Ios Hot Spots in the Nearinfrared Detected by LEISA During the New Horizons Flyby

PUBLICATIONS

Journal of Geophysical Research: Planets

RESEARCH ARTICLE Io’s hot spots in the near-infrared detected by LEISA 10.1002/2014JE004670 during the New Horizons ﬂyby Key Points: Constantine C. C. Tsang1, Julie A. Rathbun2, John R. Spencer1, Brigette E. Hesman3, and Oleg Abramov4 • Io’s nightside near-infrared thermal emissions mapped 1Department of Space Studies, Southwest Research Institute, Boulder, Colorado, USA, 2Department of Physics, University of • A total of 36 hot spots detected, 4 new 3 hot spots Redlands, Redlands, California, USA, Department of Astronomy, University of Maryland, College Park, Maryland, USA, 4 • Global coverage on the nightside Astrogeology Science Center, United States Geological Survey, Flagstaff, Arizona, USA

Abstract The New Horizons spacecraft flew past Jupiter and its moons in February and March 2007. The Correspondence to: flyby provided one of the most comprehensive inventories of Io’s active plumes and hot spots yet taken, C. C. C. Tsang, [email protected] including the large 350 km high eruption of Tvashtar. Among the suite of instruments active during the flyby was the Linear Etalon Infrared Spectral Array (LEISA), a near-infrared imaging spectrometer covering the spectral range 1.25 to 2.5 μm. We have identified 37 distinct hot spots on Io in the nine LEISA spectral Citation: fl fi Tsang, C. C. C., J. A. Rathbun, J. R. Spencer, imagecubestakenduringthe yby. We describe the thermal emissions from these volcanoes and t B. E. Hesman, and O. Abramov (2014), single-component blackbody curves to the hot spot spectra to derive eruption temperatures, areas, and Io’s hot spots in the near-infrared power output for the hot spots with sufficient signal-to-noise. Of these, 11 hot spots were seen by LEISA more detected by LEISA during the New Horizons flyby, J. Geophys. Res. Planets, than once, and East Girru showed short-term variability over a few days, also seen by other New Horizons 119, 2222–2238 doi:10.1002/ instruments. This work presents a comprehensive look at the global distribution of Io’s volcanism at the time of 2014JE004670. the flyby. From these measurements, we estimate the global power output of high-temperature (>550 K) volcanism on Io to be ~8 TW. This work provides the first short-wavelength near-infrared survey with global Received 19 MAY 2014 Accepted 17 SEP 2014 coverage at all longitudes on the nightside of Io without sunlight contamination at these wavelengths. A major Accepted article online 19 SEP 2014 conclusion from this study is that 90% of all the volcanoes observed in the New Horizons LEISA near-infrared Published online 16 OCT 2014 data in 2007 were also observed during the Galileo epoch, suggesting these are all long-lived hot spots.

1. Introduction Since the discovery of active volcanism on Io in 1979 by the Voyager 1 spacecraft [Pearl et al., 1979], Io has been extensively observed and studied in an attempt to understand the source and the driving mechanisms of its widespread volcanism. Numerous studies of Ionian volcanoes have focused on their spatial distribution [Williams et al., 2011; Veeder et al., 2012], physical characteristics such as temperature and emitted areas [e.g., Allen et al., 2013], temporal variability [e.g., Davies and Ennis, 2011; Rathbun and Spencer, 2010], eruption style [e.g., Davies et al., 2006], and emitted power [e.g., Davies, 2003]. The Galileo spacecraft notably provided the most detailed snapshot of these volcanoes over 20 orbits between 1996 and 2001. Continued long-term ground-based observations of Io’svolcanism,ateverimproving spatial resolutions, especially in the near-infrared [e.g., Marchis et al., 2005], can provide the ability to study the temporal variability of hot spots over decadal timescales [e.g., Rathbun and Spencer, 2006]. However, only a few ground-based observatories have the ability to spatially resolve more than a few of the brightest individual hot spots at all Io longitudes. In this study, we present estimates of temperatures, emitting areas, and powers of 17 volcanoes, from a total of some 37 detected volcanoes observed at all longitudes, from the New Horizons ﬂyby of Io in 2007. By doing so, we provide a rare instantaneous homogeneous inventory of the global distribution of high-temperature Io volcanism, as the ﬂyby covered all longitude ranges on Io.

2. New Horizon LEISA Observations The New Horizons spacecraft was launched in January 2006 and will arrive at Pluto in 2015. One of the instruments on board is Ralph, which consists of two subsystems, MVIC, the Multispectral Visible Imaging Camera and LEISA, the Linear Etalon Infrared Spectral Array. The analysis of the MVIC data on Io’s hot spots during the New Horizons ﬂyby of Io is presented in Rathbun et al. [2014]. Here we present the LEISA observations from that same ﬂyby. LEISA covers the wavelength region between 1.25 and 2.5 μm, at spectral resolving powers of ~240, with a higher-resolving power segment (R ~ 540) covering 2.1 to 2.25 μm. Both

segments share the same detector array. Each row on the detector is sensitive to a particular wavelength. The target is scanned across the rows, resulting in a 256 × 256 × N (spatial [x], spatial + spectral [y + λ], temporal [t]) pixel image cube with a 62 μrad per pixel field of view, where N is depends on the scan rate and duration of each scan. For the Io observations, N = 356 temporal frames. These cubes can be rectified to produce conventional x, y, λ spatial/spectral image cubes, though we did not do this for the present analysis. In this work, we only use the lower resolution 1.25–2.5 μm segment due to the wider spectral sample [Reuter et al., 2008]. New Horizons flew by Io in February 2007, with closest approach occurring on 28 February at a range of 2.26 million km, with LEISA observations ranging from this closest approach to a distance of 3.04 million km. Other remote sensing observations of Io including monochromatic and color images [Spencer et al., 2007; Rathbun et al., 2014] and ultraviolet spectra [Retherford et al., 2007] were taken. Initial analysis of the detected hot spots and temperatures from LEISA was presented in Spenceretal.[2007].LEISAtooka total of nine spectral image cubes in the near-infrared, presented here in full. Io was observed at solar phase angles ranging from 36° to 120°, some when Io was in eclipse (Table 1).

