planning settings were anisotropic analytic algorithm stereotactic beam with and without couch structure in grid size of 1mm and 2mm, non stereotactic beam, Acuros algorithm . Their geometry means of the variable sets for these plans were 112.3%, 111.9%, 112.09%, 111.75%, 111.28%, and 112.05%. And the corresponding harmonic means were 2.02%, 2.16%, 3.15%, 4.74%, 5.47% and 5.55%. Conclusions: In this study, the algorithm difference shows relatively larger harmonic mean between prostate SBRT VMAT plans. This study provides a methodology to find sensitive combined variables related to clinical analysis, and similar approach could be applied to the whole treatment procedure from simulation to treatment in radiotherapy for big clinical data analysis.

60 APPLIED LIFE SCIENCES↗

The Shortwave Cloud-SST Feedback Amplifies Multi-Decadal Pacific Sea Surface Temperature Trends: Implications for Observed Cooling

Climate models struggle to produce sea surface temperature (SST) gradient trends in the tropical Pacific comparable to those seen recently in nature. Here, we find that the magnitude of the cloud-SST feedback in the subtropical Southeast Pacific is correlated across models with the magnitude of Eastern Pacific multi-decadal SST variability. A heat-budget analysis reveals coupling between cloud-radiative effects, circulation, and SST gradients in driving multi-decadal variability in the Eastern Pacific. Using this relationship and observed feedback estimates, we find that internal Eastern Pacific SST variability is underestimated in most models. Adjusting for model bias increases the likelihood of generating a cooling trend at least as large as observations in preindustrial control simulations by ~56% on average. If models underestimate climate “noise,” as our results suggest, this bias should be accounted for when attributing the relative importance of forced versus unforced changes in the climate.

58 GEOSCIENCES↗

AN INDEPENDENT ANALYSIS OF THE BROWN DWARF ATMOSPHERE MONITORING (BAM) DATA: LARGE-AMPLITUDE VARIABILITY IS RARE OUTSIDE THE L/T TRANSITION

Observations of variability can provide valuable information about the processes of cloud formation and dissipation in brown dwarf atmospheres. Here we report the results of an independent analysis of archival data from the Brown dwarf Atmosphere Monitoring (BAM) program. Time series data for 14 L and T dwarfs reported to be significantly variable over timescales of hours were analyzed. We confirm large-amplitude variability (amplitudes >2%) for 4 out of 13 targets and place upper limits of 0.7%-1.6% on variability in the remaining sample. For two targets we find evidence of weak variability at amplitudes of 1.3% and 1.6%. Based on our revised classification of variable objects in the BAM study, we find strong variability outside the L/T transition to be rare at near infrared wavelengths. From a combined sample of 81 L0-T9 dwarfs from the revised BAM sample and the variability survey of Radigan et al., we infer an overall observed frequency for large-amplitude variability outside the L/T transition of 3.2{sub −1.8}{sup +2.8}%, in contrast to 24{sub −9}{sup +11}% for L9-T3.5 spectral types. We conclude that while strong variability is not limited to the L/T transition, it occurs more frequently in this spectral type range, indicative of larger or more highly contrasting cloud features at these spectral types.

79 ASTRONOMY AND ASTROPHYSICS↗

SU-C-304-02: Robust and Efficient Process for Acceptance Testing of Varian TrueBeam Linacs Using An Electronic Portal Imaging Device (EPID)

