Solar activity is closely related to the near earth environment – summarized descriptively as space weather. Changes in space weather have adverse effect on many aspects of life and systems on earth and in space. Real-time, high-quality data and data processing would be a key element to forecast space weather promptly and accurately. Recently, we obtained a funding from US National Science Foundation to apply innovative information technology for space weather prediction.
(1) We use the technologies of image processing and pattern recognition, such as image morphology segmentation, Support Vector Machines (SVMs), and neural networks to detect and characterize three important solar activities in real-time: filament eruptions, flares, and emerging flux regions (EFRs). Combining the real time detection with the recent statistical study on the relationship among filament eruptions, flares, EFRs, coronal mass ejections (CMEs), and geomagnetic storms, we are establishing real time report of solar events and automatic forecasting of earth directed CMEs and subsequent geomagnetic storms. (2) We combine state-of-art parallel computing techniques with phase diverse speckle imaging techniques, to yield near real-time diffraction limited images with a cadence of approximately 10 sec. We utilize the multiplicity of parallel paradigms to optimize the calculation of phase diverse speckle imaging to improve calculation speed. With such data, we can monitor flare producing active regions continuously and carry out targeted studies of the evolution and flows in flare producing active regions. (3) We are developing Web based software tools to post our processed data, events and forecasting in real time, and to be integrated with current solar activity and space weather prediction Web pages at BBSO. This will also be a part of Virtual Solar Observatory (VSO) being developed by the solar physics community. This research is supported by NSF ITR program.