2.1. Data Calibration We use Level 2 calibrated data taken from NASA’s Planetary Data System (PDS). The Level 2 data include the observed radiances and wavelengths at each pixel, and the ephemeris required to interpret the data. Further detail can be found in Reuter et al. [2008]. The PDS calibration is inaccurate, and the LEISA spectra presented in Spenceretal. [2007, hereinafter referred to as S07] used a different method of arriving at the radiometric calibration that does not provide the best possible calibration for ﬂuxes.Wehavetherefore improved the radiometric calibration for this work. An accurate radiometric calibration is important because any uncertainty in the absolute radiances can lead to uncertainty in the derived area of hot spots, while wavelength-dependent changes in the radiance lead to changes in the derived temperatures. To better determine the absolute radiances, we compared LEISA spectra taken from the standard star Vega, between 2001 and 2006 during cruise, with its well characterized 1–2.5 μm spectra from other works [Castelli and Kurucz, 1994]. The LEISA spectrum of Vega was divided by the standard Vega spectrum to produce a wavelength-dependent calibration correction. This was done three separate times during annual checkout (ACO) observations 2 (18 October 2008), 4 (24 June 2010), and 6 (1 June 2012). The average of these spectra was used as the ﬁnal conversion (Figure 1), which we applied to all Level 2 (L2) radiometric values from the PDS.

2.2. Data Reduction We extract hot spot spectra in each observation using the following procedure. We take the spectral image cube (x, y + λ, t) and adjust for the motion of Io along the time (t) dimension. We then coadd the registered (x, y + λ) frames to produce a color composite image to maximize the signal-to-noise ratio (SNR) in order to identify the locations of the hot spots. We produce image composites (Figures 3–9) using approximately 100 frames in total per observation, with frames between approximately 1.7–1.9 μmfor blue, 1.9–2.2 μm for green, and 2.2–2.4 μm for red. The contrast is stretched to highlight the hot spots on the nightside of Io, saturating the dayside. We calculate the pixel positions of each hot spot relative to a bright reference hot spot such as Tvashtar or East Girru such that the bright reference hot spot can be tracked in each of the individual spectral frames. We do this because faint hot spots are difficult to observe and track at individual frames, particularly at shorter wavelengths. The tracking of the bright reference offset hot spot is usually done by manually stepping through each spectral frame and identifying the bright reference hot spot. This is difficult to do with an automatic algorithm because of confusion from “hot” pixels caused by cosmic rays or instrument noise. The positions of each hot spot are accurate to a pixel. As described below, this precision is sufficient to identify these hot spots, but not to determine if these hot spots have moved in location since Galileo observations. Observations by the Long-Range Reconnaissance Imager (LORRI) and MVIC instruments of Io’s hot spot in a companion work [Rathbun et al., 2014] were able to distinguish spatial movement of volcanic features. For each frame, we estimate and subtract the background using a median of 50 frames before and after the frame of interest. This large number is necessary in order to avoid the bright dayside, which occupies a significant number of pixels per frame in some cases, contributing to the median background.

Table 1. All Observations Made by LEISA of the Io in the 2007 New Horizons Flyby, Ordered by Observation Date and Timea Observation Date Io Solar Range to Io Number of Hot Spots Detected Subspacecraft Latitude Observation Name File Name and Mid-Time (UTC) Phase (deg) Surface (km) (and With Power Spectrums) and Longitude (deg)

IHIRESIR01 lsb_0034820549 2007-02-26 18:31:10 36.3 3,042,371.1 0 8.6 S, 147.7 W IHIRESIR02 lsb_0034860149 2007-02-27 05:31:10 36.5 2,645,468.7 0 8.7 S, 240.7 W IECLIPSE03_01 lsb_0034890239 2007-02-27 13:52:40 46.9 2,749,419.8 3 (1) 7.5 S, 300.9 W IECLIPSE03_02 lsb_0034894319 2007-02-27 15:00:40 49.3 2,762,739.6 8 (6) 7.4 S, 300.9 W IECLIPSE03_03 lsb_0034896119 2007-02-27 15:31:44 50.4 2,768,004.0 13 (7) 7.3 S, 311.4 W IHIRESIR03 lsb_0034943249 2007-02-28 04:36:10 85.6 2,670,274.3 3 (0) 6.0 S, 26.7 W ISHINE01 lsb_0034974059 2007-02-28 13:09:39 106.7 2,417,226.2 6 (1) 5.5 S, 77.9 W IHIRESIR04 lsb_0035014949 2007-03-01 00:31:10 117.9 2,260,853.8 12 (5) 4.1 S, 163.1 W IECLIPSE04 lsb_0035048279 2007-03-01 09:46:45 119.9 2,812,918.5 8 (7) 2.1 S, 239.8 W aObservation durations were 79 s. Number of hot spots detected is the total number seen in the coadded color images. The bracketed number indicates the total number of hot spots for which this analysis was able to produce power spectra.

Finally, we sum all the radiances around the hot spot encompassed by a square box of a fixed width in each frame to produce the spectrum of the hot spot. We investigated the size of the box in which we integrate the fluxes for each hot spot by increasing the size of the box until the blackbody fluxes do not change as a function of wavelength. Increasing the box further only adds noise to the blackbody curve, and the integration box size limit is reached. All but the brightest hot spots use the 3 × 3 pixel box, except Tvashtar, East Girru, and Pele that use a 9×9 pixel due to the greater brightness of the hot spots. A sampling correction factor of 1.2 can be used to correct for the bias in the increased sampling size, but has not been used in this paper. We determine the location of hot spots on Io as follows. We calculate the angular size of Io and its pole orientation relative to the LEISA frame using information from the data headers. The only unknown remaining is the position of the center of the disk. For the observations that include part of the sunlit disk, we can fit the limb of the disk and therefore determine the position of the hot spots. For the few observations of Io in eclipse, this is not possible and we rely on a “boot-strapping” methodwhereweuse for positional reference points one or two bright hot spots, such as East Girru and Pele, whose location is known from other New Horizons remote sensing instruments, such as MVIC and the Long-Range Reconnaissance Imager (LORRI) [Rathbun et al., 2014]. As is shown in section 4, using this method, only four of the total of 37 hot spots detected by LEISA are shown to be new, with the remaining being long-persistent hot spots, giving us confidence that we are identifying these hot spots correctly. We estimate the accuracy of determining locations of hot spots from LEISA is ± 45 km due to the relative small angular size of Io and the lack of accurate limb fitting in eclipse. In comparison, LORRI images of Io provide positional accuracy to better than Correction Factor from Vega ACO2, ACO4, ACO6 20 km [Rathbun et al., 2014]. Typical ground-based adaptive optics telescopes 1.4 achieve resolutions of a few degrees latitude and longitude (20 km) 1.2 [De Kleer et al., 2014].

