Solar Proton Event Risk Modeling for Variable Duration Human Spaceflight

Transcription

1 Solar Proton Event Risk Modeling for Variable Duration Human Spaceflight D. Fry 1, Anne Adamczyk 2, S. Blattnig 2, M. Clowdsley 2, K. T. Lee 1, J. Norbury 2, N. Zapp 1 1 Space Radiation Analysis Group Lyndon B. Johnson Space Center Houston, TX 2 Langley Research Center Hampton, VA Introduction The purpose of this white paper is to outline a strategy for near-term development of modeling tools needed to mitigate Solar Proton Event (SPE) impact for future longduration human exploration missions. Such a strategy would facilitate greater use of numerous past and current satellite missions, possibly identify new directions for physics-based modeling efforts and foster increased collaboration between the research and operations community. Moreover, many of these same tools would be directly amenable to the needs of National technology infrastructure such as GPS telecommunications, commercial air flight and power grids. There is a pressing short-term need to develop tools that assess SPE risk on multiple time scales. Identified are the following four areas that should be pursued, with relatively short lead-time, in parallel to physics-based model development: 1. Probabilistic models of SPE occurrence covering mission durations of 1 week to 1 year useful to mission planners, flight control teams and design trades. 2. Probabilistic models of exceeding SPE energy-dependent flux covering mission durations of 1 week to 1 year useful as input to transport codes to assess expected exposure (design trades). 3. Probabilistic models of exceeding a threshold dose for a select time period during an event useful for modifying crew activities to adhere to dose limits and design trades. 4. Probabilistic models of SPE spectral characteristics capable of generating risk estimates over a 12-hour period prior to and during an EVA. Of the various sources of radiation exposure humans experience during exploration missions, SPEs represent the largest challenge operationally. SPEs, high intensity bursts of energetic ions (primarily protons) with energies exceeding 500 MeV can persist for weeks. There is no known method to predict during any given 24-hour time frame when an event will occur, it's duration, intensity and time profile, or it's spectral

2 characteristics. SPE mission impact is not known at any point other than in real-time, minute-by-minute as the event commences and evolves. As a result, risk to human crews is frequently addressed by defining an upper bound on SPE fluence using several of the recorded large events in the historical record - a worst case analysis that typically incurs additional parasitic mass for vehicle designs, constrains planning of mission profiles, reduces flexibility and increases complexity in both planning and realtime modification of crew activities. Although useful for short-duration mission, this may not be valid for long-duration missions during which crews will spend a substantial fraction of time away from mass shelters. For these missions, cumulative effects of the smaller events that occur with higher frequency may be non-negligible. Models that relax this constraint and incorporate the entire distribution of historical events as well as arbitrary mission duration and potential precursors to SPE onset are needed. Historically, events occur with a mix of rise and decay times. Impulsive events for instance frequently show a 6-order of magnitude increase in proton fluence in the first several hours after event onset. In general, time profiles show variability from relatively smooth exponential decay over the course of several days to a monotonic decay accompanied by a series of additional bursts anywhere from one to two orders of magnitude in fluence intensity. Physics-based modeling of solar dynamics has continuously progressed over the last 15 years, largely due to the success of numerous NASA Science Mission Directorate (SMD) missions. These same missions supply a great deal of daily situational awareness to space weather operations. In addition, our understanding of fundamental physical processes driving solar activity has drastically increased. From an operational perspective, these models are highly desired, but utility depends on transitioning from research to operations, which is complex and time-consuming. Apart from the efforts of the NASA Community Coordinated Modeling Center (CCMC), there has been limited dedicated community involvement in tool development. Tools to forecast SPE occurrence and characteristics remain elusive. Mitigation of exposure risk from SPEs for any human exploration mission occurs on the ground pre-mission through design trades to assess vehicle shielding, and in real-time during mission operations. It is important to understand what is needed both by mission planners and real-time operations to mitigate SPE impact. For example, mission planners require the ability to assess risk of SPE occurrence over the duration of the mission, when planning EVA activities, and manipulating daily crew activities. In realtime, operators need the ability to assess risk on time frames from 24 hours down to nominal 6 hour EVA duration, with advanced knowledge of event particle flux spectra and time evolution. In all, tools are needed to inform crews and flight teams over a large dynamic time range of potential high-risk conditions. Also, architecture and system designers need tools to perform trade studies. Given near-term development timelines, augmentation of past risk mitigation strategies is of utility to both mission design and operations. Tools are needed to allow for more comprehensive design trades to be conducted in radiation shielding and to help mission

