New Global Map of Io’s Volcanic Thermal Emission and Discovery of Hemispherical Dichotomies

By combining multiple spacecraft and telescope data sets, the first fully global volcanic heat flow map of Io has been created, incorporating data down to spatial resolutions of ∼10 km pixel−1 in Io’s polar regions. Juno Jovian Infrared Auroral Mapper data have filled coverage gaps in Io’s polar regions and other areas poorly imaged by Galileo instruments. A total of 343 thermal sources are identified in data up to mid-2023. While poor correlations are found between the longitudinal distribution of volcanic thermal emission and radially integrated end-member models of internal heating, the best correlations are found with shallow asthenospheric tidal heating and magma ocean models and negative correlations with the deep-mantle heating model. The presence of polar volcanoes supports, but does not necessarily confirm, the presence of a magma ocean on Io. We find that the number of active volcanoes per unit area in polar regions is no different from that at lower latitudes, but we find that Io’s polar volcanoes are smaller, in terms of thermal emission, than those at lower latitudes. Half as much energy is emitted from polar volcanoes as from those at lower latitudes, and the thermal emission from the north polar cap volcanoes is twice that of those in the south polar cap. Apparent dichotomies in terms of volcanic advection and resulting power output exist between sub- and anti-Jovian hemispheres, between polar regions and lower latitudes, and between the north and south polar regions, possibly due to internal asymmetries or variations in lithospheric thickness.


Introduction
Volcanic activity on Io is driven by tidal dissipation within Io (Peale et al. 1979;Yoder & Peale 1981;Segatz et al. 1988;Lainey et al. 2009;Tyler et al. 2015;Fuller et al. 2016).As a result, Io is the most volcanically active body in the Solar System (Davies 2007;Lopes & Spencer 2007;Cantrall et al. 2018;de Kleer et al. 2019a).Historical estimates of the total thermal energy emitted by volcanic sources have been larger, by up to an order of magnitude, than explained by equilibrium tidal heating models (e.g., Matson et al. 1981;Veeder et al. 1994;Spencer et al. 2000;Rathbun et al. 2004).Estimates of heat flow per unit area therefore span a broad range from region to region.
Io's orbital evolution is tightly locked to those of Europa and Ganymede.Cyclic feedback between thermal and orbital evolution is predicted to result in synchronized oscillations in the tidal heating of all three moons with periods on the order of 100 Ma (Ojakangas & Stevenson 1986;Hussmann & Spohn 2004).De Kleer et al. (2019b) found that understanding the evolution of the system required unique spacecraft observations of Io's volcanic, geophysical, and orbital processes to understand Io's coupled thermal and orbital evolution.This would significantly improve our understanding of how other tidally heated satellites (especially the largely silicate moons Europa and Enceladus) and extrasolar planets have evolved.At the same time, Io is an extraordinary planetsized laboratory for examining large-scale volcanic activity in real time.Understanding Io's volcanic processes and global heat flow is therefore of supreme importance, requiring mapping all of Io's volcanic activity.
The data returned from NASA's Juno spacecraft from its polar orbit around Jupiter have revealed Io's polar volcanoes in the infrared at spatial scales (up to orbit PJ49) as high as 13 km pixel −1 .New Juno Jupiter Infrared Auroral Mapper (JIRAM) hot spot detections (Adriani et al. 2014;Mura et al. 2020;Davies et al. 2024) have been added to previous analyses to create an updated Io volcanic thermal emission map (Figures 1  and 2).JIRAM observations obtained from 2017 March 27 (Juno orbit PJ05) through 2023 March 1 (Juno orbit PJ49) have allowed the identification of 273 active volcanic thermal sources (Davies et al. 2024) and the quantification of volcanic thermal emission from Io's major polar volcanoes.The JIRAM data provide a global snapshot of where effusive high-temperature (silicate) volcanic activity is currently taking place on Io.Significantly, the JIRAM polar observations of Io have allowed us to fill in gaps in the global map of volcanic activity (Davies et al. 2015;Veeder et al. 2015) to create the first truly global map of Io's ongoing volcanic activity from hot spots detected in improved data with spatial resolutions down to ∼10 km pixel −1 .