1.0 3. Blackbody Fits We fit the shape and absolute brightness 0.8 of the observed hot spot spectra, deriving their emitting temperatures 0.6 and areas. Given the limited spectral window afforded to us by LEISA, 1.4 1.6 1.8 2.0 2.2 2.4 between 1.25 and 2.5 μm, we fit for a Wavelengths (µm) single-temperature component for the Figure 1. Conversion factors as a function of wavelength applied to the hot component of a volcanic hot spot. LEISA level 2 radiometric data as a result of three Vega scans (ACO2, Works that use other data, such as ACO4, and ACO6) to convert from L2 units to flux values. from Galileo-Near-Infrared Mapping

Table 2. A List of Hot Spots Identiﬁed From All Nine LEISA Flyby Observations of Io, Ordered by Increasing West Longitude, and Where Possible the Derived Hot a Spot Temperatures, Areas, and Total Powers (in log10 Space so as to Present Symmetric Errors) Latitude Longitude Radiant Flux (GW/sr/μm) Log10 Hot Spot Name Observation (deg) (deg) Temperature (K) Area (km2) at 2.2 μm (Total Power (GW))

Ukko Patera IECLIPSE03_03 32 10 - - - - Ukko Patera IHIRESIR03 32 10 - - - - Kanehekilli ISHINE01 15 31 - - - - Janus ISHINE01 5 34 706 ± 105 8.2 (+18/5.7) 1.23 ± 0.26 2.078 ± 0.282 Masubi ISHINE01 43 54 - - - - Zal ISHINE01 40 75 - - - - Estan ISHINE01 22 75 - - - - Tawhaki ISHINE01 3 75 - - - - Amirani IHIRESIR04 21 115 695 ± 149 2.5 (+15/2) 0.51 ± 0.08 1.472 ± 0.490 Unidentified 4646 IHIRESIR04 60 120 - - - - Tvashtar IHIRESIR04 61 121 1239 ± 19 37 (+4/3) 384.8 ± 0.44(at 1.8 μm) 3.695 ± 0.020 Malik IHIRESIR04 33 127 486 ± 120 92 (+1679/88) 0.25 ± 0.07 2.516 ± 0.949 Thor IHIRESIR04 41 133 - - - - Tupan IHIRESIR04 17 139 553 ± 165 25 (+520/25) 0.25 ± 0.06 2.066 ± 0.872 Sobo IHIRESIR04 14 150 - - - - Shamash IHIRESIR04 35 152 - - - - Prometheus IHIRESIR04 1.4 154 498 ± 158 69 (+1682/67) 0.20 ± 0.06 2.336 ± 0.961 Culann IHIRESIR04 20 160 - - - - NIMS I32D IHIRESIR04 42 172 - - - - Zamama IHIRESIR04 17 173 - - - - NIMS I32D IECLIPSE04 45 180 - - - - Gabija/Sethlaus IECLIPSE04 55 200 - - - - Isum IECLIPSE04 31 211 781 ± 179 2 (+11/1) 1.09 ± 0.16 1.615 ± 0.473 Marduk IECLIPSE04 26 214 807 ± 165 1.5 (+6/1.2) 1.03 ± 0.15 1.53 ± 0.393 Kurdalagon IECLIPSE04 50 218 533 ± 220 95 (+3313/92) 0.74 ± 0.21 2.474 ± 1.044 East Girru IECLIPSE04 21 236 1149 ± 44 1 (+0.2/0.2) 8.73 ± 0.24 2.027 ± 0.034 East Girru IECLIPSE03_03 21 236 1152 ± 40 1.9 (+0.4/0.4) 15.81 ± 0.43 2.281 ± 0.031 East Girru IECLIPSE03_02 21 236 1017 ± 59 3.6 (+1.6/1.1) 14.22 ± 0.59 2.347 ± 0.066 East Girru IECLIPSE03_01 21 236 976 ± 109 5.2(+2.6/5.1) 14.06 ± 0.67 2.391 ± 0.118 Reiden IECLIPSE04 13 237 748 ± 178 2 (+13/2) 0.68 ± 0.19 1.489 ± 0.531 Reiden IECLIPSE03_01 13 237 - - - - Pele IECLIPSE03_03 19 256 1097 ± 40 0.8 (+0.2/0.2) 4.84 ± 0.14 1.807 ± 0.036 Pele IECLIPSE03_02 19 256 943 ± 73 1.8 (+1.3/0.7) 4.03 ± 0.21 1.888 ± 0.097 Pele IECLIPSE03_01 19 256 - - - - Pele IECLIPSE04 19 257 1023 ± 46 1.4 (+0.5/0.3) 5.39 ± 0.17 1.932 ± 0.055 Hephaestus IECLIPSE03_03 3 288 - - - - Ulgen IECLIPSE03_03 38 290 - - - - Ulgen IECLIPSE03_02 38 290 681 ± 205 6.3 (+92/5.9) 0.63 ± 0.15 1.677 ± 0.791 N Lerna IECLIPSE03_03 55 290 - - - - N. Lerna IECLIPSE03_02 55 290 842 ± 246 1.4 (+8.1/1.2) 1.11 ± 0.17 1.494 ± 0.426 N. Lerna IHIRESIR03 45 300 - - - - Dazhbog IECLIPSE03_03 54 302 864 ± 126 0.8 (+1.6/0.5) 0.99 ± 0.11 1.406 ± 0.219 Dazhbog IECLIPSE03_02 54 302 812 ± 321 2.7 (+56/2.5) 0.98 ± 0.23 1.495 ± 0.787 Loki IECLIPSE03_03 12 309 641 ± 92 7.6 (+23.6/5.7) 0.62 ± 0.07 1.82 ± 0.394 Loki IECLIPSE03_02 12 309 680 ± 249 8.6 (±8.2) 0.56 ± 0.16 1.629 ± 0.900 Mazda IECLIPSE03_03 10 310 - - - - Heno IECLIPSE03_03 56 310 - - - - Unidentified 6119 IECLIPSE03_03 30 311 - - - - Unidentified 3249 IHIRESIR03 25 315 - - - - Surt IECLIPSE03_03 41 334 619 ± 298 12 (+707/12) 0.20 ± 0.06 1.65 ± 0.54 Surt IECLIPSE03_02 41 334 - - - Unidentified 4319 IECLIPSE03_03 24 335 - - - Unidentified 4319 IECLIPSE03_02 24 335 - - - - aThe 2.2 μm radiant fluxes and powers have been corrected for emission angle.

