Corresponding author: C. J. Joyce, Physics Department, Space Science Center, University of New Hampshire, 39 College Road, Durham, NH 03824-3525, USA. (firstname.lastname@example.org)
 PREDICCS (Predictions of Radiation from Release, EMMREM, and Data Incorporating the CRaTER, COSTEP and other SEP measurements, prediccs.sr.unh.edu) is an online system designed to provide a near real-time characterization of the radiation environment of the inner heliosphere. PREDICCS utilizes data from various satellites in conjunction with numerical models such as the Earth-Moon-Mars Radiation Environment Module (EMMREM) to produce dose rate and particle flux data at the Earth, Moon and Mars. The Cosmic Ray Telescope for the Effects of Radiation (CRaTER) instrument launched aboard the Lunar Reconnaissance Orbiter (LRO) spacecraft in 2009 and designed to measure energetic particle radiation, offers an opportunity to test the capability of PREDICCS to accurately describe the lunar radiation environment. We provide comparisons between dose rates produced by PREDICCS with those measured by CRaTER during three major solar energetic particle (SEP) events that occurred in 2012. In addition, using EMMREM data products together with our archive of measured CRaTER dose rates, we compute the modulation potential at the Moon throughout the LRO mission and, using this, compute the background GCR dose rate during each event. We demonstrate reasonable agreement between PREDICCS and CRaTER dose rates and come to the conclusion that PREDICCS provides credible characterization of the lunar radiation environment. This study represents the first multi-event validation, via in situ measurement, of radiation models such as EMMREM, which should prove to be valuable in future efforts in risk assessment and in the study of radiation in the inner heliosphere.
 Radiation in the form of galactic cosmic rays (GCRs) and solar energetic particles (SEPs) poses a serious threat to future manned missions outside of low Earth orbit [Cucinotta et al., 2010]. Characterization of the radiation environment in the inner heliosphere is an essential step toward future manned missions. PREDICCS (Predictions of Radiation from Release, EMMREM, and Data Incorporating the CRaTER, COSTEP and other SEP measurements, http://prediccs.sr.unh.edu) is an online system that was proposed as a resource for researchers to predict near real-time radiation exposure at the Earth, Moon and Mars, as well as interplanetary space. PREDICCS utilizes satellite measurements along with Earth-Moon-Mars Radiation Environment Module (EMMREM), a numerical model that solves for the propagation of SEPs through the heliosphere and the subsequent transport of energetic particles through various levels of shielding, to produce dose rate and particle flux data in the Earth-Moon-Mars system [Schwadron et al., 2010]. PREDICCS has the potential to be an invaluable tool for its use as a means of forecasting events and providing an increased capacity for risk assessment for future missions.
 In order for PREDICCS to be used as a credible source for describing the radiation environment of the inner heliosphere, it must be demonstrated via in situ measurements that its predictions are sufficiently accurate. Cosmic Ray Telescope for the Effects of Radiation (CRaTER), an instrument on the Lunar Reconnaissance Orbiter (LRO) spacecraft designed to measure the energy deposited by energetic particles, offers an excellent opportunity to test PREDICCS. Using CRaTER data, we compute dose rates which are compared to those produced by PREDICCS, giving us the ability to gauge the effectiveness of the radiation model and determine its shortcomings.
 In this paper, we offer comparisons between CRaTER and PREDICCS dose rates during three major solar events occurring in January, March, and May of 2012. We describe the agreement and point out the discrepancies between CRaTER and PREDICCS dose rates. In addition, we calculate accumulated doses for both PREDICCS and CRaTER for each event and compare them to each other as well as to NASA Permissible Exposure Limits (PELs) [Cucinotta et al., 2010] in order to demonstrate the threat of exposure each event poses. For further comparison, we include dose rates and accumulated doses for each event as measured by the microdosimeter within the CRaTER instrument. Finally, in order to investigate possible causes for discrepancies between CRaTER and PREDICCS, we use ACE flux data as well as CRaTER Lineal Energy Transfer (LET) spectra in order to determine if compositional differences may have an effect on the agreement.