3 operators make more informed decisions in assessing real-time impact to crews. Physics-based models are clearly the preference, but other short-term methods need to be investigated in the interim. Past work and gaps are first identified below and then a brief outline of how to augment these efforts in the future efforts is given. Past Work and Gaps Several gaps exist in pursuing a direct translation from hardware assessment models to those applied to human spaceflight. There is limited spectral information contained in risk estimates developed with any of the existing models [1-7] covering the range (>1 MeV to > 500 MeV) needed for human space flight. A second issue is the statistical quality of the data. Detector design has evolved over the last several decades resulting in increased energy resolution capability and dynamic range, but there are gaps. Piecing together proton fluence spectra from different satellites that have different orbital tracks is time consuming, and has potential to introduce uncertainties that skew model results. Observations over different time spans have typically been conducted over different regions of the proton spectral energy range needed for quantitative dose calculations. There has not been any study to date that systematically compares fluence spectra from the early 1960s to the present day that ascertains whether or not data gaps or assumption made to fill gaps induces negligible uncertainty in calculated crew dose. There is also variance in which time span is used to calculate integral fluence values. Of the known models, each utilized annual proton fluence, daily fluence and/or event-integrated fluence. In general, for human exploration missions models need to be constructed from integral fluence values where the limits of integration δt are limited to δt T the mission duration. Daily fluence should be sufficient for any human mission durations considered, but a full comparison of available models with the same integration range are lacking. Dedicated space-based observation platforms for measuring SPE spectra out to >500 MeV are lacking. There has been a fair amount of high quality efforts aimed at assessing risk of SPEs for robotic missions as well as design trades for human rated vehicles. Unfortunately, the majority of risk models are not directly amenable as tools for human spaceflight and those that have been attempted are limiting when considering long-duration missions. There are multiple factors of influence. Much more shielding is needed for human rated vehicles, which make large safety factors less of an option, and thus more detailed and accurate modeling is needed. Mission durations for robotic and human space flight missions are not commensurate robotic missions are typically more than a year in duration whereas long-duration human missions are being considered at most a year in length. Considering the growth in observational platforms over the last three decades

