Jim Myers

NCSA environmental cyberinfrastructure demonstration project: creating cyberenvironments for environmental engineering and hydrological science communities

NCSAs Environmental Cyberinfrastructure Demonstration (ECID) project is developing a cyberenvironment to support environmental science and hydrology research. The ECID Cyberenvironment is based on a set of coordinating technologies that together provide unique capabilities for integrating local and remote work and for capturing and exploiting data and interaction provenance. The CyberCollaboratory portal provides collaboration tools to discover, share, analyze, and discuss data and information. The CyberIntegrator is a meta-workflow engine combining heterogeneous (local and remote) tools into workflows to support complex scientific analyses and simulations. Both of these tools record rich provenance into the RDF-based Tupelo 2 Toolkit where it can be graphically browsed or used through CI-KNOW, social network analysis software, to provide context-specific recommendations within the CyberCollaboratory and CyberIntegrator. These capabilities, and the ability to dynamically connect new tools into group spaces, workflows, and provenance trails, will be critical to the next-generation of community-scale, persistent cyberinfrastructure efforts.

Web 2.0 geospatial visual analytics for improved urban flooding situational awareness and assessment

Situational awareness of urban flooding during storm events is important for disaster and emergency management. However, no general purpose tools yet exist for rendering rainfall accumulations in real-time at the resolution of hydrologic units used for modeling. This demonstration will exhibit a novel web 2.0 visual analytical approach for understanding and adaptively managing urban flooding issues. The approach generates a geospatial-temporal map of rainfall within urban hydrologic units (sewer-sheds) in real-time. The polygon-averaged rainfall data is generated using virtual sensors which provide customized real-time data products derived from National Weather Service weather radar data using NCSAs workflow tools. Time-series KML (Keyhole Markup Language) layers are generated, where each KML layer represents a particular slice of the geospatial color-coded sewershed map. Such time-aware KML can be replayed as a movie in the web-based Google Earth environment. This geospatial visual analytic approach can provide decision markers and communities a powerful resource for assessment of neighborhood flooding issues. We will demonstrate our technology using historical and real-time rainfall data in the metropolitan Chicago area to show the effectiveness of such approach. Future work by combining additional ground-truth flooding data will allow us move towards real-time improved decision support for flooding and stormwater management.

Investors’ Misweighting of Firm-level Information and Systematic Errors in Statistical Forecasts as Proxies for the Market’s Expectations of Earnings

Because investors’ expectations of future earnings are unobservable, a large body of literature uses the firm-specific consensus analyst forecast to proxy for these expectations. However, prior research documents systematic differences between market expectations and analyst forecasts with respect to their propensities to incorporate some firm-level information. Using data from 1986 through 2012, we examine the extent to which market expectations differ from analyst forecasts by developing an alternative proxy for the market’s expectations of future earnings as reflected in stock returns (i.e., the implied market forecast). Based on predictable errors in the implied market forecast, we create a trading strategy that generates excess returns (of about 8% per year) which cannot be explained by predictable errors in the consensus analyst forecast. In addition, our strategy, which uses predictable errors in the implied market forecast, outperforms a strategy that uses predictable errors in the consensus analyst forecast, especially in recent years. Overall, our results reveal the extent to which market expectations differ from analyst forecasts and underscore the importance of developing alternative proxies for market expectations.