We present systematic spectral analyses of GRBs detected with the Burst and Transient Source Experiment (BATSE) onboard the Compton Gamma-Ray Observatory (CGRO) during its entire nine years of operation. This catalog contains two types of spectra extracted from 2145 GRBs and fitted with five different spectral models resulting in a compendium of over 19000 spectra. The models were selected based on their empirical importance to the spectral shape of many GRBs, and the analysis performed was devised to be as thorough and objective as possible. We describe in detail our procedures and criteria for the analyses, and present the bulk results in the form of parameter distributions. This catalog should be considered an official product from the BATSE Science Team, and the data files containing the complete results are available from the High-Energy Astrophysics Science Archive Research Center (HEASARC).

The distribution of GRB durations is bimodal, but there is little additional evidence to support the division of GRBs into short and long classes. Based on simple hardness ratios, several studies have shown a tendency for longer GRBs to have softer energy spectra. Using a database of standard model fits to BATSE GRBs, we compare the distributions of spectral parameters for short and long bursts. Our preliminary results show that the average spectral break energy differs discontinuously between short and long burst classes, but within each class shows only a weak dependence on burst duration.

Gamma-ray bursts provide what is probably one of the messiest of all astrophysical data sets. Burst class properties are indistinct, as overlapping characteristics of individual bursts are convolved with effects of instrumental and sampling biases. Despite these complexities, data mining techniques have allowed new insights to be made about gamma-ray burst data. We demonstrate how data mining techniques have simultaneously allowed us to learn about gamma-ray burst detectors and data collection, cosmological effects in burst data, and properties of burst subclasses. We discuss the exciting future of this field, and the web-based tool we are developing (with support from the NASA AISR Program). We invite others to join us in AI-guided gamma-ray burst classification (http://grb.mnsu.edu/grb/).

We describe the design of a suite of software tools to allow users to query Gamma-Ray Burst (GRB) data and perform data mining expeditions. We call this suite of tools a shed (SHell for Expeditions using Datamining). Our schedule is to have a completed prototype (funded via the NASA AISRP) by February, 2002. Meanwhile, interested users will find a partially functioning tool shed at http://grb.mankato.msus.edu.

We use ESX, a product of Information Acumen Corporation, to perform unsupervised learning on a data set containing 797 gamma-ray bursts taken from the BATSE 3B catalog. Assuming all attributes to be distributed logNormally, Mukherjee et al. (1998) analyzed these same data using a statistical cluster analysis. Utilizing the logarithmic values for T90 duration, total fluence, and hardness ratio HR321 their results showed the instances formed three classes. Class I contained long/bright/intermediate bursts, class II consisted of short/faint/hard bursts and class III was represented by intermediate/intermediate/soft bursts. When ESX was presented with these data and restricted to forming a small number of classes, the two classes found by previous standard techniques were determined. However, when ESX was allowed to form more than two classes, four classes were created. One of the four classes contained a majority of short bursts, a second class consisted of mostly intermediate bursts, and the final two classes were subsets of the Class I (long) bursts determined by Mukherjee et al. We hypothesize that systematic biases may be responsible for this variation.

The fluence duration bias causes fluences and durations of faint gamma-ray bursts to be systematically underestimated relative to their peak fluxes. Using Monte Carlo analysis, we demonstrate how this effect explains characteristics of structure of the fluence vs. 1024 ms peak flux diagram. Evidence of this bias exists in the BATSE fluence duration database, and provides a partial explanation for the existence of burst class properties.

The three gamma-ray burst (GRB) classes identified by statistical clustering analysis (Mukherjee et al. 1998) are examined using the pattern recognition algorithm C4.5 (Quinlan 1986). Although the statistical existence of Class 3 (intermediate duration, intermediate fluence, soft) is supported, the properties of this class do not need to arise from a distinct source population. Class 3 properties can easily be produced from Class 1 (long, high fluence, intermediate hardness) by a combination of measurement error, hardness/intensity correlation, and a newly-identified BATSE bias (the fluence duration bias). Class 2 (short, low fluence, hard) does not appear to be related to Class 1.

We present results of correlation studies, examining the association between the peak of the nu F_nu spectrum of gamma ray bursts, E_p, with the burst's energy fluence and photon peak flux. We discuss methods to account for data truncation in E_p and fluence or flux when performing the correlation analyses. However, because bursts near the detector threshold are not usually able to provide reliable spectral parameters, we focus on results for the brightest bursts in which we can better understand the selection effects relevant to E_p and burst strength. We find that there is a strong correlation between total fluence and E_p. We discuss these results in terms of both cosmological and intrinsic effects. In particular, we show that for realistic distributions of the burst parameters, cosmological expansion alone cannot account for the correlation between E_p and total fluence; the observed correlation is likely a result of an intrinsic relation between the burst rest-frame peak energy and the total radiated energy. We investigate this latter scenario in the context of synchrotron radiation from external and internal shock models of GRBs. We find that the internal shock model is consistent with our interpretation of the correlation, while the external shock model cannot easily explain this intrinsic relation between peak energy and burst radiated energy.