Observations of Io's Volcanic Activity
JIRAM can collect image data at 3.5 μm (L band) and 4.8 μm (M band).These wavelengths are particularly sensitive to  As a suitable heat flow model discriminator, 60°latitude is used to demarcate Io's polar regions.This latitude was chosen because this was where the greatest changes in heat flow occurred when comparing models, and previous works used this latitude (Milazzo et al. 2005;Davies et al. 2024) to demark the polar boundary, allowing direct comparison of results.
thermal emission from ongoing or recent high-temperature (silicate) volcanic activity that results in the eruption or exposure of hot lava at Io's surface (Davies et al. 2010), making JIRAM an excellent instrument for detecting and observing volcanic activity.The JIRAM data can therefore be directly compared with observations of volcanic activity obtained by the Galileo Near Infrared Mapping Spectrometer (NIMS; Davies et al. 1997;Lopes-Gautier et al. 1999;Davies et al. 2001Davies et al. , 2012aDavies et al. , 2012bDavies et al. , 2014;;Lopes et al. 2001Lopes et al. , 2004)), which obtained data between 0.7 and 5.3 μm from 1996 through 2001; ground-based telescopes such as NASA's InfraRed Telescope Facility (IRTF; at low spatial resolution), which has collected data for decades (from Veeder et al. 1994to Tate et al. 2023); the ground-based Keck (Marchis et al. 2000(Marchis et al. , 2002;;de Pater et al. 2004;Marchis et al. 2005;de Kleer & de Pater 2016a, 2016b;Cantrall et al. 2018) andGemini (de Kleer et al. 2014) telescopes (at spatial resolutions comparable to much of the NIMS data); and the James Webb Space Telescope (JWST), whose NIRspec instrument obtains relatively low spatial resolution but extremely high spectral resolution data (de Pater et al. 2023).In addition to the near-infrared data, Galileo Photo-Polarimeter Radiometer (PPR) data (e.g., Rathbun et al. 2004) were obtained between 1996 and 2002.The PPR, operating at longer wavelengths than NIMS, was sensitive to areas of recent volcanic activity too cool (∼<200 K) for NIMS to detect (Davies et al. 2010).Io has been observed by other flyby missions: Cassini in 2000 and New Horizons in 2007.Both obtained data of Io's hot spots (e.g., Radebaugh et al. 2004;Spencer et al. 2007).Since 2017, Io has been imaged regularly by instruments on the JIRAM spacecraft (Adriani et al. 2014;Bolton et al. 2017;Hansen et al. 2017;Mura et al. 2020;Becker 2021;Davies et al. 2024).
Recent analysis of Juno JIRAM data identified 266 hot spots in data collected up to Juno orbit PJ43 (Davies et al. 2024).Another seven hot spots were subsequently identified in JIRAM data up to orbit PJ49 (2023 March 1) and are listed in Table 1.Analysis of these data has shown that the thermal emission from Io's active volcanoes is not uniform across the moon, both longitudinally and latitudinally.Examining the 4.8 μm spectral radiance from Io's active polar volcanoes (using >60°latitude to define polar regions) showed that, while found to be as numerous per unit area as at lower latitudes, Io's polar region volcanoes emit less 4.8 μm spectral radiance than those at lower latitudes.Additionally, volcanoes in Io's north polar cap emit twice as much 4.8 μm spectral radiance as those in the south polar cap (Davies et al. 2024).
JIRAM, preferentially observing Io's polar regions rather than lower latitudes, has filled in important coverage gaps left over from previous missions that generally imaged Io from the equatorial plane.The hot spots observed across Io by JIRAM do not represent the full extent of volcanic activity, as JIRAM is not sensitive to cool thermal anomalies on Io's surface (Davies et al. 2024).We have identified all new JIRAM hot spot detections, calculated the total thermal emission from each hot spot, and combined the thermal emissions with previous detections (Davies et al. 2015;Veeder et al. 2015;Cantrall et al. 2018) to create a more complete global inventory and distribution of volcanic heat flow.

Io's Volcanic Heat Flow: Previous Work
Previous analyses have examined the latitudinal and longitudinal distribution of Io's hot spots using incomplete knowledge of the full hot spot catalog (Lopes-Gautier et al. 1999;Hamilton et al. 2013;Tyler et al. 2015).The first attempt at a global accounting of Io's volcanic heat flow was performed by Veeder et al. (2015), who quantified thermal emissions from 250 locations on Io, a list compiled from all available sources.This total included eight "outburst" eruptions, now known to be characterized by voluminous lava fountains (Davies 1996;Keszthelyi et al. 2001) that, while immensely powerful, are transient and so do not contribute much (<2%) to Io's yearly heat flow budget (Davies 2007).The volcanic heat flow map based on the Veeder et al. (2015) data set was published in Davies et al. (2015).The analyses of Veeder et al. (2012Veeder et al. ( , 2015) ) and the volcanic map showed that Io's volcanic thermal emission was far from uniform (Figure 3), with longitudinal peaks offset eastward some 40°-60°from the predicted location from a radially integrated asthenospheric tidal heating model (Ross et al. 1990).A similar analysis was performed on an extensive catalog of observations obtained using groundbased telescopes equipped with adaptive optics (AO; de Kleer & de Pater 2016a, 2016b; Figure 3).Although the AO data analysis was limited to 62 volcanoes, these were Io's most prominent thermal sources in terms of magnitude of thermal emission in the near-infrared.The resulting distribution of volcanic heat flow (de Kleer & de Pater 2016a) closely matched the Veeder et al. (2015) distribution-the smaller volcanoes not detected by telescopes utilizing AO contributed relatively little to Io's overall thermal emission pattern.

