The nature of the progenitors of short duration, hard spectrum, gamma-ray bursts (GRBs) has remained a mystery. Even with the recent localizations of four short-hard GRBs, no transient emission has been found at long wavelengths that directly constrains the progenitor nature. Instead, as was the case in studying the different morphological subclasses of supernovae and the progenitors of long-duration GRBs, we suggest that the progenitors of short bursts can be meaningfully constrained by the environment in which the bursts occur. Here we present the discovery spectra of the galaxies that hosted three short-hard GRBs and the spectrum of a fourth host. The results indicate that these environments, both at the galaxy scale and galaxy-cluster scale, differ substantially from those of long-soft GRBs. The spatial offset of three bursts from old and massive galaxy hosts strongly favors an origin from the merger of compact stellar remnants, such as double neutron stars or a neutron-star black hole binary. The star-forming host of another GRB provides confirmation that, like supernovae of Type Ia, the progenitors of short-hard bursts are created in all galaxy types. This indicates a class of progenitors with a wide distribution of delay times between formation and explosion.

None

We describe a stochastic inversion method for mapping subsurface regions where CO{sub 2} saturation is changing. The technique combines prior information with measurements of injected CO{sub 2} volume, reservoir deformation and electrical resistivity. Bayesian inference and a Metropolis simulation algorithm form the basis for this approach. The method can (a) jointly reconstruct disparate data types such as surface or subsurface tilt, electrical resistivity, and injected CO{sub 2} volume measurements, (b) provide quantitative measures of the result uncertainty, (c) identify competing models when the available data are insufficient to definitively identify a single optimal model and (d) rank the alternative models based on how well they fit available data. We use measurements collected during CO{sub 2} injection for enhanced oil recovery to illustrate the method's performance. The stochastic inversions provide estimates of the most probable location, shape, volume of the plume and most likely CO{sub 2} saturation. The results suggest that the method can reconstruct data with poor signal to noise ratio.