4 and inherent variance in spectral range of proton fluence measured, the dataset of historical events is sparse and has numerous gaps for which spectral measurements are not available. Past missions have been conducted with the majority of crew time spent inside spacecraft, where parasitic shielding mass is designed to standards set by considering an upper bound to proton fluence determined by several historically intense events. Long-duration space flight that includes landing on other planetary/near Earth Object (NEO) surfaces will likely include more time outside of well-shielded environments, and thus tools are needed that address SPE risk over time frames of hours to total mission duration. In general, risk mitigation strategies can be grouped into two categories: (1) Physicsbased models aimed at forecasting SPEs, and; (2) Historical data-driven statistical models. Physics-based models are capable of producing the most compete picture of solar dynamics. However, they are highly complex, typically have long lead times to develop, require extensive data sets for verification and validation, and as a result are costly. Conversely, data-driven models, built upon statistical analysis of what has occurred during past events, can be developed in ways analogous to other risk mitigation and forecasting strategies. This approach is less time consuming, has a tried-and-tested history of use in conventional terrestrial weather forecasting, but is highly dependent on the size and quality of historical observations. A clear advantage however is that the detailed underlying physical processes generating SPEs and propagation throughout the heliosphere are not a priori required. These methods could provide a way to develop valuable tools at relatively low cost, building on observations already made, making use of an extensive history of use in other fields such as economics, engineering design, real estate market prediction and epidemiology, and possibly identifying new direction for development of physics-based modeling. Past statistical approaches have several commonalities. All models make use of historical event data casting probability in terms of distributions of proton fluence. The models of Feynman et al. [1-2] consolidated proton flux measurements made between 1963 and 1991 (solar cycles 19-21, IMP and OGO spacecraft) and assess distributions of daily fluence values for integral proton energies from >1 MeV to >60 MeV. Results showed that over a cumulative probability range from 25% - 99% the distribution of fluence at fixed integral energy is nearly log-normal, facilitating sampling from a known distribution. Other studies have found similar results for integral fluences out to >100 MeV [3-6]. Risk estimates for exceeding a preset fluence for mission durations 1 year are then determined by assuming individual events are Poisson-distributed in time over a solar active period centered on the peak in annual sunspot number, and sampling the log-normal distribution of fluence. Modifications to this approach have also been conducted [7] but with incomplete spectral information. An alternate methodology is to use Maximum Entropy Theory (MET). Particularly suited for dealing with incomplete data sets and analysis of rare events, Xapsos et al. applied MET to produce a model of probability of exceeding a given fluence for missions 1

5 year in duration at a select confidence level [3-6]. Considering that roughly 100 out of 6386 ( ) days resulted in SPE onset (total of ~4.2% of days at elevated proton fluence levels), it is feasible that MET might work quite well for projections of exceeding a pre-determined cumulative fluence threshold. How this can be mapped into risk assessment for crew dose is unclear at the present time. Design trades are critical for assessing the level of shielding any vehicle will provide, feeding input in real-time during the design cycle. Due to a lack of sufficient event data and available tools, the vast majority of past work has dealt with SPE impact by assuming an upper bound set by either the August 1972 (King), the November 1960, the February 1956, or the September 1989 events, four characteristically large events recorded during the modern space flight era [10-18]. In addition, combinations of these events have been used to represent an upper bound in particle fluence covering the largest spectral proton energy range available [15]. Neglecting the remaining events in the historical records however explicitly constrains the trade space to worst-case scenarios, which typically result in extra parasitic mass. Additional efforts have been made that address crew risk from models that incorporate biological effects [19-21]. The relatively large uncertainty in biological endpoint, limited data sets covering particle type and energy similar to the interplanetary environment, and high variability of genetic factors between crewmembers present a large challenge towards effective utilization of these models. In general assessment has been limited to integral proton energy of >30 MeV. This is due to a tradeoff between expanding the number of historical events observed to decrease statistical error and the cost of working with gaps in the proton energy range measured over the last 60 years. Attempts have been made to circumvent the issue of limited observations by assuming a particular spectral form and fitting observations to extend to higher proton energy [22]. Next Steps It is expected that exploration of solar dynamics will continue with new and advanced missions being launched in upcoming years. However, other avenues are potentially available to bring to fruition in the near term tools of utility to both operations and vehicle design by incorporating analysis techniques similar to terrestrial weather forecasting and risk mitigation in other disciplines. Risk modeling is not new science. There are numerous risk modeling strategies available that should be carefully looked at, with the potential to produce within the next several years tools that will assist operators and mission designers in assessing SPE risk [23-30]. Physics-based modeling is currently a focus of heliophysics. It is proposed here that efforts also be dedicated to probabilistic risk-based modeling that address multivariate correlations and trending to identify characteristics such as event size, evolution, and occurrence that have direct overlap with operational timelines of mission planning, mission duration, EVAs, daily crew activities and design

