-------------------------------- REPORT ON 1993 SNOWMASS WORKSHOP -------------------------------- GEM WORKING GROUP 5: GGCM ASSEMBLY Co-chairs: G. Siscoe and J. Fedder GEM is a community-wide project with a programmatic goal to systematically integrate the broad, diverse results of magnetospheric research. For this it adopts the operational goal of creating a community, prediction-quality GGCM. This goal meets the integration requirement since, by its nature, a GGCM must integrate all aspects of magnetospheric research. It also meets an important validation requirement for, according to Richard Feynman, 'The test of a science is its ability to predict'. Applied to us, Feynman's criterion means that the ultimate test of our understanding of the magnetosphere is whether we can make a GGCM predict magnetospheric behavior. Most importantly, a GGCM would be a valuable research tool, a "national asset as WG 5's tutorial lecturer Mike Heinemann put it. It could be used for research, forecasting, and environment specification. Besides the GEM community, "customers" would include participants in ISTP and STEP, the Air Force Air Weather Office, and through NOAA's solar-terrestrial forecast office, space-environment-sensitive businesses like communication and power companies. Thus constructing a GGCM is important, worthy of concerted community effort; but WG 5's participants universally characterized the task as "daunting." The magnetosphere is big and complex, and we are all specialists in some aspects of its phenomena. To move forward against this challenge, we took for guidance Roger Revel's environmental slogan, "Think globally, act locally." For us this means work on interface problems that emerge when trying to assemble separate research results into a global description of magnetospheric behavior. There are three types of interface problems: cross-module (or intermodule) coupling, cross-scale coupling, and time-dependent coupling. The WG heard presentations on each type. While illustrating many crucial interface problems, the presentations were examples of individual research results more than progress reports from coordinated task groups. The WG needs such examples, however, to formulate a coordinated program. FIRST SESSION (which actually came second) Mike Heinemann led the discussion on intermodule coupling which included presentations by Dick Wolf, Cindy Cattell, Frank Toffoletto, and Tom Sotirelis. To preface these presentations we note that strong intermodule coupling gives the magnetosphere exceptional global coherence, which distinguishes it from the other geospheres. Coherence arises because magnetic field lines readily transmit currents and electric fields. Currents communicate stress, and electric fields communicate motion. The first of these, mechanical stress, is applied by magnetic strain or equivalently by the JxB force. Since field lines link the system globally, strains with their stresses operate globally. In JxB terms, strains become currents which circulate globally. Our discipline favors the JxB description, and thus speaks in terms of coupling by currents. That is, spatially separated regions are coupled by the currents that link them. In some cases, illustrated by field-aligned currents, the linking is direct; in others cases, illustrated by the Chapman-Ferraro current, the linking is inductive. Coupling currents that arise from magnetospheric convection include the regions 1 and 2 currents, cusp-mantle currents, and the tail current. Motion is the second mentioned of the signals transmitted by field lines; but it is not independent of the first, mechanical stress, since a proper description of convection requires self-consistent motion and stress fields--or equivalently, self-consistent electric fields and currents. Guaranteeing this self-consistency is the main form, in one guise or another, that interface problems take in intermodule coupling. Space physics achieved a major success in the early '70's when Vasyliunas and others solved this self-consistency problem for the inner magnetosphere. The solution linked the inner magnetosphere and the ionosphere. It predicted the region 2 currents and quantitatively described their properties. Attention now centers on developing a similar, quantitative theory for the region 1 currents, which with the tail current form the main power conduit from the solar wind to the magnetosphere. The task of developing a theory of the region 1 currents has two parts which can be expressed as two questions: where are region 1 currents generated? and what force generates them? Proposed answers to the "where?" question include the bow shock, the magnetosheath, the high-latitude boundary layer, the low-latitude boundary layer, and the plasma sheet. Proposed answers to the "what force?" question are similarly exhaustive: the inertial force, the viscous force, and the pressure-gradient force, which in more elaborate models becomes the divergence of the pressure tensor. Until this workshop, however, only a LLBL-viscous coupling model had been treated self-consistently in the manner of the region 2 currents. Starting from there, the workshop presentations included results of testing one version of this model, descriptions of extensions of it, and self-consistent alternatives to it. Dick Wolf showed results from using the Rice Convection Model (RCM) to explore possible roles of the pressure gradient force in generating region 1 currents. The