This is the first in a series of gamma-ray burst spectroscopy catalogs from the Burst And Transient Source Experiment (BATSE) on the Compton Gamma Ray Observatory, each covering a different aspect of burst phenomenology. In this paper, we present time-sequences of spectral fit parameters for 156 bursts selected for either their high peak flux or fluence. All bursts have at least eight spectra in excess of 45 sigma above background and span burst durations from 1.66 to 278 s. Individual spectral accumulations are typically 128 ms long at the peak of the brightest events, but can be as short as 16 ms, depending on the type of data selected. We have used mostly high energy resolution data from the Large Area Detectors, covering an energy range of typically 28 - 1800 keV. The spectral model chosen is from a small empirically-determined set of functions, such as the well-known `GRB' function, that best fits the time-averaged burst spectra. Thus, there are generally three spectral shape parameters available for each of the 5500 total spectra: a low-energy power-law index, a characteristic break energy and possibly a high-energy power-law index. We present the distributions of the observed sets of these parameters and comment on their implications. The complete set of data that accompanies this paper is necessarily large, and thus is archived electronically at: http://www.journals.uchicago.edu/ApJ/journal/.

All gamma-ray bursts are observed to have approximately the same characteristic gamma-ray energy. We show in this article that for bursts in the BATSE data set, this property as measured by the E-peak value is not an instrumental effect, but a physical property of gamma-ray bursts.

The Burst and Transient Source Experiment (BATSE) on the Compton Gamma Ray Observatory (CGRO) has triggered on 1637 cosmic gamma-ray bursts between 1991 April 19 and 1996 August 29. These events constitute the Fourth BATSE burst catalog. The current version (4Br) has been revised from the version first circulated on CD-ROM in September 1997 (4B) to include improved locations for a subset of bursts that have been reprocssed using additional data. A significant difference from previous BATSE catalogs is the inclusion of bursts from periods when the trigger energy range differed from the nominal 50-300 keV. We present tables of the burst occurrence times, locations, peak fluxes, fluences, and durations. In general, results from previous BATSE catalogs are confirmed here with greater statistical significance.

The synchrotron shock model (SSM) for gamma-ray burst emission makes a testable prediction: that the observed low-energy power-law photon number spectral index cannot exceed -2/3 (where the photon model is defined with a positive index: dN/dE is proportional to E^alpha). We have collected time-resolved spectral fit parameters for over 100 bright bursts observed by the Burst And Transient Source Experiment on board the Compton Gamma Ray Observatory. Using this database, we find 23 bursts in which the spectral index limit of the SSM is violated. We discuss elements of the analysis methodology that affect the robustness of this result, as well as some of the escape hatches left for the SSM by theory.

Artificial intelligence (AI) classifiers can be used to classify unknowns, refine existing classification parameters, and identify/screen out ineffectual parameters. We present an AI methodology for classifying new gamma-ray bursts, along with some preliminary results.

We report on the temporal behavior of the high-energy power law continuum component of gamma-ray burst spectra with data obtained by the Burst and Transient Source Experiment. We have selected 126 high fluence and high flux bursts from the beginning of the mission up until the present. Much of the data were obtained with the Large Area Detectors, which have nearly all-sky coverage, excellent sensitivity over two decades of energy and moderate energy resolution, ideal for continuum spectra studies of a large sample of bursts at high time resolution. At least 8 spectra from each burst were fitted with a spectral form that consisted of a low-energy power law, a spectral break at middle energies and a high-energy continuum. In most bursts (122), the high-energy continuum was consistent with a power law. The evolution of the fitted high-energy power-law index over the selected spectra for each burst is inconsistent with a constant for 34% of the total sample. The sample distribution of the average value for the index from each burst is fairly narrow, centered on -2.12. A linear trend in time is ruled out for only 20% of the bursts, with hard-to-soft evolution dominating the sample (100 events). The distribution for the total change in the power-law index over the duration of a burst peaks at the value -0.37, and is characterized by a median absolute deviation of 0.39, arguing that a single physical process is involved. We present analyses of the correlation of the power-law index with time, burst intensity and low-energy time evolution. In general, we confirm the general hard-to-soft spectral evolution observed in the low-energy component of the continuum, while presenting evidence that this evolution is different in nature from that of the rest of the continuum.