My research interests include magnetospheric physics and geomagnetism, space weather, solar wind/magnetosphere/ionosphere coupling, inverse methods for magnetospheric modeling, nonlinear dynamics, decision theory applied to rare event forecasting, and data science and management applications.

Abstract: The day-of-year (DOY) variation in the average value of a solar wind driver of geomagnetic activity has been shown to explain only a minority of the observed amplitude of DOY variation in geomagnetic activity. The proxies for solar wind driving used to show this are averages of a solar wind measurement in the same hour iphone photography or 3-hour interval as the geomagnetic activity measurement. This model of solar wind driving of geomagnetic activity does not account for the observation that the solar wind state in a given time interval can have an effect on activity that lasts many hours. In this work we show a model that includes the solar wind time history predicts a much higher DOY variation in both auroral zone and midlatitude geomagnetic activity. This model is used to estimate the solar wind contribution in two Field and Technical Services LLC geomagnetic activity measures that exhibit a semiannual DOY variation: the am index and postmidnight ground magnetic field measurements in the auroral zone. The estimated solar wind driver contribution to the DOY variation in the am index is 75%, which is approximately twice the amount of previous estimates. Solar wind driving is estimated to explain 40-60% of the DOY variation in the magnetic field measured by auroral zone ground magnetometers in the postmidnight sector, where previous estimates were near zero. [1]

Abstract: Solar wind density has been argued to have a strong effect on geomagnetic storms. Elevated solar wind density tends to occur in time in- tervals when the solar wind electric field is large. This complicates the anal- ysis required to identify a solar wind density influence, because the solar wind electric field is the dominant driver of geomagnetic storms. Statistical stud- ies have consistently shown that the independent correlation between solar wind density and geomagnetic storm intensity (via a proxy, such as the Dst index) is small. Modeling considerations predict a significant geomagnetic storm dependence on the plasma sheet density, which is indirectly connected to solar wind density. In this work, the solar wind density influence is quan- tified using two statistical measures: (1) data–derived impulse response func- tions and (2) the relationship between the integrated value of Dst to the in- tegrated value of the solar wind electric field during geomagnetic storm in- tervals. Results from both approaches indicate that the solar wind density modifies the geoefficiency, or the ability of a given value of the solar wind electric field to create a Dst disturbance. The impulse response method also predicts that solar wind density explains the difference in geoefficiency, as opposed to the solar wind dynamic pressure. Although the geoefficiency ef- fect is large, its influence is shown to be small when only large storms are considered, because large storms typically have large density. Weigel, 2010

Abstract: A statistical model of short time-scale geomagnetic fluctuations is developed and used to evaluate how geomagnetic dynamics are influenced by different solar wind controlling parameters. The functional form of the probability distribution function (PDF) that describes extreme-value (greater than 4σ) minute-to-minute changes in the ground magnetic field (Δx) at magnetometer station Sodankylä (geomagnetic latitude and longitude of [63.87,107.61]) is shown to be nearly independent of the variables solar wind (SW) forcing, local time (LT), and day of year (DOY). Instead of modifying the intrinsic dynamics, as characterized by the functional form of the PDF of Δx, these variables are shown either to amplify or reduce the absolute level of variability of the fluctuations: The primary difference in the PDF tail of Δx during weak and strong solar wind forcing is the standard deviation, σ the functional form of the PDF = f[Δx/σ(DOY,LT,SW)] is nearly invariant. In a statistical interpretation, we conclude that differences in solar-generated conductivity, seasonal effects, strength of solar wind forcing and variability, and position of the magnetometer ground station in local time do not change the structure of the extreme-value dynamics, as characterized by the probability distribution of Δx, but they serve to amplify the intrinsic variability [2].

To date, most algorithms developed to predict a magnetospheric space weather time series involve the filtering of data measured by a satellite upstream of Earth. The filters are optimized to have the minimum mean of the square prediction errors (MSE). This is in contrast to what the users of a prediction a user wants to know: will something happen? A filter optimized to give a small MSE will not generally give the best answer to the user-relevant question. These ideas have been developed in the paper Decision theory and the analysis of rare event space weather forecast

Abstract: Several basic results from decision theory as applied to rare event forecasts are reviewed, and an alternative method for comparing rare event forecasts is presented. A fundamental result is that for a large class of users only interested in economic utility, the relevant performance quantity is the number of correct and false alarm forecasts. This is contrasted with the reality that most forecast models are optimized to have a high data-model correlation, which does not always correspond to maximum economic utility. The value score (VS) developed by Wilks (2001) partially resolves this disconnect between modeler- and user-relevant metrics. Although the value score is closer to what is most likely of interest to a user, maximal VS does not necessarily correspond to maximal utility for the realistic case where the cost and benefit are dependent on the amplitude of the forecasted event. An alternative comparison and presentation method is proposed which may resolve this problem. For the class of users considered, full specification of model performance requires computation of the probability of correct, false alarm, and missed forecasts at several amplitude levels and warning time spans. Examples of the computations involved for the modeler and user are given for predictions of large-amplitude energetic electron fluence and geomagnetic storms parameterized by the Dst index.

pdf.