 The dose rates shown here are computed by EMMREM using a process in which only protons contribute to the dose. This limitation allows for reasonably accurate predictions of dose rates during SEP events, which are typically proton-dominated, but is more of a problem for predicting GCR radiation to which protons usually contribute only about half of the total dose [Schwadron et al., 2012]. In order to account for this, we propose a method for modeling GCR dose rates using EMMREM together with CRaTER data. To accomplish this, we use tables containing data for the modulation potential as a function of GCR dose rate, atmospheric density, and shielding thicknesses, which are described in Townsend et al. . Utilizing these tables with CRaTER-measured dose rates allows us to compute a time-dependent modulation potential at the Moon and, using our archive of measured CRaTER dose rates, we show the evolution of the modulation potential at the Moon throughout the duration of the LRO mission. Because the modulation can be scaled out to different locations in the heliosphere [Schwadron et al., 2010], we can reverse this process to compute time-dependent GCR dose rates at different points of interest, for example, Mars. To demonstrate this concept, we use the average modulation potential around each SEP event to compute the GCR dose rate. This sort of methodology may be utilized in the future by the PREDICCS system to provide a more robust characterization of the radiation environment that better incorporates GCR effects.
2 EMMREM/CRaTER Overview
 The PREDICCS dose rates shown here are computed using input data from GOES that has been filtered for use by the EMMREM model to produce dose rate data at the Moon. The EMMREM model is described in detail by Schwadron et al.  and consists of two essential components. The first is the Energetic Particle Radiation Environment Module (EPREM), which uses energy spectrum, composition and angular distribution data input from various spacecraft (ACE, SOHO, GOES, Mars Odyssey, Ulysses) as well as simulations and computes acceleration and propagation of solar energetic particles through the heliosphere. This process involves propagating a series of nodes outward from the Sun along magnetic field lines (taken to be a nominal Parker spiral) at the solar wind speed and solving for particle transport, adiabatic focusing, adiabatic cooling, convection, pitch angle scattering, and stochastic acceleration at each step via the methodology of Kóta et al.  [Schwadron et al., 2010]. EPREM is capable of solving for the transport of particles with energies ranging from 500 eV up to 1 GeV [Schwadron et al., 2010]. The second module utilized by EMMREM is the Baryonic Transport Module (BRYNTRN), which solves for the transport of protons and their secondary particles through shielding of differing materials and thicknesses. EMMREM utilizes these modules together to characterize the radiation environment of the Earth-Moon-Mars system, producing energy spectra and dose rate data at various locations in the inner heliosphere. EMMREM is able to use BRYNTRN to produce various data of biological significance, producing dose rates for different proxies of body organs (skin, eye, organs, etc.) through shielding of varying thicknesses (corresponding to spacesuit, spacecraft, thick shielding, etc.).
 CRaTER is an instrument on the LRO spacecraft whose goal is to characterize the lunar radiation environment. The instrument contains three pairs of thin and thick silicon detectors, which measure the energy deposited by energetic particles [Spence et al., 2010]. We compute hourly dose rates for each detector by computing the total energy measured in an hour by each detector, dividing by the detector mass to get the total dose accumulated, and dividing by the number of valid seconds from which data is used in the hour to get an average dose rate. The validity of each second is determined using secondary CRaTER data which flags any second of data that takes place when the spacecraft is in a mode unsuitable for measurement (e.g., performing a calibration).
 We have created an archive of CRaTER dose rates spanning the entirety of the LRO mission. These data contain averaged hourly dose rates for each detector, as well as a combined dose rate for the zenith-facing D1-D2 thin-thick pair, which is the primary CRaTER dose rate we use in this analysis. This dose rate is calculated by limiting the energy range of the two detectors to ensure that events do not contribute twice if the energy deposit of the particle is measured by both detectors. This calculation is described in detail by Schwadron et al. . This dose rate is then multiplied by three factors so that it may be used for comparison with the PREDICCS data products. The first factor is the conversion from the silicon material that comprises the CRaTER detectors to water, which is the material for which PREDICCS computes dose rates. This factor is found to be a constant 1.33 [Schwadron et al., 2012]. The second factor adjusts measurements made at altitude, r, to the surface of the Moon, where half of the sky is blocked by the Moon [Schwadron et al., 2012]. The formula for this factor is given by
where Rm is the radius of the Moon and r is the altitude of the spacecraft at the time of measurement.
 The third factor accounts for the fact that during major solar events, the rate of energetic particle events may be too great for the CRaTER detectors to measure them all. A separate mechanism in CRaTER counts the total number of events that occur each second prior to their analysis, including those that are not measured by the detectors. To compensate for the unmeasured events, we multiply the hourly dose rate by the ratio of the total number of events to the number of events actually measured. During times of normal activity, this factor is always 1. The highest values this factor takes on during the three events described here are ~ 22, ~ 24, and ~ 3 for the January, March, and May events, respectively.