pressure gradient force plays the dominant role in the region 2 success story, and the RCM is well suited to test how well it generates field-aligned currents in other situations. With coworkers at Rice, he ran three cases: reverse flow (northward IMF), dayside boundary layer; direct flow (southward IMF), dayside boundary layer; and direct flow, (primarily) nightside plasma sheet. Region 1-polarity currents resulted in each case. But also in each case, ionospheric currents ran out of the modeled domain down potential into the polar cap, and symmetrically down-potential out of the polar cap into the modeled domain. Wolf noted that this result implies that there must be another source of region 1-type field- aligned currents at higher latitudes. The relative strengths of the modeled and unmodeled region 1 field-aligned currents depends on such model parameters as the distribution of ionospheric conductivity. Wolf concluded that the pressure gradient force is probably supplying some of the region 1 currents, but it is evidently not supplying all of them. Although Bill Lotko's presentation was in another session, part of what he said belongs here. He and his coworkers at Dartmouth have extended the viscous low-latitude boundary layer model to include the effects of pressure gradient parallel to the boundary layer. Dartmouth has an analytical model, rather than a simulation model like the RCM. This new effort makes the model two-dimensional. For the first time it could reveal the phenomenon of entrainment of the sunward magnetospheric flow into the anti-sunward flowing boundary layer. The picture remarkably resembles one from the RCM run for the same physics (sans viscosity). Lotko's and Wolf's talks showed that the field is making steady and significant progress in determining the properties of the various options enumerated above as candidates for the source regions and generators of the region 1 currents. Undertaking a systematic and thorough determination of these properties is one way to narrow the options, for some of the predicted properties might be in essential conflict with observations. But as Cindy Cattell reported, comparing theory and data is not straight-forward and can give ambiguous results. She used magnetic field data from the S-3 satellite to look for the region 0 currents predicted by the version of the viscous, LLBL model published by Siscoe, Lotko, and Sonnerup. In some cases the currents were there but in others they were absent. A subsequent 2D treatment of the model showed that region 0 currents are restricted in local time. Siscoe was tasked with providing Cattell with model runs for conditions corresponding to the local times and transpolar potentials of the S-3 data. All these region 1 models require as a boundary condition the electrical potential around the polar cap. Specifying this potential is therefore an interface problem. Frank Toffoletto described advances in the capability of Toffoletto-Hill electric field mapping algorithm to specify the polar cap potential. These advances include the following: moveable dayside x-line for simulating IMF Bx effects; a realistic and variable magnetopause shape; arbitrary normal component to the magnetopause; expansion fan geometry with a distant x-line; return flow from the distant x-line; and the Hilmer-Voigt interior field model with flexible ring current and tail current representations. It was clear that the algorithm is already a highly flexible and powerful tool. It was the closest thing we had yet seen to a completed module for a modularized GGCM. The Toffoletto-Hill algorithm is designed for use with time independent or quasi-static convection models. We must also start thinking about the inherently time dependent convection of substorms and the interface problems associated with it. For example, research on the onset of the substorm expansive phase now focuses on the inner edge of the current sheet, where the current is strongest and from whence onset seems to emanate. One likely scenario associates onset with some threshold of current strength. The current strengthens during the growth phase until this threshold is reached, then the expansive phase begins. While determining the nature of the threshold is a matter for the substorm campaign, quantifying the strengthening of the current during the growth phase is a coupling problem. It is part of the overall configurational change attending the growth phase and involves the inductive interaction of all the major current systems: Chapman-Ferraro, region 1, and tail. So far, however, no one has managed to solve this global reconfigurational problem in 3D. (In 1968 Unti and Atkinson solved a limited, 2D version of it, and Coroniti and Kennel gave an approximate 3D solution in 1972).) But recently Tom Sotirelis has packaged some of the modeling ideas of David Stern into a promising approach to the 3D problem. He combines an elliptical magnetopause with a Tsyganenko current sheet to calculate the magnetic field everywhere within the magnetosphere and tail. The model has the flexibility of adjusting the ellipticity to give approximate pressure balance at the magnetopause between the interior magnetic field and the exterior solar wind (magnetosheath). Thus, for the first time since Coroniti and Kennel left the problem it became possible to calculate (albeit approximately) the strengthening of the current at the inner edge of the current sheet, as flux is