Abstract: The solar-wind-driven magnetosphere-ionosphere exhibits a variety of dynamical states including low-level steady plasma convection, episodic releases of geotail stored plasma energy into the ionospheric known broadly as substorms, and states of continuous strong unloading. The WINDMI model [J. P. Smith et al., J. Geophys. Res. 105, 12 983 (2000)] is a six-dimensional substorm model that uses a set of ordinary differential equations to describe the energy flow through the solar wind-magnetosphere-ionosphere system. This model has six major energy components, with conservation of energy and charge described by the coupling coefficients. The six-dimensional model is investigated by introducing reductions to derive a new minimal three-dimensional model for deterministic chaos. The reduced model is of the class of chaotic equations studied earlier [J. C. Sprott, Am. J. Phys. 68, 758 (2000)]. The bifurcation diagram remains similar, and the limited prediction time, which is in the range of three to five hours, occurs in the chaotic regime for both models. Determining all three Lyapunov exponents for the three-dimensional model allows one to determine the dimension of the chaotic attractor for the system. [3].

Abstract: A spatial and dynamic study of solar wind driving of high-latitude ground magnetic fields and their time derivatives is given. Nonlinear, data-derived basis functions are used to determine the optimal solar wind driven dynamic coupling function at high-latitude locations. The method does not assume a solar wind driving function a priori so that an unbiased determination of the relative influence of different driving processes can be made. Using this method, the locations where high-latitude models capture signatures of different driving processes, such as reconnection and the Kelvin-Helmholtz instability, are revealed for both the amplitude of the field and its time derivative. The time rate of change of the field is a disturbance measure related to geomagnetically induced currents. For the amplitude of the ground magnetic field the primary driver is reconnection; the time rate of change of the field has signatures of both reconnection driving in the nightside sector and Kelvin-Helmholtz driving in the prenoon sector. The independent influence of the solar wind ion density and azimuthal component of the magnetic field in driving magnetic field fluctuations is found to be small at auroral-zone latitudes. The data-derived coupling functions are also used to estimate the expected prediction error of a general class of models that specify the ground magnetic field or its time derivative given solar wind plasma measurements. Prediction efficiencies as small as zero and as large as 0.7 for both the amplitude of the field and its time derivative are possible. The prediction efficiency is highly dependent on spatial location and the direction of field being predicted. [4]

http://tsds.net/ Abstract: Time Series Data Server (TSDS) is a software package for implementing a server that provides fast super-setting, sub-setting, filtering, and uniform gridding of time series-like data. TSDS was developed to respond quickly to requests for long time spans of data. Data may be served from a fast database, typically created by aggregating granules (e.g., data files) from a remote data source and storing them in a local cache that is optimized for serving time series. The system was designed specifically for time series data, and is optimized for requests where the longest dimension of the requested data structure is time. Scalar, vector, and spectrogram time series types are supported. The user can interact with the server by requesting a time series, a date range, and an optional filter to apply to the data. Available filters include strides, block average/minimum/maximum, exclude, and inequality. Constraint expressions are supported, which allow such operations as a request for data from one time series when a different time series satisfied a specified relationship. TSDS builds upon DAP (Data Access Protocol), NcML (netCDF Mark-up language) and related software libraries. In this work, we describe the current design of this server, as well as planned features and potential implementation strategies.

In space physics, there is a very large effort to create standard metadata for data products [5]. The most important reason for this is to enable search; the number of data products is increasing, and with the advent of small satellites [6], [7], we will likely double the number of available data products in a very short amount of time (I talk more about this here [link]). In our proposal for the ViRBO project, we laid out a plan to build on an existing code base that had metadata search, discovery, access, version control, and edit functions. The developers of this code base are from GC RAS and NGDC, and as part of the ViRBO project, we have significantly extended its capabilities.

Abstract: The recent Heliophysics Virtual Observatory (VxO) effort involves the development of separate observatories with a low overlap in physical domain or area of scientific specialization and a high degree of overlap in metadata management needs. VxOware is a content and metadata management system. While it is intended for use by a VxO specifically, it can also be used by any entity that manages structured metadata. VxOware has many features of a content management system and extensively uses the W3C recommendations for XML (Extensible Markup Language), XQuery (XML Query), and XSLT (Extensible Style Sheet Language Transformations). VxOware has features such as system and user administration, search, user-editable content, version tracking, and a wiki. Besides virtual observatories, the intended user-base of VxOware includes a group or an instrument team that has developed a directory structure of data files and would like to make this data, and its associated metadata, available in the virtual observatory network. One of the most powerful features of VxOware is the ability to link any type of object in the observatory to other objects and the ability for every object to be tagged.

http://bu.edu/cism/cismdx

Abstract Among the many challenges facing space physics today is the need for an analysis and visualization package which can examine the results from numerical and empirical models as well as observational data. In this paper we introduce the CISM-DX package as a possible answer to this problem. We have created a tool which is capable of completing highly sophisticated visualizations of the results from a diverse set of models throughout the connected Sun-Earth system. We begin with an overview of the open source software programs, OpenDX and Octave, which form the basis of our package. Next we discuss the extensions to these programs that we have added to enhance their utility to the space physics community. We also discuss the extensive database of observations that is included with CISM-DX. Results from heliospheric simulations are used to highlight the ease at which a novice user can begin to utilize these tools. An energy budget calculation during a magnetospheric substorm simulation illustrates the more advanced analysis possible with the CISM-DX package. Next we highlight the extensive database of observational measurements included with the distribution, along with tools for displaying them. This package is being released under an open source licensing agreement to facilitate and encourage community use and contribution. Full Article