 The CRaTER instrument also contains a microdosimeter, which measures dose rates in silicon behind a higher level of shielding than do the CRaTER detectors. The properties of the CRaTER microdosimeter are discussed in detail by Mazur et al. . As a means of comparison and evaluation of the PREDICCS dose rates for higher levels of shielding, we include dose rate data from the CRaTER microdosimeter in our plots of solar events.
3 Computation of the Modulation Potential at the Moon During the LRO Mission
 In this section, we demonstrate our method of calculating the modulation potential at the Moon using CRaTER dose rates in conjunction with an EMMREM-generated table containing modulation potential data. Using the techniques described here, we generate a plot showing the evolution of the modulation potential at the Moon throughout the duration of the LRO mission. Later we utilize this computed modulation potential to calculate the background GCR dose rates during each of the three solar events described here as a way of showing how these data and techniques can be applied. Because the modulation potential can be scaled out to different locations in the heliosphere [Schwadron et al., 2010], similar methods can be used and potentially incorporated into PREDICCS to produce more accurate GCR dose rates at different points of interest such as Mars.
 The modulation potential is the approximate amount of energy lost by a galactic cosmic ray during its transit through the heliosphere. We should therefore be able to determine a functional relationship between the GCR dose rate and the modulation potential that we can use to calculate modulation potentials from measured GCR dose rates. We compute the modulation potential at the Moon using a table generated by EMMREM, which is described by Townsend et al.  and is based on the Badhwar and O'Neill GCR model described in O'Neill . The table contains data for the modulation potential as a function of GCR dose rate, atmospheric density, and shielding thickness. Using an atmospheric density value of 0 g/cm2 (corresponding to the lunar atmosphere) and shielding of 0.3 g/cm2 (closest to the 0.22 g/cm2 shielding of the CRaTER detectors), we plot modulation potential against dose rate (Figure 1). By fitting this plot with a polynomial, we develop an expression for the modulation potential as a function of dose rate. If we take a CRaTER dose rate measured during quiet time (i.e., not during a SEP event) to be a reasonable GCR rate, we can use it with this expression to compute the modulation potential. The dose rate data in Figure 1 is computed over 4π steradian exposure, and thus we multiply our CRaTER dose rates by two before entering them into the calculation to convert from approximate 2π exposure to 4π exposure.
 We then purge our CRaTER dose rate data of SEP events using a method which removes any hour of data which possesses count rates that are above a certain threshold and average the dose rates over one day increments to smooth out the data. Thus, using all of our CRaTER dose rate data with SEP events excluded in this way with the expression in Figure 1, we are able to show the evolution of the modulation potential throughout the duration of the LRO mission. Figure 2 shows the GCR dose rate and the modulation potential over this time period. We see in the plot that the modulation potential comes to a local minimum at the start of 2010 before steadily increasing in expected fashion as we enter solar maximum, a time of increased magnetic field strengths and coronal mass ejections which increase the energy lost by GCRs and thus the modulation potential [Schwadron et al., 2008, 2010]. The sunspot number is generally correlated with solar activity and, thus, to provide the reader with some context within the solar cycle, Figure 2 also shows the mean daily sunspot number averaged monthly over this time period as measured by the Solar Influences Data Analysis Center (SIDC) at the Royal Observatory of Belgium (available at: http://www.ngdc.noaa.gov/nndc/struts/results?t=102827&s=5&d=8,430,9). We see the expected correlation between increases in sunspot number and modulation potential as the Sun transitions from minimum to maximum. It is interesting to note that increases in sunspot number seem to predate similar changes in the modulation by a month or two.
4 PREDICCS/CRaTER Comparison During SEP Events
 In this section we compare the dose rates measured by CRaTER to those predicted by PREDICCS for three major SEP events in 2012 that occurred in January, March, and May. For our purposes, we define an event as an extended period containing dose rates elevated significantly above the normal background dose rate associated with GCRs. For this comparison, we use the previously discussed combined D1-D2 CRaTER dose rate and the PREDICCS dose rates for 1.0 g/cm2 H2O, which is a proxy for skin and eye dose, shielded by 0.3 g/cm2 aluminum, which is a proxy for spacesuit. This configuration is the closest analog to the CRaTER detectors we have available and should have the best agreement with the CRaTER dose rates. For the sake of comparison, we also show PREDICCS dose rates for three other levels of shielding (1.0 g/cm2 (heavy spacesuit), 5.0 g/cm2 (spacecraft), and 10.0 g/cm2 (heavy shielding)). Also shown is the dose rate for each event as measured by the microdosimeter within CRaTER, which is shielded on one side by approximately 0.89 g/cm2 and 2.28 g/cm2 on the other [Mazur et al., 2011]. We can therefore reasonably expect the microdosimeter dose rates to fall somewhere between the PREDICCS dose rates for shielding of 1.0 g/cm2 and 5.0 g/cm2.