Purpose: The purpose of this project was to develop a process that utilizes the onboard kV and MV electronic portal imaging devices (EPIDs) to perform rapid acceptance testing (AT) of linacs in order to improve efficiency and standardize AT equipment and processes. Methods: In this study a Varian TrueBeam linac equipped with an amorphous silicon based EPID (aSi1000) was used. The conventional set of AT tests and tolerances was used as a baseline guide, and a novel methodology was developed to perform as many tests as possible using EPID exclusively. The developer mode on Varian TrueBeam linac was used to automate the process. In the current AT process there are about 45 tests that call for customer demos. Many of the geometric tests such as jaw alignment and MLC positioning are performed with highly manual methods, such as using graph paper. The goal of the new methodology was to achieve quantitative testing while reducing variability in data acquisition, analysis and interpretation of the results. The developed process was validated on two machines at two different institutions. Results: At least 25 of the 45 (56%) tests which required customer demo can be streamlined and performed using EPIDs. More than half of the AT tests can be fully automated using the developer mode, while others still require some user interaction. Overall, the preliminary data shows that EPID-based linac AT can be performed in less than a day, compared to 2–3 days using conventional methods. Conclusions: Our preliminary results show that performance of onboard imagers is quite suitable for both geometric and dosimetric testing of TrueBeam systems. A standardized AT process can tremendously improve efficiency, and minimize the variability related to third party quality assurance (QA) equipment and the available onsite expertise. Research funding provided by Varian Medical Systems. Dr. Sasa Mutic receives compensation for providing patient safety training services from Varian Medical Systems, the sponsor of this study.

07 ISOTOPE AND RADIATION SOURCES↗

Carbon storage cost modeling for the offshore Gulf of Mexico

Groundbreaking for geologic carbon storage (GCS) projects in the offshore Gulf of Mexico is imminent, and there is great interest in utilizing this region for GCS projects. Offshore saline reservoirs provide a significant and accessible resource for GCS. However, conducting GCS in the offshore environment will pose distinct challenges pertaining to site selection, operations, infrastructure use, and monitoring compared to operating onshore that ultimately affect technoeconomic assessment of offshore GCS projects. Carbon storage and transport costs are critical to project developers looking to deploy carbon storage in the offshore environment. We present CO2_S_COM_Offshore, a model developed by the National Energy Technology Laboratory (NETL) as a screening-level offshore saline GCS cost modeling tool. Based on NETL’s widely used CO2_S_COM cost model for onshore saline CS, CO2_S_COM_Offshore enables technoeconomic analysis of GCS in offshore areas. This model comprehensively incorporates multiple facets of offshore GCS projects, from regional evaluation and site selection to permitting, transport, operations, monitoring, site closure, and decommissioning. In general, the model can explore the cost implications for potential offshore GCS project(s) by enabling the user to change several project operational and financial attribute configurations. Key inputs include offshore storage formation options, CO2 injection rate and duration, infrastructure types, monitoring intensity, project financing, and post-injection site care duration. Supporting cost algorithms within CO2_S_COM_Offshore were compiled utilizing S&P Global’ s QUE$TORTM cost estimation software alongside a variety of open-source scientific literature. In addition to reviewing key model components, we discuss several sensitivity analyses, input variabilities, and results on analysis of break-even CO2 price required by a project based on different regulation/policy and operational scenarios for the offshore Gulf of Mexico. These results indicate the value of modeling offshore GCS specifically, and the potential of offshore GCS within a decarbonization value chain. Presented at the 41st USAEE/IAEE North American Conference, 3-6 November 2024, Baton Rouge, LA, United States.

Mark-Moser, Mackenzie K.↗

A comparison of segmented abdominopelvic fluid volumes with conventional CT signs of abdominal compartment syndrome in a trauma population

Purpose: To compare the utility of abdominopelvic fluid volume measurements with established computed tomography signs for refractory post-traumatic abdominal compartment syndrome. Methods: This retrospective observational cohort study included 64 consecutive adult trauma patients with preoperative CT and diagnosis of refractory abdominal compartment syndrome requiring decompressive laparotomy at a level I trauma referral center between 2004 and 2014. We hypothesized that abdominal fluid volume measurements would be more predictive of the need for early laparotomy than previously described conventional CT signs of refractory ACS. Abdominopelvic fluid volumes were determined quantitatively using semi-automated segmentation software. The following conventional imaging parameters were recorded: abdominal anteroposterior:transverse ratio (round belly sign); infrahepatic vena cava diameter; distal abdominal aortic diameter; largest single small bowel wall diameter; hydronephrosis, inguinal herniation; and mesenteric and body wall edema. For outcome analysis, patients were stratified into two groups: those who underwent early (< 24 h) and late (≥ 24 h) decompressive laparotomy following CT. Correlation analysis, comparison of means, and multivariate logistic regression were performed. Results: Abdominal fluid volumes (p = 0.001) and anteroposterior:transverse ratio (p = 0.009) were increased and inferior vena cava diameter (p = 0.009) was decreased in the early decompressive laparotomy group. Multivariate analysis including conventional CT variables, fluid volumes, and laboratory values revealed abdominal fluid volumes (p = 0.012; Δ in log odds of 1.002/mL) as the only independent predictor of early decompressive laparotomy. Conclusions: Segmented abdominopelvic free fluid volumes had greater predictive utility for decision to perform early decompressive laparotomy than previously described ACS-related CT signs in trauma patients who developed refractory abdominal compartment syndrome.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Potential Basin-Scale Estimates of Arctic Snow Depth with Sea Ice Freeboards from CryoSat-2 and ICESat-2: An Exploratory Analysis