Io's Polar Volcanoes
Tidal heating drives Io's volcanism (Peale et al. 1979;Segatz et al. 1988).A total of 373 volcanic thermal sources have now been identified so far in JIRAM data, with the promise of further identifications from data already obtained and returned but yet to be released to the NASA Planetary Data System.Most if not all significant hot spots in Io's polar regions and elsewhere have likely been identified in JIRAM data (Mura et al. 2020;Davies et al. 2024).
The JIRAM data allow comparison between polar volcanoes (with latitudes >60°) and those at lower latitudes.Fundamental differences in volcanic activity were found (Davies et al. 2024).Io's polar volcanoes are less energetic than those at lower latitudes by a factor of 2, and energy released from the north polar region is twice that of the south polar region (Davies et al. 2024).The distribution of energy as measured by JIRAM suggests a preponderance of shallow (asthenospheric) tidal heating or the presence of a global magma ocean (Khurana et al. 2011) over deep tidal heating (Segatz et al. 1988; Note.Power outputs are given in the Appendix (Table A1).Ross et al. 1990;Tyler et al. 2015;Matsuyama et al. 2022).However, JIRAM is only sensitive to part of the thermal emission from Io's volcanic centers, being limited (as was NIMS) to surface temperatures above approximately 200 K.The methodology of deriving JIRAM hot spot spectral radiances is provided in Davies et al. (2024).Further processing of the JIRAM spectral radiance data was performed to derive the color temperature (T, K) of each hot spot where both L-band and M-band data were obtained contemporaneously.The L/M ratio has a unique solution to a single temperature, which for the purpose of this analysis is derived iteratively.The magnitude of spectral radiance at either the L or M band was then used to generate the emitting area of the thermal source (A, m 2 ), and the total power output (Q, W) was determined using Equation (1): where σ = Stefan-Boltzmann constant = 5.67 × 10 −8 W m −2 K −4 and ε = emissivity (=1).
Where spectral radiance for a hot spot was only available in the M band (4.8 μm), an empirical relationship (Equation ( 2)) derived from the analysis of hundreds of NIMS hot spot spectra (Davies & Veeder 2023) is used.Here, total thermal emission Q (GW) is derived from 4.8 μm spectral radiance L 4.8 (GW μm -1 ), where . 2 4.8 0.8838

Hot Spot Detections and Sources
Of the 343 hot spot power estimates included in this analysis, 242 come from Veeder et al. (2015), which was compiled mostly from Galileo NIMS and PPR data analyses but also included some ground-based telescope hot spot detections.The Veeder et al. (2015) analysis was unique inasmuch as it sought to quantify the total thermal emission from Io's volcanoes from all observed and likely sources, which necessitated assigning temperatures to a few low-albedo features based on the fact that they were sites of recent volcanic activity that were still warm enough to prevent volatiles condensing on them, even though there were no direct observations of thermal emission with NIMS.The area of the low-albedo feature at the assumed temperature of 130 K (see Veeder et al. 2009Veeder et al. , 2011Veeder et al. , 2012) ) was then used to calculate power output.About a dozen PPR detections of spectral radiance at one wavelength, where no image data showing a surface feature were available, were assigned a power based on similar detections elsewhere on Io, as a first-order estimate of likely power.Since 2015, subsequent data analyses of Galileo NIMS (Davies & Veeder 2023) and additional ground-based detections (Cantrall et al. 2018;de Kleer et al. 2019b) have allowed refinement of a number of these table entries, and these more recent, more robust temperature, area, and power estimates (Table 2) have been incorporated into the revised hot spot catalog.
Also added to the hot spot catalog are 14 detections from ground-bases telescopes post-2015 (Cantrall et al. 2018).To the growing list were then added 87 detections from previously unknown thermal sources identified in JIRAM data (Davies et al. 2024), including nine sources identified in PJ47 and PJ49 data.The largest group of new hot spots (Figure 4) was located at longitudes 0°W-60°W, a region covered at relatively low spatial resolution by Galileo images, and in the polar regions above 60°latitude.The Appendix contains the power emitted from the 343 distinct thermal sources plotted in Figure 1.

Data Processing
The JIRAM data processing methodology we use to derive spectral radiance (GW μm -1 ) is described in detail in Davies et al. (2024).In short, each JIRAM observation was geolocated within an Io ground-control framework informed by NASA Navigation and Ancillary Information Facility Spacecraft, Planet, Instrument, C-matrix, Events ("SPICE") kernels and carefully adjusted to account for SPICE kernel position uncertainties, which were small (usually requiring a shift of  Veeder et al. (2015).The long arrows indicate the longitudes of maximum heat flow predicted using an asthenospheric heating model (Segatz et al. 1988;Ross et al. 1990).The shorter arrows are the locations of secondary maxima.30°bins are used.Loki Patera (not binned) is plotted to show its position and average thermal emission (9600 TW; Matson et al. 2006).The total thermal emission shown excludes outbursts.Figure from Davies et al. (2015).The data here compare well with a plot of longitudinal power output derived from Keck and Gemini telescope data (de Kleer & de Pater 2016a; see below).
no more than a few pixels) but highly significant for identifying the location of hot spots on Io's surface.The band radiance for each hot pixel was converted to a spectral radiance and corrected for spacecraft distance to target pixel and emission angle, assuming Lambertian emission.If the hot spot pixel band radiance exceeded the equivalent of an Experiment Data Record data number (DN) value of 12,000, above which the instrument detector is not calibrated (Adriani et al. 2014;Davies et al. 2024), the spectral radiance is flagged as "saturated."This is a misnomer, as the detector is not technically "saturated": saturation occurs at a DN value of ∼16,000.Instead, this 12,000 DN limit-equivalent to an M-band radiance of 0.005 84 W m −2 sr −1 for a 1 s exposure-is the upper limit of DN values where the JIRAM detector response is still linear (Adriani et al. 2014).Above this DN value, processing and calibration of JIRAM EDR data to band radiance sets the upper bounday of band radiance to the equivalent of the EDR DN value of about 12,000.In our processing pipeline "saturation masks" are created for each JIRAM observation, where all pixels above this limit are flagged.For the purposes of temperature derivation, we use maximum values of spectral radiance below this limit, as did Davies et al. (2024).