added to the tail during the growth phase, self-consistently with the increase in solar wind pressure acting on an expanding tail boundary. Sotirelis showed that the procedure worked, but he had too little time before the workshop to make detail comparisons with observations. This work is continuing, however. During the final plenary session on substorms, Charlie Kennel described perhaps the most interesting example of intermodule coupling. As he expressed it, we now have three distinct and separate empirically based pictures of the substorm, all of them highly detailed and highly articulated. One picture refers to the ionosphere and includes the equatorward, premidnight auroral brightening at onset and the subsequent auroral bulge leading to a double oval, all under a polar bounding arc. The second picture refers to synchronous altitude and includes the inner edge of the current sheet and current disruption with attendant dipolarization, injection, and current wedge. The third picture refers to the tail and includes plasma sheet thinning followed by expansion (though an alternative to this has been suggested) and heating, with reconnection and plasmoid or flux rope creation and ejection also involved but with disputed timing. The problem now is to mesh and synchronize the three pictures to make a single video of the whole process. SECOND SESSION (which actually came first) Bob Lysak led the discussion of cross-scale and time-dependent coupling, which featured presentations by Jim Drake, Nancy Crooker, Bill Lotko, Mike Heinemann, and Joachim Birn. Cross-scale coupling can also be described as intramodule coupling. That is, within one module--the boundary layer, say--macroscale behavior might depend fundamentally on microscale properties, and vice versa. In macroscale formulations like MHD, microscale processes are parameterized. That is, they appear as transport coefficients multiplying some combination of macroscale variables, for example, the coefficient of viscosity multiplies the Laplacian of the velocity. But the microphysics of collisionless plasmas is far richer than the microphysics of collisional gases, on which the standard repertory of microscale parameterizations is based. One issue that needs to be addressed, therefore, is the degree to which we must reformulate microphysical parameterizations to understand macroscale magnetospheric behavior. Two schools of thought were expressed at the working group. From the audience Chuck Goodrich and Harlan Spence volunteered that much can be understood with standard formulations. (Though Goodrich warned that little can be understood without treating nonuniform plasmas.) Jim Drake argued the other side. He used the magnetopause as an example of a region where non-MHD processes dominate. Since the magnetopause acts as both barrier against and portal for solar wind entry into the magnetosphere, prima facie, magnetopause physics affects macroscopic magnetosphere properties. Drake demonstrated that magnetopause physics well illustrates the richness of microphysical processes available to collisionless plasmas. The simplest physics--reconnection based on electron inertia--gives too thin a layer compared to observations. Observations also show the presence of broadband turbulence. Drake described a nonlinear feedback scenario of turbulence generation and layer thickening based on the current convective instability. The instability, which is driven by the perpendicular gradient of the parallel current, generates vorticity. This leads to vortex interactions and merging of vortex tubes. This in turn leads to the transport of current, but also to local steepening, which drives the instability. According to Drake, Vortices in a plasma have a life of their own, independent of their source. Independent vortex dynamics interacting with a source of instability leads to a dynamic, turbulent current layer. The scenario was illustrated with results from a numerical simulation. In response to the question, Is microphysics governing the qualitative behavior of dayside reconnection? Drake answered, Yes. The problem of how the described microphysics could be parameterized for use in a macroscale code was, however, left as a future research topic. Since this report is growing too long, I must, with apologies to the other speakers, compress the remainder of the session into headlines. Lotko agreed with Drakes assessment that it is important to tailor the macroscopic transport terms to the operative microphysics. He illustrated the sensitivity of the Dartmouth viscous LLBL model to microphysics by showing model results in which the coefficient of viscosity had been given different forms. Birn made a similar point regarding the onset of tearing instability in the tail. He showed that a pressure anisotropy can stabilize the instability. This opens the door to time dependent reconnection scenarios that modulate tearing by modulating anisotropy rather than resistivity. Regarding time- dependent coupling, Crooker showed that the time-dependent dayside reconnection model of Lockwood and Cowley gives rise without further assumptions to the By-dependent mantle currents. Heinemann was the first to use the word daunting in describing the task we face in trying to couple time scales. He noted that the present magnetosphere-ionosphere coupling scheme is quasi-static and cannot capture such things as traveling ionospheric vortices. He described an alternative