Tvashtar Power Spectrum Comparison 300 Spectrometer (NIMS) (e.g., Davies et al. and works therein) and ground-based adaptive optics telescopes [e.g., De 250 Spencer et al. (2007) This Work Kleer et al., 2014], are able to access wavelengths as long as 5 μm, allow the 200 determination of the “warm” m) µ component that emits at lower temperatures than the hot component 150 by ﬁtting with a two-component blackbody. Since we do not have data 100 Power (GW/sr/ at those longer wavelengths, we are limited to ﬁtting with a single-

50 temperature component only. The fitting algorithm minimizes the residual between the observed 0 1.3 1.4 1.5 1.6 1.7 1.8 1.9 spectrum and the calculated blackbody Wavelengths (µm) spectrum using the Interactive Data Language (IDL) AMOEBA routine. Figure 2. A comparison between the power spectrum of Tvashtar from Once the best fit spectrum has been observation IHIRESIR04 presented in Spencer et al. [2007] (black) and from this work (blue) using the same observation with the new calibration in found, we determine uncertainties Figure 1. The best fit spectrum to our data is given in red, yielding an by generating synthetic noise with 2 emitted temperature and area of 1239 (±19) K and 39 (+4/3) km , comparable magnitude to the residuals, respectively. The new LEISA Tvashtar spectrum is missing spectra beyond and adding it to the best fit blackbody 1.9 μm due to saturated pixels, which are not possible to rectify with the curve. We fit for temperature and area current calibration. The dips at 1.4 and 1.65 μm in the newly calibrated spectrum (blue) are likely due to artifacts introduced from the Vega scans again, and using this Monte Carlo that are not included in the original Spencer et al. spectrum. Improved method repeated 200 times, calculate calibration will continue throughout the New Horizons mission. uncertainties in temperature and areas. Because the blackbody curve scales linearly with area, but temperature to the fourth, we are more sensitive to temperature than emitted area. Because flux per unit area is a strong function of temperature, the total emitting area is extremely dependent on the temperature and is thus in general poorly constrained. We therefore take the standard deviation in the Monte Carlo fits for the error in temperature, but use the standard deviation of the logarithm for the errors in area. We also take the retrieved temperature and area and convert it to radiated power, assuming Lambertian emission. Note that the powers derived here are lower limits to total power, as they include only the surface hot enough to radiate at the short LEISA wavelengths. The areas presented here are corrected for projection effects, assuming cosine dependence of flux on emission angle.

Figure 3. Observation IHIRESIR04 showing 11 distinct hot spots spread across the nightside of Io. (left) Image of the dayside crescent of Io and the massive eruption of Tvashtar. (right) The same image but with the nightside contrast stretched toreveal the 10 other hot spots. Hot spots are (1) Tvashtar (61N, 121W), (2) Amirani (24N, 115W), (3) Sobo (14N, 150W), (4) Prometheus (1°S, 154 W), (5) Tupan (18°S, 139 W), (6) Malik (33°S, 127 W), (7) Unidentiﬁed 4646 (60°S, 120 W), (8) NIMS I32D (42°S, 172 W), (9) Shamash (35°S, 152 W), (10) Culann (19°S, 160 W), and (11) Zamama (17 N, 172 W). Thor is located at (41 N, 133 W) and is not plotted here due to the saturated thermal bloom of Tvashtar but can just be made out in Figure 10. The images are a composite of approximately 100 spectral images taken by LEISA, with wavelengths from 2.2 to 2.4 μm for red, 1.9 to 2.2 μm for green, and 1.7 to 1.9 μm for blue.

Figure 4. Observation ISHINE01 showing at least six distinct hot spots spread across the nightside of Io. The right side of the image suffers from increased noise at the detectors’ edge, making identiﬁcation of speciﬁc hot spots more suspect.

4. Results The nine LEISA spectral observations of Io detected a total of 37 hot spots, with 11 of them observed more than once. Table 2 shows the hot spots identified, their locations and retrieved temperatures (K), emitting 2 areas (km ), 2.2 μm radiant fluxes (GW/sr/μm), and total power outputs (in log10 GW so that errors are symmetric). Of the 37 hot spots, we were able to fit single-temperature blackbody curves to 17, 4 of which were measured more than once at different times during the flyby sequence. The rest were too faint for useful blackbody fits. Most hot spots are long lived, allowing comparisons with ground-based and spacecraft observations, although some have retrieved temperatures and areas with significant errors that are too large to be directly compared to previous observations. Figure 2 shows a comparison of the power spectrum of Tvashtar produced in this work and that from S07. S07 derived a temperature of 1287 K for Tvashtar from the LEISA spectrum directly. Our new Tvashtar spectrum yields a temperature of 1239 (±19) K and an area of 37 (+4/3) km2. S07 also calculated an area of 49 km2 using the best estimate of Tvashtar’s temperature (1200 K, an average of LEISA- and MVIC-derived temperatures) and the flux observed from the LORRI instrument. This was done because of the preliminary nature of the LEISA calibration at the time. Therefore, the temperature and areas of Tvashtar derived here are more self- consistently derived, and explain why the spectrum from this work and that of S07 is visibly different, despite the similarities of the retrieved temperatures and areas. Figures 3–9 shows seven of the nine spectral images taken by LEISA during the 2007 Io flyby. The three otherimageseithershownohotspotsorjustasinglehotspot(EastGirru)andareomitted,although they were analyzed. For observations that include part of Io in sunlight, two panels are drawn, with brightness and contrast adjusted for the day and night sides. All seven images from Figure 3–9allhave the same red-green-blue (RGB) scheme and scaling. From each observation, an RGB image was created that stacked 100 spectral images (red: 2.2–2.4 μm, green: 1.9–2.2 μm, and blue: 1.7–1.9 μm) together to

Figure 5. Observation IHIRESIR03 reveals three distinct hot spots. (1) Ukko Patera (32 N, 10 W), (2) Unidentiﬁed 3249 (25 N, 315 W), and (3) N. Lerna (55°S, 290 W).

Figure 6. Observation IECLIPSE03_01 of Io in its gibbous phase, with three hot spots detected on the nightside. The brightest is likely East Girru near 21 N, 236 W. A weaker hot spot can be seen at 13°S, 237 W is likely Reiden, with Pele (20°S, 260 W) just visible on the terminator.

increase the SNR for the observation, such that low thermal emission hot spots could be detected. The hot spots seen in each image are shown and numbered. Figure 10 shows the nonsunlit portions of these images presented as a single global mosaic, and individual observations to illustrate coverage of Io’s surface. Figures 11–14 show, for each observation, the names of each hot spot where blackbody fits could be obtained, the derived LEISA spectra (blue), the best fit single-temperature blackbody curves (red), and the retrieved best fit temperatures and areas. These plots do not correct for emission angle. We plot the power output of all the hot spots as a function of their latitude and longitude (Figure 15). Both Figure 15 and Table 2 have fluxes, areas, and powers corrected for emission angle. Most of the hot spot temperatures range from 500 K to 900 K, with Pele, East Girru, and Tvashtar at approximately 1000, 1150, and 1240 K, respectively. Of the 11 hot spots seen more than once with LEISA, Ukko Patera, NIMS I32D, East Girru, Reiden, Pele, Ulgen, North Lerna, Dazhbog, Loki, and an unidentified hot spot at longitude 335°, only East Girru, seen in three observations, showed temporal variations (Figure 16). Pele was observed on four separate occasions, but from the three observations when power output could be determined. Pele’s power remained constant over the short period of time (~1.5 days) it was observed. North Lerna and Dazhbog were each observed twice, and their power also remained constant over the few days of observations. There were four potential new hot spots identified in the images that have not been seen in previous observations. We identify them here with their last four digits of the file name of the first observation in which they appear. All four of these hot spots do not have sufficient SNR to retrieve their blackbody emissions, and are only weakly detectable in the coadded spectral images.