Hot Spot Value Selection
When the position of a JIRAM hot spot coincided with a previous detection in the same location, and the prior detection was derived from Galileo data, the prior thermal emission value was used.Where the prior detection used an estimate of power based on instrument detection limits, or power output was assigned, the JIRAM estimate was used.This led to the downsizing of some hot spot thermal emission numbers.A few hot spots, listed in Table 3, that were detected by JIRAM are not included in the new hot spot list.These new thermal emission values for hot spots are in a slightly different location on the volcano than the original detection.As the goal is to determine the most appropriate value of total thermal emission from the hot spot and not precisely where on the volcano the thermal emission might be sourced, the original, 2015 derived value is used if the average thermal emission is larger than the JIRAM-derived estimate.Portions of hot spots in other colors reflect probable changes of hot spot thermal emission location at a given volcano where the nearest pre-Juno hot spot thermal emission is preferentially used, in most cases where hot spot power is derived from Galileo NIMS spectra.A large cluster of post-2015 hot spots is found in the northern hemisphere at longitudes between 0°W and 60°W, a region poorly covered by Galileo.

Results
The Appendix contains the thermal emission from 343 locations.The range of power outputs spans 5 orders of magnitude.The most powerful hot spot in the list is Loki Patera, which varies between about 6000 and 12,000 GW.We use the average output value of 9600 GW (Matson et al. 2006).The least powerful hot spot identified in the data set is a JIRAM detection, designated JRM111 by Davies et al. (2024) and located at 162.6°W, 46°N.This hot spot is small but hot, with an area of only 0.0029 km 2 and a color temperature derived from the L-band/M-band ratio of 1035 K. Power output is 0.186 GW.The hot spot was observed twice, during Juno orbits PJ7 and PJ43.There are seven hot spots with thermal emission less than 1 GW, 76 hot spots with powers between 1 and 10 GW, 150 hot spots with powers from 10 to 100 GW, 101 hot spots with powers between 100 and 1000 GW (1 TW), and nine hot spots with powers greater than 1 TW.The average hot spot thermal emission is 168 GW, with the large standard deviation of 607 GW expected from thermal emissions spanning so many orders of magnitude.
We note that "one-temperature, one-area fits" to spectra may underestimate the total thermal emission, as the total emitted power from a volcano is the result of combining thermal emission from multiple areas at different temperatures.Nevertheless, a onetemperature, one-area fit is generally robust to within a factor of 2 (Davies & Veeder 2023) and often, based on the specific style of volcanic activity, may yield the same power output as a more sophisticated two-temperature, two-area thermal model fit.This occurs at volcanoes where there is a high effective temperature, typically above ∼800 K. Given the JIRAM hot spots to which this one-temperature, one-area model is applied and the relatively small power outputs that result, the overall conclusions of this analysis are not significantly changed when considering a factor of 2 increase in thermal emission from these hot spots.
In the Davies et al. (2024) analysis of JIRAM 4.8 μm spectral radiance, an empirical relationship derived by Davies & Veeder (2023) from the analysis of Galileo NIMS data was used to estimate the total thermal emission for JIRAM-detected hot spots.The resulting power outputs were on occasion greater by more than a factor of 2 than those subsequently derived from L-band/M-band ratios, yet the results still supported a global magma ocean and asthenospheric model over deep-mantle heating, as does the analysis in this paper.
Table 4 shows hot spot numbers and total power output by hemisphere.Hot spot number alone shows more hot spots on the anti-Jovian hemisphere than on the sub-Jovian hemisphere, outside the standard deviation of the average.Leading, trailing, northern, and southern hemispheres show almost exactly the same number of thermal sources.
In terms of emitted volcanic power, the anti-Jovian hemisphere, while having a greater number of hot spots, outputs only 75% (25 TW) of the power emitted from the sub-Jovian hemisphere (33 TW).

Global Volcanic Heat Flow
Veeder et al. (1994) estimated that Io's total thermal emission was 105 ± 12 TW, derived from an analysis of multiyear, multiwavelength observations of Io's infrared spectral radiance as measured by the IRTF.Of this total, Veeder et al. (2015) estimated that Io's volcanoes contributed 54.5 TW.Table 5 shows the revised volcanic heat flow after updating some entries in Veeder et al. (2015) and adding new detections and some improved values from telescope and JIRAM observations.The revised total thermal emission is 57.7 TW, an increase of 3.1 TW.