formulation that can handle faster phenomena. THIRD SESSION The final Working Group session dealt with GEM milestones and products. A full-up GGCM is years away. In the meantime, the GEM project will have generated valuable intermediate products of the same kind but not as complete--stand-alone subroutines of the GGCM, coupled subroutines, models, and data products that can be distributed to the community for use as research tools. Several candidates for such products have been suggested. Most are already existing codes or parts of codes that might be modified for incorporation in a GGCM. With approval of interested parties, these value-added versions could be labeled GEM products. The Working Group heard presentations on candidate GEM products from Dick Wolf, John Lyon, John Spreiter, and Steve Stahara. Wolf gave his assessment of the advantages and feasibility of using the magnetic field of an MHD global simulation in the RCM. (One viewgraph showed the 10 Re domain of the RCM inside the 200 Re domain of the Lyon-Fedder MHD simulation. "A humbling comparison," said Wolf.) He concluded that the self-consistency of the magnetic field--MHD's great virtue--constitutes a significant improvement over the field the RCM uses, and that it is worthwhile to work toward merging them. He estimates that two years would be needed to make an operational merged model. Such a merged model would constitute a GEM milestone. Lyon displayed results of a recent run of the Lyon-Fedder global MHD code showing the evolution of the convection electric field in the tail following a southward turning of the IMF. We recognized the standard substorm sequence: growth phase and expansive phase. But the comprehensiveness of the view and the details were remarkable. Throughout the growth phase the electric field at the cross-tail current sheet was negligible, corresponding to flux buildup in the tail. But suddenly "the dam burst." The electric field in the current sheet jumped to full convection value simultaneously along a strip running down the center of the tail from at least 10 Re to perhaps 60 Re. Besides this beautiful result, the presentation demonstrated two things: global MHD simulations constitute the closest analog to a truly global GGCM that exists; and they are powerful tools for validating and even revealing concepts. For example, this run strikingly illustrates the substorm scenario that is sometimes mentioned (e.g., in an invited AGU talk by Coroniti) but is still undeveloped in which the onset of the expansive phase simultaneously engages the near and midtail regions, including, presumably, all processes operating therein, whose relative timings are now in dispute. The lessons for WG 5 are twofold: because of their power, global MHD simulations are too computer intensive--especially one that could resolve boundary layers--to be packaged into portable, workstation codes for community use; and consequently their principal--and highly valuable--role in helping bring about such portable codes is in testing the outputs of prototypes and perhaps in providing data for look-up subroutines, as in the proposed RCM-MHD merger. Spreiter gave an update on the code he and Stahara are developing that computes the magnetic field generated by currents flowing on a doubly connected 3D surface shaped like the magnetopause with a cross-tail current sheet. The virtue of their code is that the magnetopause shape is adjusted to give pressure balance between the solar wind (magnetosheath) and the interior field at the boundary. This is a more accurate and comprehensive code than the one being developed by Sotirelis, but it is unclear which code will come on line first. The Spreiter-Stahara code has a singularity at the inner edge of the current sheet--precisely where there answer is most crucially needed. Spreiter says the singularity is in principle removable by a technique well known in aerodynamics. When this is done--or when Sotirelis is done- -current sheet instability theorists will finally have a plot of the globally self-consistent current sheet strength as a function of tail flux and solar wind pressure. This will allow them to follow the approach to instability during the growth phase build up of tail flux and to study the triggering of the instability by solar wind pressure changes. Stahara amazed everyone by displaying the remarkable ability of the Spreiter-Stahara magnetosheath code to duplicate highly variable changes of the magnetosheath magnetic field. Changes that casual inspection would label turbulence turn out to be rapid readjustments of magnetosheath flow in response to solar wind variations, especially of wind direction. The result calls for a re-evaluation of magnetosheath turbulence studies. By allowing investigators to subtract off the large, externally caused fluctuations, the code opens the door to a new and potentially more incisive effort to understand the nature and origin of magnetosheath turbulence. The demonstration also showed that the code is ready for community use. It would be nice to appropriate it as a GEM Milestone. In any case. it exemplifies the type of code that we would like to label GEM products and get into the hands of space physicists. The session continued with audience participation in identifying other potential GEM Milestones and WG 5 projects. GEM Milestones can come from any working group. For example, Working Group 2-- Particle Entry, Boundary Structure, and Transport--would