5. Discussion From the LEISA observations (Table 2), there are as many active volcanoes from 0°–180°W as there are from 180°–360°W, although the number of volcanoes that are bright enough for blackbody ﬁts is greater from 180° to 360°W. This may be an artifact as the 180°–360°W hemisphere is where all the eclipse observations were made, which may be more sensitive to fainter emissions due to the lack of obscuration by scattered light from the dayside crescent. With the exception of Tvashtar at 61°N, all hot spots seen by LEISA are located between ~55°S and ~55°N. All LEISA observations were taken close to the equator (subspacecraft latitude from 8.7°S to 2.1°S), so hot spots should be visible at any latitude except the extreme north. No obvious Figure 7. Observation IECLIPSE03_02 of Io in eclipse. Eight separate hot spots are seen, although observation IECLIPSE03_03, taken soon dependency of temperature or area on after, shows more hot spots. East Girru and Pele are most notable in latitude or longitude is noticeable within the this image. relatively high uncertainties of the LEISA ﬁts.

5.1. Comparable Data Sets: Galileo and Ground Based The New Horizons LEISA data provide one of the most comprehensive global sets of observations of Io’s nighttime volcanic activity over a short period of time, at these short near-infrared wavelengths, covering all longitudes and latitudes between 75°N and 90°S. Figure 17 shows the locations of the hot spots detected by LEISA found in this study Figure 8. Observation IECLIPSE03_03: (1) East Girru (21 N, 236 W), compared to those found by Galileo-NIMS (2) Pele (19°S, 256 W), (3) Ulgen (38°S, 290 W), (4) N. Lerna (55°S, 290 W), from Lopes and Spencer [2007, Table A1]. (5) Heno (56°S, 310 W), (6) Hephaestus (3 N, 288 W), (7) Mazda (10°S, 310 W), (8) Unidentified 6119 (30°S, 310 W), (9) Dazhbo (54 N, The majority of LEISA hot spots are long-lived 302 W), (10) Loki (12 N, 309 W), (11) Surt (41 N, 334 W), (12) Unidentified sources that were also active during the 4319 (24 N, 335 W), and (13) Ukko Patera (32 N, 10 W). Galileo epoch. The observations by Galileo were biased in sampling the leading and anti- Jovian hemispheres of Io, leading to the clustering of identifications from 90 to 180°. This study also identified four potential new hot spots that were not observed by Galileo. These were identified by the end four digits of their file names; UID-4646 (60 S, 120 W), UID-6119 (30 S, 311 W), UID-3249 (25 N, 315 W), and UID-4319 (24 N, 335 W). None of these hot spots were listed in Lopes and Spencer [2007, Table A1.], Radebaugh [2005] provides additional unnamed active patera, seen by Voyager and Galileo instruments, that augment the list of hot spot detections in Lopes and Spencer [2007]. UID-4646 was not seen by Radebaugh [2005] either, but UID-6119 correlates to PV89 (31 S, 312 W), UID-3249 to either PV87 (26 N, 312 W) or PV98 (27 N, 316 W), and UID-4319 to PV129 (25 N, 335 W). However, none of these four hot spots were sufficiently bright for their radiated power to be measured. One of the drawbacks with using the LEISA data set for studying Io hot spots is that the data only extend down to 2.5 μm, which is short compared to other instruments. Systematic differences between the observations by LEISA will occur because Galileo-NIMS and ground-based Keck observations incorporate longer wavelengths. When we quote a range of surface temperatures, we are presenting shorter wavelength data that will selectively detect the higher temperatures and smaller area components, while the converse is true for longer wavelength data. In addition, we have calculated the limit of LEISA to detect faint hot spots in section 5.4. It is important to keep these issues in mind as we proceed to compare specific hot spots. We can compare our derived temperatures and areas for hot spots to previous observations by Marchis et al. [2005], Davies [2003], and Veeder et al. [2012] of temperatures and area for a large survey of Ionian hot spots. The LEISA observations presented here have spatial resolutions of 140 km, which is similar to the Marchis et al. [2005, hereinafter referred to as M05] observations of Io that had resolutions of 160 km, that used the Keck Adaptive Optics system at discrete near-infrared band-passes centered on 2.3, 3.7, and 4.8 μm. While LEISA observations have the advantage of continual wavelength coverage from 1.25 to 2.5 μm, M05 covers longer wavelengths, allowing lower temperature hot spots to be identified. The LEISA observations also have the advantage that there is no reflected sunlight contamination, because unlike M05, we are observing all longitudes at night or in eclipse. The M05 observations spanned 10 nights in 2001 and almost covered the entire surface, interrupted briefly by Jupiter. They detected 26 hot spots, two of which were new. Veeder et al. [2012] identified 240 volcanic Figure 9. Observation IECLIPSE04 showing eight distinct hot features in a study that estimated the total spots spread across the nightside of Io. (1) Isum (31 N, 211 W), (2) East Girru (21 N, 236 W), (3) Reiden (13°S, 237 W), (4) Pele (20°S, power output on Io from Galileo observations. 260 W), (5) Marduk (26°S, 214 W), (6) NIMS I32D (45S, 180 W), (7) They found 50% of the total heat flow from Io Gabija/Sethlaus (55°S, 200 W), and (8) Kurdalagon (50S, 218 W). emanated from just 1.2% of the total surface.

Figure 10. (top) A mosaic of nightside images that encompass all Ionian hot spots detected by LEISA, compiled from four LEISA observations, reprojected onto a latitude and longitude grid. Two eclipse observations (IECLIPSE03_03 and IECLIPSE04) and three crescent observations (IHIRESIR03, IHIRESIR04, and ISHINE01) have been used. Other observations either overlap directly with these images or view only the dayside of Io. The hashed grid lines provide the locations not mapped on the nightside during the ﬂyby. Each image from each observation are a stack of multiple individual spectral images from 2.2 to 2.4 μm. (bottom rows) Individually reprojected observations to show the spatial coverage as the ﬂyby progressed.