Longitudinal Distribution
Table 5 shows how volcanic thermal emission is distributed in terms of longitude, using 30°bins extending from pole to pole.Using updated or new estimates of individual hot spot thermal emission produces similar results to previous analyses (Davies et al. 2015;Veeder et al. 2015; de Kleer & de Pater 2016a), as shown in Figure 5. Like the previous analyses cited above, we find peaks in thermal emission at around 310°W, even when excluding Loki Patera, and a broader peak between 75°W and 120°W.A secondary peak is found centered on 225°W.Gemini/Keck and JIRAM observations added new detections mostly between longitudes 340°W and 160°W on Io's sub-Jovian/leading hemisphere.Also as previously found, there is a distinct minimum in thermal emission around the sub-Jovian point, despite many new detections falling into this longitudinal range -most of these hot spots are relatively small in terms of thermal emission.Power output around 310°W is shown without Loki Patera, which is shown separately-inclusion raises the bin value to nearly 18 TW.The new longitudinal power distribution analysis is in broad agreement with previous analyses.

Heat Flow Distribution with Latitude
Table 6 and Figure 6 show the power distribution as a function of latitude as well as where changes occurred as the result of adding recent hot spot identifications.Gemini/Keck and JIRAM observations added most new energy at midlatitudes.Table 7 shows the distribution of hot spots by polar cap and lower latitudes, as well as the thermal emission and volcanic heat flow per unit area.These numbers and their implications are discussed below.JIRAM added numerous new detections in Io's polar regions (Davies et al. 2024).The discrepancy noted by Davies et al. (2024) between the north and south poles is evident, with north polar cap volcanoes cumulatively emitting twice as much total energy as south polar cap volcanoes.The longitudinal band including Loki Patera accounts for nearly 20 TW.
Table 7 also contains a line of results that includes a revised thermal output from Dazhbog Patera.Located at 301.5°W, 55°N, this is a large patera with a floor area of 8783 km 2 (Veeder et al. 2012).Thermal emission indicative of ongoing volcanic activity was detected from Dazhbog Patera by Galileo PPR (Spencer et al. 2000;Rathbun et al. 2004), Keck (de Pater et al. 2004;Marchis et al. 2005;Cantrall et al. 2018), and New Horizons (Spencer et al. 2007).Keck 4.8 μm spectral radiances were 14 ± 2.5 GW μm -1  ), which yields an effective temperature of the patera floor of 235 K and a total power output of 1508 GW.This has a noticeable but small effect on total regional thermal emission, average power per hot spot, and energy emitted per unit area for low-latitude volcanoes, insufficient to affect our conclusions.

Polar Regions Comparison with Lower Latitudes
A comparison of the distribution of volcanic power between polar regions and lower latitudes and between the two polar regions is shown in Table 7.The results closely follow those derived from 4.8 μm spectral radiance (Davies et al. 2024).The hot spot density-the number of hot spots per unit area-is about the same regardless of latitude, with slightly lower numbers in the polar regions.However, in both polar caps, hot spots emit, on average, less energy than those at lower latitudes by a factor of about 3, and north polar cap volcanoes emit about twice as much energy as south polar cap volcanoes.In terms of hot spot power density-the energy emitted per unit areanorth polar volcanoes have a power density about half that at lower latitudes and twice that of the south polar region.