6 trades. To accomplish this task there needs to be a concerted effort from all agency stakeholders in SPE mitigation (SMD/SOMD/ESMD) to: 1.) Ensure dedicated measurement of SPE fluence spectra out to >500 MeV (preferably 1 GeV), at a maximum cadence of 5-min, by leveraging future robotic missions. 2.) Glean as much information as possible from the historical event record to support development of near-term tools characterizing event probability of occurrence, size and spectral characteristics over proton energies from >1 MeV to >500 MeV, time evolution and duration. 3.) Develop near-term SPE forecasting tools based entirely on SPE statistics, and feedback to the long-term development of physics-based forecasting models 4.) Maximally leverage risk model development in other disciplines. Risk assessment and forecasting of extreme/rare events is a specific area of research in terrestrial weather, financial markets and geology (earthquakes) that have direct application to SPEs. 5.) Develop dedicated funding to support the transition from research to operations fostering collaboration between the research and operations communities extending beyond agency efforts. There are essentially two mutually exclusive paths to deal with any risk limit the risk, and mitigate a specific endpoint of the risk. It is widely accepted that the largest source of uncertainty in assessing risk of exposure is uncertainty in biological endpoint [19-21]. Shielding, mission planning, and real-time operational contingencies are used to limit exposure risk. Efforts to minimize biological uncertainty and develop better tools should be worked in parallel. This will allow agency science to directly feed operations with tools based on the current state of knowledge and build strategies for transition from research to operations when more mature physics-based models are available. Current space weather operations are centered on minimizing crew exposure levels, i.e. ALARA, or As Low As Reasonably Achievable. This will likely remain an overarching principle for future long-duration missions. Operators need quick, relatively short outlook tools to better inform both the crew and mission planners of impending SPE impact. Architecture design trades need statistical tools that allow for a probabilistic sampling over larger range of proton energy and fluence. Future efforts aimed at long-term development of physics-based models and near-term development of probabilistic tools should be a key focus of the next decade of heliophysics. Ensuring a continual measurement of needed spectral ranges of proton fluence and fostering a robust collaboration between research and operations will support both long-term and near-term goals of the agency to protect crews on longduration missions.

7 Bibliography 1. J. Feynman, T. P. Armstrong, L. Dao-Gibner and S. Silverman, New Interplanetary Proton Fluence Model, Journal of Spacecraft, 27, 403 (1990). 2. J. Feynman, G. Spitale and J. Wang, Interplanetary Proton Fluence Model: JPL 1991, Journal. of Geophysical Research: Space Physics, 98, (1993). 3. M. A. Xapsos, G. P. Summers, P. Shapiro and E. A. Burke, New Techniques for Predicting Solar Proton Fluences for Radiation Effects Applications, IEEE Transactions on Nuclear Science, 43 (1996). 4. M. A. Xapsos, G. P. Summers and E. A. Burke, Extreme Value Analysis of Solar Energetic Proton Peak Fluxes, Solar Physics, 183, (1998). 5. M. A. Xapsos, J. L. Barth, E. G. Stassinopoulos, E. A. Burke and G. B. Gee, Space Environmental Effects: Model for Emission of Solar Protons (ESP) - Cummulative and Worst-Case Event Fluences, NASA/TP (1999). 6. M. A. Xapsos, J. L. Barth, E. G. Stassinopoulos, S. R. Messenger, R. J. Walters, G. P. Summers and E. A. Burke, IEEE Transactions on Nuclear Science, 27 (2000). 7. R. A. Nymmik, Probabilistic Model for Fluences and Peak Fluxes of Solar Energetic Particles, Radiation Measurements, 30, (1999). 8. I. Jun, R. T. Swimm, A. Ruzmaikin, J. Feynman, A. J. Tylka and W. F. Dietrich, Statistics of Solar Energetic Particle Events: Fluences, Durations, and Time Intervals, Advances In Space Research, 40, (2007). 9. M. Y. Kim, J. W. Wilson, F. A. Cucinotta, L. C. Simonsen, W. Atwell, F. F. Badavi and J. Miller, Contribution of High Charge and Energy (HZE) Ions During Solar- Particle Event of September 29, 1989, NASA/TP (1999). 10. NASA s Exploration Systems Architecture Study Final Report, NASA/TM (2005). 11. T. T. Pham and M. S. El-Genk, Dose Estimates in a Lunar Shelter with Regolith Shielding, Acta Astronauta, 64, (2009). 12. L. C. Simonsen, J. W. Wilson, M. H. Kim, and F. A. Cucinotta, Radiation Exposure for Human Mars Exploration, Health Physics, 79, (2000). 13. M. S. Clowdsley, B. M. Anderson, N. Luetke, and J. W. Wilson, Calculation of Radiation Protection Quantities and Analysis of Astronaut Orientation