Figure 11. LEISA 4000–8000 cm 1 (1.25–2.50 μm) spectra (blue) of hot spots identiﬁed in IHIRESIR04. Best ﬁt single-component blackbody curves are plotted (red). We were only able to retrieve spectra for Amirani, Prometheus, Tupan, and Malik. Spectra for Tvashtar from this same observation are given in Figure 2. The 2.5 μm image is in the original rotation frame with North to the right of the image.

The small area, high-temperature hot spots used by their study relied on NIMS and Solid State Imaging (SSI) detected hot spots at short wavelengths, including outburst eruptions such as Tvashtar, Surt, and near the Amirani region. Their work also used average values for the power output of Pele and Janus at 280 and 320 GW, respectively (references therein for individual measurements). The LEISA observations of Pele and Janus during the 2007 ﬂyby show their respective hot component powers to þ11:9 þ110 be 75 10:2 GW and 119 57 GW. The study by Davies [2003] [hereinafter referred to as D03] analyzed 14 hot spots using 0.7–5.2 μm spectra from the Galileo-NIMS spectrometer, calculating hot spot temperatures using a two-component model.

Figure 12. LEISA 4000–8000 cm 1 (1.25–2.50 μm) spectra (blue) of seven hot spots in IECLIPSE03_02. Best ﬁt single-component blackbody curves are plotted (red).

Figure 13. LEISA 4000–8000 cm 1 (1.25–2.50 μm) spectra (blue) of seven hot spots in IECLIPSE03_03. Best ﬁt single-component blackbody curves are plotted (red).

Table 3 shows a comparison of the temperatures found in this work and those of D03 and M05. Our temperatures are all higher than those derived by M05, but much lower than those for the “hot” component derived using the two-component fits by D03, the differences likely the result from their different techniques. M05 uses longer wavelengths, so one would expect lower temperatures, and two-component fits are likely to produce higher temperatures for the hottest component than one-component fits to the same data. þ3000 þ52 We also measured the high-temperature powers of Kurdalagon (298 284 GW), North Lerna (31 20 GW), þ62 and Dazhbog (28 19 GW). Rathbun et al. [2004] measured total power for Kurdalagon and Dazhbog using Galileo Photopolarimeter Radiometer (PPR). Previous observations of both Tvashtar and Pele obtained power estimates that can be compared to this work, especially Tvashtar which is known to be highly variable. Milazzo et al. [2005] found Tvashtar had brightness temperatures greater than 1300 K, covering an area of 5.5 km2 and a total power output of 0.89 TW (November 1999 and February 2000), as seen by Galileo-SSI from 0.41 to 0.97 μm. LEISA measured a similar temperature at 1239 K but with a larger area of 37 km2 and a power of 4.95 TW for the hot temperature þ11:9 component. The hot component power measured at Pele with LEISA was 75 10:2 GW. Further inferences and comparisons are found in Rathbun et al. [2014]. Tupan Patera and Janus Patera were also measured by Galileo-NIMS [Davies and Ennis, 2011; Davies et al., 2012]. Janus Patera was measured on two separate occasions, with an average of 1075 K, 0.6 km2, and 179 GW for the hot component. We derive from the single-body fit a temperature lower than NIMS, at 706 K, an area of 8.2 km2, and a power of 119 GW, emitting less power than during the Galileo observations. The power of Tupan Patera derived from the NIMS observations was 79 GW assuming a two-component fit and showed variability even between 1996 and 2001 during the Galileo epoch. The LEISA observations showed that the single hot component of Tupan had a power of 116 GW. Dazhbog Patera (54 N, 301 W) was not detected by NIMS but was seen by Photopolarimeter Radiometer (PPR) data [Rathbun et al., 2004] that covers the much longer wavelengths from 17 to 110 μm. We see Dazhbog

Figure 14. LEISA 4000–8000 cm 1 (1.25–2.50 μm) spectra (blue) of hot spots identiﬁed in IECLIPSE04. Best ﬁt single-component blackbody curves are plotted (red). East Girru and Pele are distinctly brighter, leading to better SNR spectra. The 2.5 μm image is in the original rotation frame with North to the right of the image.

active in the LEISA data that covers the NIMS spectral range, which indicates Dazhbog has become more active in the intervening years (or that LEISA was more sensitive than NIMS at these longitudes, where NIMS did not get good data). Mazda Catena (10 S, 310 W) and Heno Patera (56 S, 310 W) show a similar pattern, both were not detected by NIMS but were seen by PPR and again with LEISA at shorter wavelengths.

5.2. Comparison With Other New Horizons Observations The LEISA observations described here are contemporary with other remote sensing data from the New Horizons flyby of Io, which were recently analyzed by Rathbun et al. [2014], expanding upon the initial analysis by Spencer et al. [2007]. Rathbun et al. [2014] analyzed data from (1) the four-color MVIC instrument, which obtained 17 sets of images at wavelengths spanning the visible through to the near-infrared, and (2) the LORRI instrument, which observed Io at high spatial resolution at broadband monochromatic wavelengths between 400 and 900 nm. Rathbun et al. analyzed a total of 190 LORRI images, providing New Horizon’s most accurate estimates of locations of the detected hot spots. The MVIC observations have lower spatial resolution but provide the possibility of a partial reconstruction of the blackbody emission by using its four-color-filtered channels centered 502, 647, 875, and 890 nm. Despite the broad and overlapping filter responses, the broad wavelength coverage of MVIC and LORRI allowed Rathbun et al. [2014] to constrain temperatures and areas for Tvashtar from those data sets. They measured the color temperature from the four individual MVIC filters. Temperatures and areas of 1240 (±60 K) at 50 km2 flow or 1280 K (±100) and 30 km2 were derived for Tvashtar, depending on which color ratio was used. Their spatial and temporal resolution is significantly superior to the LEISA data set. The LEISA data set provides much better estimates of hot spot temperatures and areas, while MVIC and LORRI data sets are better suited for identifying hot spots locations and describing any short temporal variability. From the MVIC and LORRI data set, Rathbun et al. [2014] determined brightness of Reiden, Pele, and East Girru at timescales of seconds to hours. Only East Girru showed short-term variability at the 10–20% level. They also

Figure 15. The longitude (blue) and latitude (red) distribution of hot spots detected by LEISA and where temperature and area could be determined.

showed East Girru changed brightness over the course of 6 days, decreasing by a factor of 4, from 27 February to 3 March. Our four LEISA observations of East Girru span a 2 day period, from 27 February to 1 March (Figure 16). The LEISA data also detected a drop in the power of East Girru in this time period, by a factor of 2.5, qualitatively consistent with the drop detected by LORRI and MVIC. LEISA also made three independent measurements of Pele, but no change in power is noted outside of the measured errors. Only Figure 16. The temporal variability of East Girru in the LEISA data sets from one measurement of Reiden’s power February and March 2013. was retrieved.