Correlation with Heat Flow Models
Figure 7 shows the hot spot thermal emissions overplotted on end-member heat flow models.Spearman rank correlation coefficients between quantified thermal emission from 343 sources and heat flow models are shown in Table 8.Correlations between hot spot number distribution and endmember surface heat flow models are shown in Table 9.Values of Spearman ranked correlation range from 1 (very strong correlation) to −1 (very strong anticorrelation), with no correlation returning a value of 0. As in the Io global volcanic heat flow map published in Davies et al. (2015), which used the hot spots identified in Veeder et al. (2015), 15°longitude and latitude bins are used.To ensure accurate correlations, as bin areas get smaller as latitude increases, binned heat flow is converted into heat flow per unit area (W m −2 ) and compared with the expected heat flow in that bin (also in W m −2 ) expected from the end-member models shown in Figure 7.
Global correlation results broadly confirm what was already inferred from the longitudinal and latitudinal distribution of both volcanic heat flow and hot spot numbers across Io.While the highest correlations are for the magma ocean and asthenospheric models, the correlation values are not high.There is a strong anticorrelation with the deep-mantle heating model.The inclusion of additional sources has actually decreased the correlations from those using the Davies et al. (2015) hot spot thermal emission values.Regarding hot spot numbers, the correlations are stronger, showing some preference for the magma ocean and asthenospheric heating models.There is a stronger anticorrelation with the deep-mantle model.Testing a uniform heat flow and a uniform distribution of hot spots returned values of 0, showing no correlation.
Might there be stronger correlations on a hemispherical basis?Results are shown in Tables 8 and 9.In terms of heat flow (Table 8), the strongest correlations are found for the magma ocean and asthenospheric models on the leading hemisphere, followed by the correlations for the same heat flow models on the trailing hemisphere, although the largest correlation values are still weak for both the magma ocean model (0.4694) and the asthenospheric heating model (0.3790).The hemispherical pairing showing the largest difference is the sub-Jovian/anti-Jovian pairing, with the sub-Jovian hemisphere returning the highest correlation coefficient (albeit a weak 0.3605 for the magma ocean model).The difference in total power output between these two hemispheres is ≈8 TW (Table 4).Nevertheless, the distribution of volcanic heat flow suggests a dichotomy between sub-Jovian and anti-Jovian hemispheres.Considering the leading/trailing hemispheres, the difference power output between hemispheres is a considerable 12.2 TW, suggesting a leading/trailing dichotomy in volcanic  2) to the thermal emission curve published by Veeder et al. (2015).Most added thermal emission is at latitudes below 60°.advection.The apparent dichotomy between the north and south polar regions (above 60°latitude) and between the polar regions and lower latitudes quantified by Davies et al. (2024) appears in the full hemispherical numeration only for the magma ocean model (Table 7).The northern hemisphere outputs ≈10 TW more energy than the southern hemisphere, with the caveat that the difference in energy output could be entirely attributed to the presence of Loki Patera (which has an assigned value of 9.6 TW).
In terms of hot spot numbers, the highest correlations are found for the leading hemisphere for both the magma ocean model (0.6072) and the asthenospheric heating model (0.6069), and the biggest difference in correlation between the magma ocean model and asthenospheric model is between the leading (0.6072) and trailing (0.3913) hemispheres.The northern/ southern hemisphere coefficients are more or less the same, given that the number of thermal sources in these hemispheres is almost identical (Table 4).
The amplitude of Io's tidal deformation is strongly related to the rheological properties of Io's interior and to the degree of melting (Kervazo et al. 2022).Calculation of the distribution of tidal heating between the mantle and the asthenosphere is highly sensitive to the assumed melt fraction in the asthenosphere to the extent that for a melt fraction smaller than a critical value that corresponds to a transition from a soliddominated behavior to that of a liquid-dominated one, dissipation is confined mostly to the mantle (Kervazo et al. 2022).As described above, tidal heating in the deep mantle results in polar regions that are preferentially heated, and tidal heating in the asthenosphere results in equatorial regions that are preferentially heated.If the partial melt fraction in the asthenosphere is larger than ∼30%, dissipation associated with volume changes modifies the surface tidal heating pattern, but equatorial regions remain preferentially heated (Kervazo et al. 2022;Matsuyama et al. 2022).
The results presented in this paper have implications for Io's interior structure.The observed pole-to-pole thermal emission is consistent with a dissipation of about 75% in the asthenosphere and the rest in the deep mantle, which is broadly consistent with previous modeling (e.g., Tackley 2001;Hamilton et al. 2013), which suggested a partition of two-thirds shallow heating, one-third deep heating.Applying the model of Kervazo et al. (2022), this partitioning of heating location would imply a silicate viscosity for the deep mantle of the order of 10 19 Pa s and a melt fraction of about 20%-30% in a 50-100 km thick asthenosphere.