achieve a GEM Milestone if it definitively resolved the issue of the nature of the dayside cusp: is it steady or blinking or some specific combination? Another WG 2 example would be the resolution of magnetic shear versus position factors in structuring dayside reconnection. Other candidates already mentioned include quantifying the global determinants of tail current strength (Sotirelis and Spreiter) and the RCM-MHD merger (Wolf, Lyon, and Fedder). We need candidate GEM Milestones from other working groups. This is an open solicitation to the GEM community for such candidates. By creating a list of achieved and actual milestones, the project is able to document its progress and prospects. This is more than public relations; though good public relations are important. A list of milestones gives the community a sense of progress and of working toward feasible, intermediate goals in support of the general goal. The following is a list of WG 5 project titles with the names of people who expressed interest in working on these projects. Neither the projects nor the names are meant to be exclusive, however. Suggestions for other projects and volunteers to work on them and those listed below are openly solicited. Coupled RCM-MHD code Wolf-Lyon-Fedder Coupled aerodynamic-MHD tail code Spreiter-Birn-Toffoletto Cross-tail current Spreiter-Sotirelis-Pulkkinen Magnetosheath residuals and turbulence Stahara-? B-normals to magnetopause in MHD models Lyon-Goodrich-Luhmann Polar cap-LLBL boundary condition Toffoletto-Lotko- Wolf Cusp-LLBL boundary condition Crooker-Lotko-Siscoe pV^5/3 maps Spence-Wolf MHD model-IMP magnetic field comparisons Luhmann-Kaymaz-Siscoe-Lyon-Fedder Roles of convection and stress balance in plasma sheet thinning Siscoe-Schulz RECOMMENDATIONS Besides the specific research-directed recommendations that take the forms of candidate GEM milestones and projects, Working Group 5 makes four programmatic recommendations: 1. There should be more working group interactions at GEM workshops. Because of its desire to identify GEM Milestones, WG 5 might be more sensitive to the need for such interactions than other working groups. Nonetheless, perhaps joint sessions should be built into future meetings. 2. The GEM project should pay attention to the issue of visibility. Many important results are coming out of the project, but they are not generally identified with GEM. Possible remedies for this lack of recognition include special AGU sessions highlighting GEM results, special GEM sections in JGR, and EOS articles reporting on the program. Of course GEM products, when they become available, will also help. At each Snowmass workshop, volunteers should be identified to implement such projects. 3. While retaining its NSF identification, GEM should continue to broaden its constituency. Even now probably less than half GEM workshops participants are funded out of NSF's GEM budget. GEM is inter-institutional and international. As Brian Fraser put at the workshop: "It's the project. It's well organized; it has goals; it reviews itself annually; and it operates in workshop mode." These virtues transcend institutional identification and funding opportunities. The key phrase is, "It's the project." 4. GEM is one of five projects with related goals. The others are ISTP, CEDAR, STEP, and Phillips Lab's Space Weather Forecasting Project. GEM should coordinate its activities with these other projects to bring about an appropriate division of labor. For example, ISTP is developing CRAY-level magnetospheric models. GEM should be developing complementary workstation-level models--which it is doing. It is important to consciously work together to coordinate these efforts. This is happening to a degree now through representatives from one project participating in the activities of another project. For example, Chuck Goodrich represented ISTP at this workshop. Working Group 5 is also coordinating with the Phillips Lab effort (as the next item illustrates). The recommendation is to continue inter- project coordination and to be ready to increase it in the interest of optimizing the use of resources. NEXT MEETINGS OF WORKING GROUP 5 There will be a meeting of Working group 5 at Phillips Lab on October 26 and 27. The point of the meeting is to define workstation level, community usable codes that can be packaged as GEM products over the next two years. Examples of such codes include the Rice Magnetosphere Specification Model (MSM), HAO's Assimilated Mapping of Ionospheric Electrodynamics (AMIE), the Spreiter- Stahara magnetosheath model, the Toffoletto-Hill electric field mapping algorithm, and codes based on David Stern's models (e.g., Sotirelis work) and Mike Schulz's models. The GEM spin on these already existing codes is to provide interface codes that allow the researcher to use them in mix-and-match fashion. Anyone interested in attending should contact George Siscoe at UCLATM::GSISCOE or Mike Heinemann at PH4000::HEINEMANN. Confirmed and tentative participants include Wolf, Toffoletto, Richmond, Schulz, Fedder, Lotko, and Hughes. A fuller list of attendees and an agenda will be published in a subsequent GEM Messenger. There will also be a meeting of Working group 5 at the Fall AGU meeting, as part of the general GEM WG meetings at this time. Details will be announced later. If you would like to be added to the Working Group 5 mailing list please send an email message to guan@igpp.ucla.edu.