The potential of deriving snow depth estimates using differences in freeboard heights from CryoSat-2 (CS-2) and ICESat-2 (IS-2) is examined. In our analysis, we use lidar freeboard from the Airborne Topographic Mapper (ATM) on Operation IceBridge (OIB) as proxy of IS-2 total (snow plus ice) freeboard. Snow depths are estimates from the OIB snow radar. Differences in height between the total (ATM) and ice (CS-2) freeboards are related to snow depth by the refractive index of the snow layer, which is dependent on snow density. For two years (2014 and 2015), regression of the ATM and CS-2 freeboard differences against OIB snow depth gives correlations of approximately 0.80, estimated eta (sub s) of 1.21, and standard errors of 8 centimeters. The resulting refractive index, eta (sub s), can be compared to that expected of the Arctic snow cover in early spring (1.25 plus or minus 0.05). The expected biases and variability in the regression analysis are discussed. Results suggest that snow depth can be estimated from the freeboard differences. The benefits of adjusting the orbit of CS-2 for providing more optimized overlaps between IS-2 and CS-2 are considered.

Sea ice↗

The RR Lyrae stars: New perspectives

We demonstrate that the Oosterhoff II (Oo II) RR Lyrae ab variables are hotter by ∼270 K, at the same period, than Oo I variables. Or, at the same ((B) – (V)){sub 0} value the Oo II variables have larger radii than Oo I variables. This accounts for the reason Oo II variables are brighter (0.12-0.20 mag) than Oo I variables. The dependence of the light amplitude of RR Lyrae variables on temperature is independent of Oo type. This makes it possible to derive an accurate set of equations to relate intrinsic (B – V){sub 0} color indices to light amplitudes, which in turn can be used to determine the interstellar reddening (E (B – V)). With just a few variables (∼5), it is possible to determine the E (B – V) to an accuracy of <0.01 mag in the absence of systematic photometric errors. We discuss the errors introduced in color excess determinations by including the Blazhko stars in a solution. A comparison of color excess values of 23 globular clusters and two regions of the Large Magellanic Cloud (LMC), determined with the aid of our newly developed equations, are found to compare favorably (∼0.01 mag) with color excess values found in the literature. Four new Oo III variables, some found in metal-poor clusters, are discussed. An analysis of the galactic-field variables indicates the majority are Oo I and Oo II variables, but a few short-period (log P < –0.36) metal-strong variables, so far not found in galactic globular clusters are evidently ∼0.30 mag fainter than Oo I variables. Oo III variables may also be present in the field. We conclude that the RR Lyrae ab variables are primarily restricted to four sequences or groups. If we assume that the Oo I variables' mean absolute magnitude is Mv = 0.61, the mean absolute magnitudes of the other three sequences are: short-period variables Mv ∼ 0.89 mag, Oo II Mv ∼ 0.43 mag, and Oo III Mv ∼ 0.29 mag. The Oo I fundamental RR Lyrae ab red edge (FRE) and fundamental blue edge (FBE) occur at approximately the following temperatures: FRE T ∼ 6180 K and FBE T ∼ 6750 K. There is a strong dependence of Mv on [Fe/H] as we proceed from the short-period variables to the Oo I variables and to the Oo II variables, but there seems to be little or no dependence of Mv on [Fe/H] for stars within a group, at least for the Oo I and Oo II groups. The Oo II variables exhibit a weak period luminosity relation in V in many globular clusters unlike the Oo II-like variables in Oo I clusters which do not exhibit a P-L relation. The properties of some intermediate LMC clusters are discussed.