 In addition, we compute the total dose accumulated during each event using both the CRaTER and PREDICCS dose rates, as well as the microdosimeter. These accumulated doses are calculated by multiplying the dose rate at each data point by the time step between it and the previous data point. In the plots, we include the NASA 30 day dose limits from the PELs for skin (150 cGy) and eye/lens (100 cGy) [Cucinotta et al., 2010] in order to provide a sense of the threat each event could pose to astronauts. The higher levels of shielding are once again shown to demonstrate how they can reduce the threat of radiation. The periods shown contain three data gaps, one for CRaTER and two for PREDICCS. To account for missing data in the calculation of accumulated doses, we compute data values during the data gap using an exponential function that connects the data values on either side of the gap, which represents a straight line in the log scale dose rate plots. In each case we point out the data gap in the figure captions.
4.1 January SEP Event
 The 23 January SEP event lasted approximately 10 days, spanning 23 January to 1 February and reaching dose rates of up to ~150 cGy/day. Figure 3 shows the dose rates measured by CRaTER and the dose rate predicted by PREDICCS during this event. We see excellent agreement between CRaTER and PREDICCS at the two peaks of the event; however, PREDICCS seems to underestimate the dose rate immediately after the peaks, dropping off much more rapidly than the measured dose rate. The PREDICCS dose rates for higher levels of shielding are also shown in the plot, and we see that the CRaTER data are, as expected, closest to the dose rate with 0.3 g/cm2 shielding. The microdosimeter dose rate reaches dose rates as high as ~2 cGy/day, falling between the PREDICCS dose rates for 1.0 and 5.0 g/cm2 shielding as we would expect.
 Using Figure 2, we estimate a modulation potential during this event of approximately 725 MV. Then by interchanging the axis of Figure 1 and fitting the data to get an expression for dose rate as a function of modulation potential, we compute the background GCR rate during the event (red dotted line) using the estimated modulation potential for the event. This computed GCR rate is consistent with the background CRaTER and microdosimeter dose rates as it should be, and it exceeds the PREDICCS dose rates before and after the event as expected. The agreement with CRaTER is trivial, since the modulation potential is derived from CRaTER GCR rates, but it shows how it could be used to improve PREDICCS GCR dose rates, which would be useful at other locations in the heliosphere where there is no spacecraft to make measurements.
 The accumulated doses for CRaTER and PREDICCS for the 23 January event are shown in Figure 4. We see that for this event, CRaTER measures a total dose of 185 cGy, which is comparable to the Halloween Storms of 2003, for which EMMREM produced a total accumulated dose of almost 300 cGy for the same 1.0 g/cm2 H2O behind 0.3 g/cm2 shielding [Schwadron et al., 2010]. For this event PREDICCS somewhat underestimates the total dose accumulated, predicting a dose of 118 cGy. We see that the CRaTER accumulated dose exceed the 30 day limits for both lens and skin dose. Also shown in the plot are accumulated doses calculated by PREDICCS for the higher levels of shielding. Since the CRaTER dose is very close to the EMMREM dose for 0.3 g/cm2 shielding, we can assume that the PREDICCS doses accumulated for the higher levels of shielding are reasonably accurate, and this shows the degree to which increased shielding can reduce radiation exposure. The microdosimeter measures a total dose of 3.58 cGy for the event, which falls between the PREDICCS dose rates for the appropriate levels of shielding.
4.2 March SEP Event
 The 7 March SEP event lasted approximately 10 days, spanning from 7–17 March and reaching dose rates of up to ~100 cGy/day. Figure 5 shows the dose rates measured by CRaTER and predicted by PREDICCS. During this event, we see reasonable agreement between CRaTER and PREDICCS, particularly at the first peak. Following the first peak, we do not see as steep a drop-off for PREDICCS as that in the previous event, with PREDICCS not dropping significantly below CRaTER until roughly 1 day before the second peak. The second peak is underestimated somewhat by PREDICCS, and the steep drop off following it is similar to what we see in the first event. Once again we show PREDICCS dose rates for higher levels of shielding as well as the background GCR dose rate computed using an approximate modulation potential of 775 MV. The microdosimeter shows dose rates closer to the 5.0 g/cm2 than to the 1.0 g/cm2, shielded PREDICCS dose rates than shielding during the two peaks of the event and reaches dose rates as high as 7 cGy/day.