Discussions
The heat emitted from active and recently active surface volcanism is not distributed evenly across Io.This was already known (Davies et al. 2015;Veeder et al. 2015;de Kleer et al. 2016ade Kleer et al. , 2019b)), but now we have a more complete robust data set that includes long-sought-after polar infrared hot spot detections.As things have it, the additional detections do not change the previous analyses very much, adding 3.1 TW to Io's active surface volcanism heat flow component, making up about 56% of Io's total thermal emission.This is offset somewhat by a decrease of some 2.5 TW in thermal emission from Dazhbog Patera.Estimates of Io's global mean surface heat flux range from 1.5 to 4.0 W m −2 (e.g., Moore et al. 2007), with more recent observations supporting a value of 2.24 ± 0.45 W m −2 (Lainey et al. 2009), close to that of Veeder et al. (1994;≈2.5W m −2 ).This heat flow is far from uniform in distribution.Io appears to be an exemplar of heat pipe volcanism, where heat is transferred from Io's interior to the surface by the movement of magma at active volcanoes (Moore 2001).Away from volcanoes, there should be no observable endogenic heat flow because Io's resurfacing rate is faster than heat can be conducted, effectively burying heat.Based on the quantification of thermal emission from all sources identified over more than 30 yr, this is far from the case.Accounting in the most liberal way for volcanic heat flow from active or recent sources, including the sources newly identified in Juno data (updating Davies et al. 2024), yields 58 ± 1 TW of Io's 105 ± 12 TW of endogenic heat (Veeder et al. 1994).The distribution of the remaining ∼47 TW of Io's observed thermal emission-a truly vast amount of energy, equivalent to the Earth's total endogenic heat flow (47 ± 2 TW; Davies & Davies 2010)is unknown.It is likely that this thermal emission component originates from near-surface intrusive activity (Veeder et al. 2012;Spencer et al. 2020;Steinke et al. 2020;Spencer et al. 2021) or from shallow, recently buried volcanic deposits.It is also expected that the distribution and magnitude from these sources are a function of local volcanic activity, which, in turn, is a function of internal heating pattern.If this is the case, then the mapping of this "background" heat flow distribution may be the only way to determine if current heat flow models are indeed applicable to Io.The possibility cannot be ruled out that such mapping will show a strong correlation between global heat flow and one of the models, which would then open up a new line of investigation as to why heat flow from active volcanoes does not follow the trend.
Prior to the Juno mission, Io's polar volcanoes were thought to be few in number, with eruptions more episodic in nature than at lower latitudes.Polar volcanoes were thought to be larger in terms of thermal emission (and, by inference, volumetric effusion rate) and with possibly higher-temperature lava (ultramafic).Previous analyses (Milazzo et al. 2005) found that Io's polar hot spots known at that time had half the number per unit area (1 per 5.1 × 10 5 km 2 ) than found for Io as a whole (1 per 2.5 × 10 5 km 2 ) and those located at lower latitudes (below 60°), determined to be (1 per 2.3 × 10 5 km 2 ).We find very different results based on total power output that match the results of examining 4.8 μm spectral radiance (Davies et al. 2024) in that the distribution of volcanoes by number is about the same regardless of latitude, and Io's polar cap volcanoes are less energetic, on average, than volcanoes at lower latitudes.The hypothesis of "more heat flow is the result of more regional heating" fails, as Io's poles appear to be anomalously warm (Rathbun et al. 2004).Either the implied (but unproven for Io) link between enhanced volcanic activity and enhanced heat flow is incorrect or the warm poles are instead caused by some local physical property of the polar region surface material (such as a high thermal inertia; e.g., Rathbun et al. 2004).
JIRAM data of volcanic thermal emission suggest that Io has a global magma ocean in slight preference to relatively shallow heating or a global magma ocean.There is a strong  Kervazo et al. (2022) and the interior partition of tidal energy and rheology this analysis suggests, the expected value of k 2 would be between 0.08 and 0.1.
It is likely that an Io-dedicated mission such as the Io Volcano Observer (McEwen et al. 2014; McEwen & The IVO Science Team 2020), designed for Io's specific environment and unique challenges (such as imaging the surface with a high dynamic range of temperatures from <100 K to >1400 K), would be needed to map Io's background thermal emission and refine measurements of thermal emission distribution.Additional modeling of Io's interior is needed.The controls on lithospheric thickness and structure and how these may be changing across Io need to be better understood, as is the movement of magma in a global or partial magma ocean.

Conclusions
Io's volcanic activity, in terms of power output and distribution of currently or recently active locations, is far from uniform, as others have discovered previously (Veeder et al. 2012;Hamilton et al. 2013;Veeder et al. 2015;de Kleer et al. 2019b).This analysis of total power output from Io's hot spots supports the polar dichotomy found from examination of 4.8 μm spectral radiance found in JIRAM data (Davies et al. 2024), with additional possible sub-Jovian/anti-Jovian and leading/trailing dichotomies as well.Io's interior and volcanic advection is more complex than allowed by current models.
Despite the relatively low correlations, comparison of the distribution of thermal emission from currently active volcanoes on Io with proposed models of internal heat flow shows a slight preference for a global magma ocean model over a heat flow model dominated by shallow asthenospheric tidal heating.
The thermal emission from active volcanoes shows a negative correlation with a deep-mantle tidal heating model.Results are consistent with prior analyses using estimates of total thermal emission.We find that using hot spot numbers, rather than volcanic thermal emission, does not allow differentiation between magma ocean and asthenospheric models.It is becoming clear that Io's heat flow is not well served using these models; that Io's interior heating is more complex than previously thought, likely involving a global or partial magma ocean; and that apparent structural or compositional dichotomies exist between sub-and anti-Jovian hemispheres, between polar regions and lower latitudes, and between the north and south polar regions.
In short, the same trends are found with hot spot total power output estimates as with the 4.8 μm spectral radiance analysis of Davies et al. (2024), namely, 1. the hot spot density does not vary greatly between the polar regions and lower latitudes, 2. the polar regions combined emit only ∼one-third of the energy per unit area seen at lower latitudes, 3. the north pole alone emits about 44% less energy per unit area than that emitted at lower latitudes, 4. the north polar region emits twice as much energy per unit area as the south polar region, and 5. the south polar region emits only one-fifth of the energy emitted per unit area at lower latitudes.
The distributions of volcanic heat flow and hot spots are not fitted well with any current model of tidal heating and volcanic advection.Measurement of the distribution of "background" thermal emission (that is, not obviously emanating from current or recent volcanic activity) is a crucial measurement that would provide further constraints on future modeling of Io's interior.As it currently stands, the new map of Io's volcanic thermal emission nevertheless is an important boundary condition that models of Io's heat flow need to replicate.

Figure 1 .
Figure1.The updated Io volcanic thermal emission map, combining nearly 30 yr of observations.This Mollweide projection is centered on the anti-Jovian point (180°W).Some prominent volcanoes are labeled.North is up.

Figure 2 .
Figure2.Orthographic projections of the volcano distribution and estimated total hot spot heat flow in Io's northern hemisphere (left) and Io's southern hemisphere.As a suitable heat flow model discriminator, 60°latitude is used to demarcate Io's polar regions.This latitude was chosen because this was where the greatest changes in heat flow occurred when comparing models, and previous works used this latitude(Milazzo et al. 2005;Davies et al. 2024) to demark the polar boundary, allowing direct comparison of results.