79 ASTRONOMY AND ASTROPHYSICS↗

Software tools for the integration and execution of multidisciplinary analysis programs

The Environment for Application Software Integration and Execution (EASIE) software furnishes a methodology and a set of utility routines for design teams to use in building, maintaining, and applying CAD systems consituted by a large number of diverse, stand-alone analysis codes. EASIE uses a central data base that contains all variables required as inputs to the analysis programs that will make up the integrated system. Attention is given to the illustrative cases of three operational design systems based on EASIE.

Rowell, L. F.↗

Polarimetric Radar Variables in Snowfall at Ka- and W-Band Frequency Bands: A Comparative Analysis

Dual-frequency millimeter-wavelength radar observations in snowfall are analyzed in order to evaluate differences in conventional polarimetric radar variables such as differential reflectivity (ZDR) specific differential phase shift (K DP ) and linear depolarization ratio (LDR) at traditional cloud radar frequencies at Ka and W bands (~35 and ~94 GHz, correspondingly). Low radar beam elevation (~5°) measurements were performed at Oliktok Point, Alaska, with a scanning fully polarimetric radar operating in the horizontal–vertical polarization basis. This radar has the same gate spacing and very close beam widths at both frequencies, which largely alleviates uncertainties associated with spatial and temporal data matching. It is shown that observed Ka- and W-band Z DR differences are, on average, less than about 0.5 dB and do not have a pronounced trend as a function of snowfall reflectivity. The observed Z DR differences agree well with modeling results obtained using integration over nonspherical ice particle size distributions. For higher signal-to-noise ratios, KDP data derived from differential phase measurements are approximately scaled as reciprocals of corresponding radar frequencies indicating that the influence of non-Rayleigh scattering effects on this variable is rather limited. This result is also in satisfactory agreement with data obtained by modeling using realistic particle size distributions. Observed Ka- and W-band LDR differences are strongly affected by the radar hardware system polarization “leak” and are generally less than 4 dB. Smaller differences are observed for higher depolarizations, where the polarization “leak” is less pronounced. Realistic assumptions about particle canting and the system polarization isolation lead to modeling results that satisfactorily agree with observational dual-frequency LDR data.

42 ENGINEERING↗

Tumor Control Outcomes After Hypofractionated and Single-Dose Stereotactic Image-Guided Intensity-Modulated Radiotherapy for Extracranial Metastases From Renal Cell Carcinoma

Purpose: To report tumor local progression-free outcomes after treatment with single-dose, image-guided, intensity-modulated radiotherapy and hypofractionated regimens for extracranial metastases from renal cell primary tumors. Patients and Methods: Between 2004 and 2010, 105 lesions from renal cell carcinoma were treated with either single-dose, image-guided, intensity-modulated radiotherapy to a prescription dose of 18-24 Gy (median, 24) or hypofractionation (three or five fractions) with a prescription dose of 20-30 Gy. The median follow-up was 12 months (range, 1-48). Results: The overall 3-year actuarial local progression-free survival for all lesions was 44%. The 3-year local progression-free survival for those who received a high single-dose (24 Gy; n = 45), a low single-dose (<24 Gy; n = 14), or hypofractionation regimens (n = 46) was 88%, 21%, and 17%, respectively (high single dose vs. low single dose, p = .001; high single dose vs. hypofractionation, p < .001). Multivariate analysis revealed the following variables were significant predictors of improved local progression-free survival: 24 Gy dose compared with a lower dose (p = .009) and a single dose vs. hypofractionation (p = .008). Conclusion: High single-dose, image-guided, intensity-modulated radiotherapy is a noninvasive procedure resulting in high probability of local tumor control for metastatic renal cell cancer generally considered radioresistant according to the classic radiobiologic ranking.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Search for Hard X-Ray Emission from Aquila X-1: High Energy Emission from Gamma-ray Radio Star 2CG 135+1/LSI 61 305

