GAIA is a virtual observatory designed to facilitate science. As the program evolves, we will develop modules that provide information and tools to carry out analysis and other technical tasks that are widely applied to data that falls under the GAIA umbrella. These tasks are being carried out by researchers around the globe, and it is not our intention to repeat such work. Rather, we want to bring such efforts to the fore within an environment that brings together those who are develpoing the tools and those who use them. The GAIA Science Center is a long-term program that will include at least the following:
Calibration - As the number of instruments grows the problem of how to meaningfully use, integrate, and compare data collected by different optical instruments and riometers is becoming acute. Accurate calibration of optical instruments is a cornerstone of quantitative auroral observations and fundamental to the science that follows from those observations. Calibration workshops are held in conjunction with the Annual European Meeting on Studies of the Atmosphere by Optical Methods. Calibration facilities and expertise exist in the UK, Scandinavia, Finland, the USA, and Japan. The Calibration Project will provide a forum for calibration resources and information to be shared and dissemintated.
Identification and Tracking of Injections - We are now learning how to identify dispersionless and disperseds injections in riometer and optical data. Particularly during the THEMIS era, characterizing the spatio-temporal evolution of injections is a valuable capability. The Injection Project will (1) provide tools for inferring injection boundaries from riometer and optical data and (2) regularly post analysis of riometer and optical data for specific injection events. The project is based on a number of published studies, and draws on data from single beam and imaging riometers, MSPs, ASIs, and satellite-borne global imagers.
Data Integration - A key element of the virtual observatory concept is the ability to use multiple observations from different instruments in common scientific investigations. For example, a typical investigation might compare space based auroral images with ground based all sky imagers to provide a multi-scale picture of auroral activity that is then used with ground radar and magnetometer observations and in situ satellite observations to study the nature of high-latitude forcing of the coupled ionosphere and thermosphere. There is currently no common methodology for easily sharing and comparing auroral image data. Furthermore, the technical capabilities for online access to auroral images varies widely between instrument teams and data repositories. GAIA provides a mechanism for finding and accessing multiple data sets, but the task of integrating multiple data into a common data set must still be done on an ad hoc basis. We plan to work with the space physics research community to develop tools, methodologies, and procedures to allow end users to seamlessly combine data sets from different instruments into a single dataset for scientific studies.
Identification of Auroral Boundaries - Auroral boundaries are a unifying theme for observations from different kinds of instruments. Boundaries can be found from imagery, precipitation, magnetometer and radar observations from both the ground and space. Boundaries have many uses, from indexing the state of the magnetosphere, to positioning models of auroral precipitation, to providing context for other observations, et cetera. Direct particle precipitation observations provide the most sensitive determinations of auroral boundaries, though the coverage is thin. The goal of the Auroral Boundary project is to provide similar (inter-calibrated) auroral boundary determinations from dissimilar observations. Benchmarking various boundary schemes versus particle precipitation based determinations provides the unifying glue. Progress can best be achieved through collaborations involving the relevant expertise.
Computer Vision and Pattern Recognition - Auroral observations produce enormous quantities of image data. These data are routinely used to extract boundaries and absolute intensities. Automatic algorithms exist to carry out such analysis. There is, however, additional important information in the auroral distribution in terms of texture and type. Presently, analysis of type and texture is both qualitative and done manually. While this is valuable, it is of course subjective (What is a patch? What is an arc? etc.). Furthermore, this qualitative analysis does not lend itself to using information about texture and type in conjunction with the outputs of, for example, global models. In the Computer Vision Project, we are developing tools for auromatic classification of the texture and type of aurora. The Computer Vision Project provides users of auroral data with access to these tools. Furthermore, it is developing classification of all auroral images that will be included in the GAIA metadata.The current state of Computer Vision as applied to auroral data is discussed on Mikko Syrjaesuo's Machine Vision webpage. You can also find an interactive example of Computer Vision applied to auroral images on Mikko's Content Based Image Retrieval site.