In all, the LORRI observations (which only observed 60% of the surface, missing longitudes 30°–170°W) revealed a total of 54 distinct emission sources on Io, 13 of which are extended sources and might not be thermal in nature. The LEISA observations detected fewer hot spots, 37 in total, despite their global coverage, showing that a signiﬁcant number of hot spots were not detected by LEISA but were detected by LORRI. The 41 thermal features seen by LORRI and 37 by LEISA can therefore be considered lower limits on the total number of eruptions that are taking place on Io at any given time. Figure 18 showsthelocationsofhotspotsdetectedbybothLORRIandMVICfromRathbun et al.[2014]andthose found by LEISA from this study.

5.3. Power Emitted From Volcanoes Observed by LEISA The total high-temperature power output of hot spots observed by LEISA is 7.6 × 1012 W, with Tvashtar contributing 65% of this total. The total is well below the estimated total thermal output from Io, from other studies listed in Veeder et al.[2012],Table2,whicharebetween5×1013 W and 1.1 × 1014 W, and signiﬁcantly below the power predicted by Veeder et al. [1994] of ~ 1.25 × 1014 W. The much lower power seen by LEISA is to be expected given that we are insensitive to low-temperature volcanic emissions which peaks at much longer wavelengths, and which contributes the bulk of the thermal emission from Io. The LEISA derived value of 7 × 1012 W can be viewed as an estimate of the instantaneous hot component emitted by Io.

0 Latitude

-30

-60 LEISA sources NIMS sources -90 360 330 300 270 240 210 180 150 120 90 60 30 0 W. Longitude

Figure 17. Hot spots detected by LEISA from this study (red) compared with those from Galileo-NIMS (blue). The hot spots locations from Galileo are from Lopes and Spencer [2007, Appendix 1]. The base map is from Galileo-SSI.

Table 3. A Comparison of the Derived Temperatures of Hot Spots From This Work (T14) Using New Horizons LEISA Data From 1.25 to 2.5 μm, and Those Derived From Marchis et al. [2005] (M05) and Davies [2003] (D03) From Keck-AO Filtered Channel and Galileo-NIMS 0.7–5.1 μm Data, Respectivelya Hot Spot Name T14 Temperatures (K) M05 Temperatures (K) D03 Temperatures (K)

Janus 706 ± 105 557 ± 103 - Amirani 695 ± 149 364 ± 93 1022 Tvashtar 1239 ± 19 - - Malik 486 ± 120 - 586 Tupan 553 ± 165 470 ± 157 974 Prometheus 498 ± 158 396 ± 98 1263 Isum 781 ± 179 406 ± 52 - Marduk 807 ± 165 469 ± 69 - Kurdalagon 533 ± 220 - - East Girru 1149 ± 44 - - Reiden 748 ± 178 - - Pele 1097 ± 40 - - Ulgen 754 ± 85 378 ± 120 - N Lerna 642 ± 88 - - Dazhbog 864 ± 126 - - Loki 641 ± 92 342 ± 22 - Surt 628 ± 102 440 ± 60 - aFor each hot spot from M05, we took the average temperature derived across the number of days of observations. For the D03 temperatures, we show the “hot” component values derived from the two-body component ﬁt across the 1–5 μm range.

5.4. LEISA Detectability Limit We conducted a test of the LEISA instrument’s detectability limit for Io hot spots for the time of the New Horizons flyby. We produced a series of synthetic blackbody spectra from 1.25 to 2.5 μm, at 50 different emitting temperatures and 50 different emitting areas, from 300 to 800K and 0.2 to 119km2, respectively. The hot spots were placed at an average distance of 2,500,000 km, comparable to the range oftheLEISAimagecubes,andallhotspotswereassumedtobeatthenadirpoint.Theaveragenoise level of LEISA (5 × 10 10 W/cm2/sr/μm), determined by looking at the background flux from 1.25 to 2.0 μm, was added to these spectra. Each of these spectra was then fitted to retrieve temperatures and areas as was done with the real LEISA observations of Io. Figure 19 shows the results of this simulation. If the retrieved temperatures are the same as the temperatures used to create the synthetic spectrum, then they will lie on the 1:1 line (black), indicating perfect precision. For guidance, we also plot the ±25K

0 Latitude -30

-60 LORRI sources LEISA hotspots MVIC hotspots -90 360 330 300 270 240 210 180 150 120 90 60 30 0 W. Longitude

Figure 18. Hot spots detected by LEISA (red) compared with those found by the MVIC (yellow) and LORRI (blue) instruments on New Horizons [Rathbun et al., 2014]. The sizes of the circles from LORRI, MVIC, and LEISA are representative of their relative brightnesses. The base map is from Galileo-SSI.

deviation from the 1:1 in blue. At high temperatures, the spectra can be ﬁtted accurately, even for low emitting areas. At lower temperatures, the precision with which temperature can be retrieved decreases, until the noise dominates the spectrum and no meaningful quantity can be retrieved. The temperature lower limit for useful temperature constraints with LEISA is near 550 K: if a hot spot with this temperature has an emitting area of 9 km2 or larger temperatures can be retrieved with a precision Figure 19. A test of the ability of LEISA to detect hot spots on Io, showing better than 25 K. The lowest a hot spot at ~550K or higher, and an area of ~9km2 or larger has a blackbody emission that is retrievable to a precision better than 25 K temperature we measured in this (blue lines). study during the 2007 New Horizons ﬂyby was Malik, which had a poorly determined temperature of 486±120K, but had a very large emitting area (92km2) that allowed it to be detected. The LEISA detection limit is at least 200 K higher than that of M05 (Table 3), owing to the longer wavelength observations afforded by Keck-NIR. A detectability limit analysis is useful for determining what level of volcanic activity was present in the past that should be powerful enough to be detected by LEISA, but was not, indicating a lessening of activity of a particular hot spot. For example, Kanehekili Fluctus was seen to have a hot component of 1000K, area of 0.75 km2, and output power of ~500 GW [Davies et al., 2012], which should have been detected by LEISA if it had those powers, but was not, indicating a lessening in its activity. Finally, we suggest future observations of hot spots on Io include a similar study on detectability to better understand the detection threshold and how many potential hot spots are truly variable from year to year and not just missed due to instrument limitations.