Figure 3 .
Figure 3. Io's volcanic heat flow pre-Juno: the thermal emission as a function of longitude for 242 dark paterae, dark flows, and other volcanic features on Io as determined byVeeder et al. (2015).The long arrows indicate the longitudes of maximum heat flow predicted using an asthenospheric heating model(Segatz et al. 1988;Ross et al. 1990).The shorter arrows are the locations of secondary maxima.30°bins are used.Loki Patera (not binned) is plotted to show its position and average thermal emission (9600 TW;Matson et al. 2006).The total thermal emission shown excludes outbursts.Figure fromDavies et al. (2015).The data here compare well with a plot of longitudinal power output derived from Keck and Gemini telescope data (de Kleer & de Pater 2016a; see below).

Figure 4 .
Figure 4. Hot spot detections cataloged by Veeder et al. (2015) and mapped by Davies et al. (2015) are shown as white circles.Post-2015 detections, including JIRAM detections up to 2023 March (PJ49, inclusive), are shown as black circles.Portions of hot spots in other colors reflect probable changes of hot spot thermal emission location at a given volcano where the nearest pre-Juno hot spot thermal emission is preferentially used, in most cases where hot spot power is derived from Galileo NIMS spectra.A large cluster of post-2015 hot spots is found in the northern hemisphere at longitudes between 0°W and 60°W, a region poorly covered by Galileo.

Figure 5 .
Figure 5. Longitudinal distribution of volcanic thermal emission from all sources.This figure compares the updated hot spot list incorporating recent AO and JIRAM detections with the updated Veeder et al. (2015) hot spot power list (Table 2) and the distribution mapped by de Kleer & de Pater (2016a) from observations from 2013 to 2015.All three analyses show broad agreement.

Figure 6 .
Figure 6.Latitudinal distribution of volcanic thermal emission from 343 volcanic sources (up to and including JIRAM data obtained through orbit PJ49, 2023 March 1).This figure compares the new global hot spot list (incorporating Keck, Gemini, and JIRAM data analyses and the updated Veeder et al. 2015 hot spot power list using the values in Table2) to the thermal emission curve published byVeeder et al. (2015).Most added thermal emission is at latitudes below 60°.

Figure 7 .
Figure 7. Hot spot total power estimates (transient outburst eruptions are not plotted) in GW derived from nearly 30 yr of observations, including from the Juno spacecraft, overlaid on end-member models of interior heat flow (graphics from Matsuyama et al. 2022).Upper left: I-deep-mantle heating, favoring enhanced polar heating.Upper right: II-shallow, asthenospheric heating, with main heat flow located at sub-and anti-Jovian longitudes and equatorial to midlatitudes.Lower left: III -magma ocean model (Matsuyama et al. 2022), with peak heat flow offset eastward from the asthenospheric model.This Mollweide projection is centered on the anti-Jovian point (180°W).The Io reference image (lower right) is the Voyager-Galileo mosaic (Becker & Geissler 2005).

Table 3
JIRAM Detections Not Plotted as Covered by Previous Analyses a Davies et al. (2024).

Table 7
Distribution of Hot Spots and Total Thermal Emission (Orbits PJ5 to PJ49) Using Maximum JIRAM "Linear Detector Response" Values and 60°Latitude Caps

Table 8
(Janssen et al. 2017)eeder et al. 2015)rved Heat Flow Distribution and End-member Heat Flow Model and a Uniform Distribution Control 242 locations(Davies et al. 2015;Veeder et al. 2015).Spearman Rank Correlation Coefficients Globally and by Hemisphere for Hot Spot Number for Three Heat Flow Models and a Uniform Distribution Control with preference to a heat flow dominated by deep-mantle heating.Because of offsets in the expected peaks and troughs (both longitudinally and latitudinally), observed thermal emission does not match any single heat flow model well.Given the relatively small number of thermally weak polar hot spots, the results from this analysis match previous analyses well.Much is left to be explained.For example, why are Io's polar volcanoes smaller and less powerful than their contemporaries at lower latitudes?It is clear that a more sophisticated model of internal heating, advection, and the subsequent surface expression of heat flow is required to explain observations of thermal emission from Io's active volcanoes.Of great use would be measurements of heat flow from Io's not obviously volcanic surface-the 97% of Io's surface not currently occupied by active or recently active volcanoes, especially at Io's poles.It is possible that Junoʼs Microwave Radiometer(Janssen et al. 2017)will return data at multiple wavelengths of Io's surface structure that will allow derivation of heat flow.Additionally, Juno might obtain data from close Io flybys that will allow measurement of Io's gravitational tidal response, quantified by Love number (k 2 ), which would contribute toward establishing if Io has a global magma ocean (de Kleer et al. 2019b).k 2 places a strong constraint on a planetary body's interior state.A high k 2 would indicate that some part of Io's interior is weak enough to mechanically decouple from the lithosphere, a likely scenario if Io has a global magma ocean.A low k 2 would indicate that Io's interior behaves like a solid.Using the model of Io's interior developed by c anticorrelation