8 Dependence, in Proceedings of AIAA Space 2006 Conference, San Jose, California, September 19-21, 2006, AIAA M. S. Clowdsley, J. E. Nealy, J. W. Wilson, B. M. Anderson, M. S. Anderson, and S. A. Krizan, Radiation Protection for Lunar Mission Scenarios, in Proceeding of AIAA Space 2005 Conference, Long Beach, California, August 30 September 1, Shielding Strategies for Human Space Exploration, Edited by J. W. Wilson, J. Miller, A. Konradi, and F. A. Cucinotta, NASA Conference Publication 3360 (1997). 16. L. C. Simonsen, J. E. Nealy, L. W. Townsend and J. W. Wilson, Radiation Exposure for Manned Mars Surface Missions, NASA/TP-2979 (1990). 17. L. C. Simonsen and J. E. Nealy, Radiation Protection for Human Missions to the Moon and Mars, NASA/TP-3019 (1991). 18. W. D. Hypes, A. J. Butterfield, C. B. King, G. D. Qualls, W. T. Davis, M. J. Gould, J. E. Nealy and L. C. Simonsen, Concepts for Manned Lunar Habitats, NASA/TP (1991). 19. J. W. Wilson, J. E. Nealy, W. Schimmerling, F. A. Cucinotta and J. S. Wood, Effects of Radiobiological Uncertainty on Vehicle and Habitat Shield Design for Missions to the Moon and Mars, NASA/TP-3312 (1993). 20. F. Cucinotta, W. Schimmerling, J. Wilson, L. Peterson and J. Dicello, Space Radiation Cancer Risk Projections for Exploration Missions: Uncertainty Reduction and Mitigation, NASA/TP (2002). 21. M. Y. Kim, G. De Angelis and F. A. Cucinotta, Probabilistic Assessment of Radiation Risk for Astronauts in Space Missions, Acta Astronautica, in press (2010). 22. M. Y. Kim, M. J. Hayat, A. H. Feiveson and F. A. Cucinotta, Prediction of Frequency and Exposure Level of Solar Particle Events, Health Physics (2009). 23. D. Vose, Risk Analysis: A Quantitative Guide, 3 rd Edition, John Wiley and Sons, Ltd. (2008). 24. C. A. Doswell III, R. Davies-Jones, and D. L. Keller, On Summary Measures of Skill in Rare Event Forecasting Based on Contingency Tables, Weather and Forecasting, 5, (1990). 25. E. J. Gumbel, Statistics of Extremes, Dover Publications (2004).

9 26. G. King, L. Zeng, Logistic Regression in Rare Events Data, Society for Political Methodology (2001). 27. G. King and L. Zeng, Explaining Rare Events in International Relations, International Organization, 55, (2001). 28. R. W. Katz, M. B. Parlange and P. Naveau, Statistics of Extremes in Hydrology, Advances in Water Resources, 25, (2002). 29. T. Bedford and R. M. Cooke, Probabilistic Risk Analysis: Foundations and Methods, Cambridge University Press (2001). 30. G. R. Dargahi-Noubary, The Use of Modern Statistical Theories in the Assessment of Earthquake Hazard, with Application to Quiet Regions of Eastern North America, Soil Dynamics and Earthquake Engineering, 22, (2002).