Several investigations supported by these CCRO grant were completed or are close to completion. The study of EGRET data for the unidentified source 2CG 135+01 was very fruitful. We discovered transient gamma-ray emission by combining several data obtained since 1994 through 1997. It is the first time that time variable emission is established for this enigmatic source, and clearly an interpretation in terms of an isolated radio pulsar (Geminga-like) is disfavored now. Our preferred model is a Galactic source, probably an energetic pulsar (such as PSR129-63) in a binary system producing gamma-rays because of pulsar wind/mass outflow interaction. We also accumulated may data concerning the radio source LSI 61 303, the possible counterpart of 2CG 135+01. We show that a possible anti-correlation between radio and gamma-ray emission exists. This anticorrelation is evident only in the energy range above 100 MeV, as demonstrated by the lack of it obtained from OSSE data. If confirmed, this anti-correlation would prove to be very important for the interpretation of the hundreds of unidentified gamma-ray sources currently discovered by EGRET near the Galactic plane, and would point to a new class of sources in addition to AGNs and isolated pulsars. We also completed the analysis of several time variable gamma-ray sources near the Galactic plane, with the discussion of evidence for transient emission from 2EG J1813-12 and 2EG J1828+01. We completed several investigations regarding gamma-ray bursts (GRBs), including the study of the brightness distribution for different spectral/duration GRB sub-classes, an investigation of acceleration processes and their consequences for GRB afterglow emission [61, the application of the synchrotron shock model of GRBs to X-ray energies.

Tavani, Marco↗

A novel risk classification system for 30-day mortality in children undergoing surgery

A simple, objective and accurate way of grouping children undergoing surgery into clinically relevant risk groups is needed. The purpose of this study, is to develop and validate a preoperative risk classification system for postsurgical 30-day mortality for children undergoing a wide variety of operations. The National Surgical Quality Improvement Project-Pediatric participant use file data for calendar years 2012–2014 was analyzed to determine preoperative variables most associated with death within 30 days of operation (D30). Risk groups were created using classification tree analysis based on these preoperative variables. The resulting risk groups were validated using 2015 data, and applied to neonates and higher risk CPT codes to determine validity in high-risk subpopulations. A five-level risk classification was found to be most accurate. The preoperative need for ventilation, oxygen support, inotropic support, sepsis, the need for emergent surgery and a do not resuscitate order defined non-overlapping groups with observed rates of D30 that vary from 0.075% (Very Low Risk) to 38.6% (Very High Risk). When CPT codes where death was never observed are eliminated or when the system is applied to neonates, the groupings remained predictive of death in an ordinal manner.

60 APPLIED LIFE SCIENCES↗

Engineering Overview of a Multidisciplinary HSCT Design Framework Using Medium-Fidelity Analysis Codes