 The accumulated doses for CRaTER and PREDICCS are shown in Figure 6. For this event, CRaTER and PREDICCS are in much closer agreement, with both doses just about reaching the 30 day skin limit of 150 cGy. Once again we show the PREDICCS products for higher levels of shielding, which should be very accurate given the close agreement between the analogous dose rates. We see thatfor this event, higher levels of shielding do not reduce the dose rate quite as much as before, being only as much as a factor of ~ 25 lower than the highest rate. The microdosimeter measures a total dose of 11.6 cGy for the event, falling between the PREDICCS 1.0 g/cm2 and 5.0 g/cm2 doses.
4.3 May SEP Event
 The 17 May SEP event lasted approximately 4 days, spanning 17–21 May and reaching dose rates as high as ~8 cGy/day. Figure 7 shows the dose rates measured by CRaTER and predicted by PREDICCS. At the very peak of the event, we see very good agreement between CRaTER and PREDICCS, although we soon see the familiar drop off as PREDICCS begins to underestimate the dose rate following the peak. The microdosimeter reaches a maximum dose rate of ~0.9 cGy/day. This event differs from the previous two events in that, it is shorter and contains only one peak with dose rates comparable to the lesser second peaks of the first two events.
 Figure 8 shows the total dose accumulation for the 17 May event, which, being the shortest event and having only one, smaller peak, produces much smaller total doses. For this event, CRaTER accumulates a total dose of 3.61 cGy, more than a factor of 25 lower than the 30 day lens limit, while the analogous PREDICCS prediction underestimates the dose at 2.58 cGy. The total dose for this event is much closer to the 8 June 2011 event, which accumulated total doses of ~2 cGy [Schwadron et al., 2012], than to the much larger Halloween event of 2003. The microdosimeter records a total dose of 0.530 cGy for this event, falling slightly below the PREDICCS dose 5.0 g/cm2.
5 Discussion of CRaTER/PREDICCS Discrepancies During Events
 For each event, we observe a period, following the period of peak dose rates, in which PREDICCS substantially underestimates the dose rate measured by CRaTER. In this section, we seek an explanation for these differences. The January event in particular contains the longest period of disagreement following its first peak (approximately spanning DOY 25–27) that we see in any of the events, and as such, we study it to learn about possible causes for the discrepancy between CRaTER and PREDICCS.
 The PREDICCS dose rates shown here are computed using only dose contributions of protons, so we would therefore expect that such periods of disagreement might be caused by an increase in the influence of heavy ions, which are not accounted for in the PREDICCS model. Figure 9 shows the CRaTER and PREDICCS dose rates plotted alongside the ratio of CNO to H fluxes as measured by the SIS instrument aboard the ACE spacecraft. If heavy ions were the cause of the observed discrepancies, we would expect that during periods of discrepancy their flux would increase relative to that of protons. However, we see that this ratio in fact remains relatively constant throughout the event, and we thus conclude that heavy ions are not likely to be the cause of disagreement.
 In order to further investigate the effects of compositional differences on the observed discrepancy, we compute Lineal Energy Transfer (LET) spectra according to the methods described by Case et al. . Figure 10 shows the ratio of the LET spectra taken over the period of the discrepancy to the spectrum of the event as a whole. Though the statistics become unreliable as the LET increases, we see a region of enhanced flux with reasonable statistics spanning LET from approximately 0.3–2.0 kev/µm. This region is commonly associated with low energy or “slow” protons. It is known that the transport models used by EMMREM do not handle low energy ranges well. Therefore, given this, it seems reasonable to conclude that since we observe an increase in the flux of slow protons during the largest period of discrepancy seen during the three events studied here, the most likely cause of disagreement between CRaTER and PREDICCS is that PREDICCS is not properly accounting for the effect of slow protons.