6. Conclusions We present near-infrared (1.25 to 2.5 μm) observations of Io’s volcanic activity using the New Horizons LEISA spectrometer data from 2007. All nine LEISA spectral cubes were analyzed to locate, identify, and, where possible, retrieve blackbody curves to fit for eruption temperatures, areas, and power outputs. A total of 37 distinct hot spots where identified, 11 of which were imaged more than once. As well as confirming all hot spots detected by a previous study [Spencer et al., 2007], we detected a further four potentially new hot spots, but they were all too low in intensity to fit blackbody curves to. The majority of the observed hot spots are long lived, seen during the Galileo era and from ground-based observations in the late 1990s and mid-2000s, respectively. New calibration of LEISA was applied to more accurately measure the spectral emission of these hot spots. We were able to fit single-temperature blackbody curves to a total of 17 hot spots, four of which more than once during the flyby. Tvashtar was the most powerful emission in the New Horizons data set, at 1239 (±19) K and an area of 37 (+4/3) km2,values that are slightly different found by Spencer et al.[2007].Wefind the emitted power of East Girru fell during the 48 h LEISA observed the hot spot, decreasing in power by a factor of 2.5 in this period, an effect also identified in LORRI and MVIC data by Rathbun et al. [2014]. This work represents one of the most complete surveys in longitude at short wavelengths of Io’s nighttime volcanic activity. No discernable variation in longitude of these high temperature hot spots was observed. LEISA detected only hot spot components with temperatures above about 500 K: the total power radiated at these high temperatures was measured by LEISA to be about 8 TW, a small fraction of Io’s total volcanic output of about 100 TW.

Acknowledgments References We would like to thank the LEISA team for the archived data. We also thank Allen, D., J. Radebaugh, and D. C. Stephens (2013), Temperature and variability of Pillan, Wayland Patera and Loki Patera on Io from Cassini Cathy Olkin and Allen Lunsford for their ISS data, Icarus, 226,77–88. help with the LEISA calibration. This Castelli, F., and R. L. Kurucz (1994), Model atmospheres for Vega, Astron. Astrophys., 281, 817–832. work was funded by the NASA Jupiter Davies, A. G. (2003), Volcanism on Io: Estimation of eruption parameters from Galileo NIMS data, J. Geophys. Res., 108(E9), 5106, doi:10.1029/ Data Analysis Program NNX11AK15G 2001JE001509. and the New Horizons mission. We also Davies, A. G., and M. E. Ennis (2011), The variability of volcanic activity at Zamama, Culann, and Tupan Patera on Io as seen by the Galileo near thank D. Williams and an anonymous infrared mapping spectrometer, Icarus, 215, 401–416. reviewer for their comments and Davies, A. G., L. Wilson, D. Matson, G. Leone, L. Keszthelyi, and W. Jaeger (2006), The heartbeat of the volcano: The discovery of episodic helpful suggestions to improve activity at Prometheus on Io, Icarus, 184, 460–477. this paper. Davies, A. G., G. J. Veeder, D. L. Matson, and T. V. Johnson (2012), Charting thermal emission variability at Pele, Janus Patera and Kanehekili Fluctus with the Galileo NIMS Io Thermal Emission Database (NITED), Icarus, 221, 466–470. De Kleer, K., I. de Pater, A. G. Davies, and M. Adamkovics (2014), Near-infrared monitoring of Io and detection of a violent outburst on 29 August 2013, Icarus, doi:10.1016/j.icarus.2014.06.006, in press. Lopes, R. M. C., and J. R. Spencer (Eds.) (2007), Io After Galileo: A New of Jupiter’s Volcanic Moon, 342 pp., Springer Praxis Publishing, Chichester, U. K. Marchis, F., et al. (2005), Keck AO survey of Io global volcanic activity between 2 and 5 microns, Icarus, 176,96–122. Milazzo, M. P., L. P. Keszthelyi, J. Radebaugh, A. G. Davies, E. P. Turtle, P. Geissler, and K. P. Klassen (2005), Volcanic activity at Tvashtar Catena, Io, Icarus, 235–251. Pearl, J. C., R. Hanel, V. Kunde, W. Maguire, K. Fox, S. Gupta, C. Ponnamperuma, and F. Raulin (1979), Identification of gaseous SO2 and new upper limits for other gases on Io, Nature, 288(5725), 757–758. Radebaugh, J. (2005), Formation and Evolution of Paterae on Jupiter’s Moon of Io, PhD thesis, Univ. of Arizona, Ariz. Rathbun, J. A., and J. R. Spencer (2006), Loki, Io: New ground-based observations and a model describing the change from periodic overturn, Geophys. Res. Lett., 33, L17201, doi:10.1029/2006GL026844. Rathbun, J. A., and J. R. Spencer (2010), Ground-based observations of time variability in multiple active volcanoes on Io, Icarus, 209, 625–630. Rathbun, J. A., J. R. Spencer, L. K. Tamppari, T. Z. Martin, L. Barnard, and L. D. Travis (2004), Mapping of Io’s thermal radiation by the Galileo photopolarimeter-radiometer (PPR) instrument, Icarus, 169, 127–139. Rathbun, J. A., J. R. Spencer, R. M. Lopes, and R. R. Howell (2014), Io’s active volcanoes during the New Horizons Era: Insights from New Horizons imaging, Icarus, 231, 261–272. Retherford, K., et al. (2007), Io’s atmospheric response to eclipse: UV aurorae observations, Science, 318, 237–240. Reuter, D. C., et al. (2008), Ralph: A visible/infrared imager for the New Horizons Pluto/Kuiper Belt Mission, Space Sci. Rev., 140, 129–154. Spencer, J. R., et al. (2007), Io Volcanism seen by New Horizons: A major Eruption of the Tvashtar volcano, Science, 318, 240–243. Veeder, G. J., D. L. Matson, T. V. Johnson, D. L. Blaney, and J. D. Goguen (1994), Io’s heat flow from infrared radiometry: 1983–1993, J. Geophys. Res., 99(E8), 17,095–17,162, doi:10.1029/94JE00637. Veeder, G. J., A. G. Davies, D. L. Matson, T. V. Johnson, D. A. Williams, and J. Radebaugh (2012), Io: Volcanic thermal sources and global heat flow, Icarus, 219, 701–722. Williams, D. A., L. P. Keszthelyi, D. A. Crown, J. A. Yff, W. L. Jaeger, P. M. Schenk, P. E. Geissler, and T. L. Becker (2011), Volcanism on Io: New insights from global geologic mapping, Icarus, 214,91–112, doi:10.1016/j.icarus.2011.05.007.