An objective of the HPCC Program at NASA Langley has been to promote the use of advanced computing techniques to more rapidly solve the problem of multidisciplinary optimization of a supersonic transport configuration. As a result, a software system has been designed and is being implemented to integrate a set of existing discipline analysis codes, some of them CPU-intensive, into a distributed computational framework for the design of a High Speed Civil Transport (HSCT) configuration. The proposed paper will describe the engineering aspects of integrating these analysis codes and additional interface codes into an automated design system. The objective of the design problem is to optimize the aircraft weight for given mission conditions, range, and payload requirements, subject to aerodynamic, structural, and performance constraints. The design variables include both thicknesses of structural elements and geometric parameters that define the external aircraft shape. An optimization model has been adopted that uses the multidisciplinary analysis results and the derivatives of the solution with respect to the design variables to formulate a linearized model that provides input to the CONMIN optimization code, which outputs new values for the design variables. The analysis process begins by deriving the updated geometries and grids from the baseline geometries and grids using the new values for the design variables. This free-form deformation approach provides internal FEM (finite element method) grids that are consistent with aerodynamic surface grids. The next step involves using the derived FEM and section properties in a weights process to calculate detailed weights and the center of gravity location for specified flight conditions. The weights process computes the as-built weight, weight distribution, and weight sensitivities for given aircraft configurations at various mass cases. Currently, two mass cases are considered: cruise and gross take-off weight (GTOW). Weights information is obtained from correlations of data from three sources: 1) as-built initial structural and non-structural weights from an existing database, 2) theoretical FEM structural weights and sensitivities from Genesis, and 3) empirical as-built weight increments, non-structural weights, and weight sensitivities from FLOPS. For the aeroelastic analysis, a variable-fidelity aerodynamic analysis has been adopted. This approach uses infrequent CPU-intensive non-linear CFD to calculate a non-linear correction relative to a linear aero calculation for the same aerodynamic surface at an angle of attack that results in the same configuration lift. For efficiency, this nonlinear correction is applied after each subsequent linear aero solution during the iterations between the aerodynamic and structural analyses. Convergence is achieved when the vehicle shape being used for the aerodynamic calculations is consistent with the structural deformations caused by the aerodynamic loads. To make the structural analyses more efficient, a linearized structural deformation model has been adopted, in which a single stiffness matrix can be used to solve for the deformations under all the load conditions. Using the converged aerodynamic loads, a final set of structural analyses are performed to determine the stress distributions and the buckling conditions for constraint calculation. Performance constraints are obtained by running FLOPS using drag polars that are computed using results from non-linear corrections to the linear aero code plus several codes to provide drag increments due to skin friction, wave drag, and other miscellaneous drag contributions. The status of the integration effort will be presented in the proposed paper, and results will be provided that illustrate the degree of accuracy in the linearizations that have been employed.

Weston, R. P.↗

AUTOMATED CLASSIFICATION OF VARIABLE STARS IN THE ASTEROSEISMOLOGY PROGRAM OF THE KEPLER SPACE MISSION

We present the first results of the application of supervised classification methods to the Kepler Q1 long-cadence light curves of a subsample of 2288 stars measured in the asteroseismology program of the mission. The methods, originally developed in the framework of the CoRoT and Gaia space missions, are capable of identifying the most common types of stellar variability in a reliable way. Many new variables have been discovered, among which a large fraction are eclipsing/ellipsoidal binaries unknown prior to launch. A comparison is made between our classification from the Kepler data and the pre-launch class based on data from the ground, showing that the latter needs significant improvement. The noise properties of the Kepler data are compared to those of the exoplanet program of the CoRoT satellite. We find that Kepler improves on CoRoT by a factor of 2-2.3 in point-to-point scatter.

79 ASTRONOMY AND ASTROPHYSICS↗

ON MACHINE-LEARNED CLASSIFICATION OF VARIABLE STARS WITH SPARSE AND NOISY TIME-SERIES DATA

With the coming data deluge from synoptic surveys, there is a need for frameworks that can quickly and automatically produce calibrated classification probabilities for newly observed variables based on small numbers of time-series measurements. In this paper, we introduce a methodology for variable-star classification, drawing from modern machine-learning techniques. We describe how to homogenize the information gleaned from light curves by selection and computation of real-numbered metrics (features), detail methods to robustly estimate periodic features, introduce tree-ensemble methods for accurate variable-star classification, and show how to rigorously evaluate a classifier using cross validation. On a 25-class data set of 1542 well-studied variable stars, we achieve a 22.8% error rate using the random forest (RF) classifier; this represents a 24% improvement over the best previous classifier on these data. This methodology is effective for identifying samples of specific science classes: for pulsational variables used in Milky Way tomography we obtain a discovery efficiency of 98.2% and for eclipsing systems we find an efficiency of 99.1%, both at 95% purity. The RF classifier is superior to other methods in terms of accuracy, speed, and relative immunity to irrelevant features; the RF can also be used to estimate the importance of each feature in classification. Additionally, we present the first astronomical use of hierarchical classification methods to incorporate a known class taxonomy in the classifier, which reduces the catastrophic error rate from 8% to 7.8%. Excluding low-amplitude sources, the overall error rate improves to 14%, with a catastrophic error rate of 3.5%.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