6 Summary and Conclusions
 We report here on the radiation conditions of three major SEP events that occurred during 2012. For each event, we show dose rates as measured by CRaTER and predicted by PREDICCS in order to verify the viability of the PREDICCS system for modeling radiation in the inner heliosphere. In each case, we find that PREDICCS accurately predicts the peak dose rates and cumulative dose rate for the event, but that following the peak of the event, the PREDICCS dose rate drops off more rapidly than the dose rates measured by CRaTER. This discrepancy is likely due to the inherent limitations of the EMMREM radiation transport model, which is a one-dimensional model and only incorporates protons into the calculation of dose. In the section on Discussion of CRaTER/PREDICCS Discrepancies During Events, we have shown that the most likely cause of the disparity between CRaTER and PREDICCS dose rates is that PREDICCS does a poor job of taking into account low energy or slow protons. Additionally, we describe a method for computing the modulation potential at the Moon using measured CRaTER dose rates. Using this method, we show the evolution of the modulation potential throughout the LRO mission and use it to compute the background GCR dose rate during each event. Because the computed modulation potential can be scaled out to different locations in the heliosphere [Schwadron et al., 2010], methods such as these may prove to be useful in future research as a means of computing more accurate GCR dose rates at different points of interest in the heliosphere.
 In addition, we show the total dose accumulated during each event for CRaTER, PREDICCS, and the microdosimeter. The accumulated doses for each event as well as the percent difference between PREDICCS and CRaTER are shown in Table 1. From this we see that, based on these three events, the dose accumulated by PREDICCS underestimates the dose measured by CRaTER by as much as 36%, and overestimates it by as much as 10%. This level of deviation is comparable to that found in Schwadron et al. , where it was found that the dose accumulated during the 7 June 2011 event by EMMREM underestimated that of CRaTER by approximately 38%. We find this to be a reasonable margin of error, which demonstrates the degree to which PREDICCS is reliable in predicting the total exposure for each event.
Table 1. Peak Dose Rates and Accumulated Doses for CRaTER, the Microdosimeter and PREDICCS for Various Levels of Shieldinga
Peak Dose Rate
aThe percent difference between CRaTER and PREDICCS for the most comparable level of shielding (0.3 g/cm2) is shown to demonstrate the accuracy of PREDICCS in predicting the total dose accumulated for each event. We see that for these three events, PREDICCS underestimates the accumulated dose by as much 36% and overestimates it by as much as 10%, which is reasonably accurate given the limitations of the radiation model. Similarly the peak dose rates are underestimated by as much as 37% and overestimated by as much as 15%.
PREDICCS (s = 0.3)
PREDICCS (s = 1.0)
PREDICCS (s = 5.0)
PREDICCS (s = 10.0)
CRaTER (s = 0.22)
PREDICCS/CRaTER % Difference
Microdosimeter (s = 0.89/2.28)
 In order to provide additional characterization of the three events shown here, Figure 11 shows the LET spectra of each event normalized to that of a background spectrum taken over a period of 2 years spanning 2010 and 2011. From this we can see that the January event does not vary that much from the background spectrum, with small peaks in the regions associated with slow H and He. The March event, however, shows a large enhancement for slow H and an even larger one for slow He. This result is particularly interesting since the March event has the hardest spectrum of any of the three events. Finally, the May event shows almost no enhancement for slow He, but the largest enhancement for slow H of any of the events.
 The analysis shown here demonstrates how beneficial PREDICCS may be in risk assessment for future missions by showing how different levels of shielding can reduce radiation exposure to astronauts. Based on our findings with the three events described here, we conclude that PREDICCS provides a credible means of providing a time-dependent account of the radiation conditions of the inner heliosphere, which may be of great value to future research efforts and in the planning of future missions. The recently launched PREDICCS website offers the EMMREM-generated dose rates shown here, as well as particle flux data and plots that are updated hourly as a resource to the community. Up to date dose rate data products from CRaTER as well as LET spectra will also be available on the PREDICCS website in the near future. Thus, we present the first multi-event validation of radiation environment models (e.g., EMMREM) using observations of SEP events from LRO/CRaTER that occurred over a period of 5 months in 2012. The radiation dose rates shown here are in good agreement with observations, which paves the way for their use in a broad array of applications for situational awareness and research.
 This work is supported by NASA LRO/CRaTER/PREDICCS Project (Contract NNG11PA03C), the NSF/FESD Sun-to-Ice Project (Grant AGS1135432), and the NASA/LWS/NSF EMMREM Project (Grant NNX11AC06G). We thank the ACE SIS instrument team and the ACE Science Center for providing the ACE data used here.