We have calculated the ≈5-μm radiant flux for every volcanic hot spot in every one of the 190 GalileoNear-Infrared Mapping Spectrometer (NIMS) tube observations of Io obtained between 28 June 1996 and 16 October 2001 in order to determine the variability of thermal emission from Io's volcanoes at local, regional and global scales, and to identify individual eruption episodes where thermal emission waxes and wanes. The resulting NIMS Io Thermal Emission Database (NITED) allows the comparison of activity at individual volcanoes and different regions of Io. The database contains over 1000 measurements of radiant flux at approximately 5μm, corrected for emission angle, range to target and incident sunlight (where necessary). We examine the data for Loki Patera, Io's most powerful volcano. For data acquired in local darkness we use two-temperature fits to nighttime spectra and prior knowledge of emitting area to determine total radiated thermal emission. For other data we use the constancy of the integrated thermal emission spectrum to determine total thermal emission from measurements of radiant flux at 5μm. As seen by NIMS, total thermal emission from Loki Patera varies between 7600 GW and 17000 GW. We revise upwards previous estimates of thermal emission from NIMS data. NIMS 3.5-μm radiant fluxes (both measured and estimated) are consistent with measurements from ground-based telescopes. This work highlights the value of NITED as a research tool.
 The volcanoes of the jovian moon Io are compelling targets for missions to the Jupiter system and for ground-based telescopes (see summary in work byDavies  [see also de Pater et al., 2004; Lopes-Gautier et al., 1997; Marchis et al., 2005; McEwen et al., 1997; Veeder et al., 1994]. Lists of Io's active volcanoes have been compiled [Davies, 2007; Lopes and Spencer, 2007], and their locations plotted and compared with models of global heat flow [Lopes-Gautier et al., 1999]. The locations of active volcanoes have also been compared with distributions of mountains and paterae to seek out possible correlations in distribution at different spatial scales [Hamilton et al., 2011; Radebaugh, 2005; Schenk et al., 2001]. The spatial distribution of volcanic centers is a useful indication of underlying heat sources and processes (cf., the Pacific “ring of fire”). However, these Io analyses did not account for the magnitude of thermal emission from individual hot spots. In the hot spot ‘maps’ areal density of volcanic centers does not necessarily equate to heat flow distribution – e.g., a small hot spot (for Io) emitting 1 GW was previously treated identically to a hot spot four orders of magnitude more powerful. The next step in the geophysical interpretation process is to map the distribution of volcanic heat flow. We have been quantifying hot-spot thermal emission in order to determine the distribution of volcanic heat flow and the contribution to Io's thermal budget of heat flow from different eruption types. In order to obtain this snapshot of the magnitude and distribution of Io's thermal emission during theGalileo epoch, Veeder et al. [2009, 2011; Io: Volcanic thermal sources and global heat flow, submitted to Icarus, 2011] quantified the thermal emission from 240 ionian volcanoes, and produced plots of volcanic heat flow as a function of longitude and latitude, as well as heat emanating from lava flows, paterae, and other sources. In this paper we describe the next stage of the thermal emission analysis, the quantification of variability of thermal emission at 4.7 to 5 μm from Io's active volcanoes detected by the Galileo Near Infrared Mapping Spectrometer (NIMS). We show how radiant flux can be estimated for other wavelengths (e.g., 3.5 μm) from estimated total thermal emission and 5-μm radiance data for comparison with other datasets.
2. NIMS Observations of Io
 The GalileoNear-Infrared Mapping Spectrometer was an instrument well-suited to observing thermal emission from ongoing or recent volcanic activity. The NIMS wavelength range (0.7 to 5.2μm) meant that it was sensitive to a wide range of surface temperatures (>1000 K to ≈220 K) and lava surface exposure times [Davies et al., 2010]. The acquisition and processing of NIMS data of volcanic thermal emission, and descriptions of “tube” and “cube” products, are described in detail by Davies [2007, and references therein], but can be summarized as follows. NIMS observations were obtained at a minimum of 8 to a maximum of 408 wavelengths over a wide range of spatial resolutions (hundreds of kilometers per pixel to less than a kilometer per pixel [Davies, 2007, Table 3.2]. Observations after 11 October 1997 were limited to 12 or 15 wavelengths distributed across most of the wavelength range. A total of 190 NIMS tube products were collected of Io. These contain raw radiance data. Some of these tube products contained multiple observations of Io (e.g., the C9INWARMCV01 observation obtained on 27 June 1997 imaged the Janus/Kanehekili region of Io nine times). Temporal resolution was also highly variable. During Galileoorbit E4, for example, 27 global observations were collected between 19:10 UT on 17 December 1996 and 16:23 UT on 19 December 1996. The Loki Patera region was observed 15 times. On some other orbits only single observations of Io were obtained. The tube products were corrected to remove the instrumental spatial response and processed into 181 re-navigated and re-sampled “cube” products. Both are useful because tube radiance products yield the most accurate radiant flux data, whereas cube products yield the most accurate navigation and location data.
3. NIMS Io Thermal Emission Database (NITED)
 Prior to this work, few ionian volcanoes have been examined by utilizing all available NIMS data. Those that were are Prometheus [Davies et al., 2006], Zamama, Culann, and Tupan Patera [Davies and Ennis, 2011]. Only by examining all data are the intricacies of thermal emission revealed, and a deeper understanding of volcanic processes obtained. For example, the analysis of the Prometheus dataset revealed individual episodes of volcanic activity and a strong similarity to effusive activity observed at the terrestrial volcano Kilauea, Hawai'i, although on a much larger scale. Individual episodes at other volcanoes were also observed. In the case of Prometheus, detailed examination of one episode and estimation of the volume erupted led to the conclusion that this and other similar volcanoes on Io are fed from large magma chambers at relatively shallow depths [Davies et al., 2006].
 In order to study a larger set of active volcanoes we have therefore examined the entire NIMS Io dataset of 190 tube radiance products, obtained between 28 June 1996 (Galileo orbit G1) and 16 October 2001 (Galileo orbit I32). Where necessary, cube products were used to obtain accurate locations of thermal sources. The USGS Io mosaic [Becker and Geissler, 2005] was used to identify likely active features. We have measured the thermal emission at 5 μm wherever possible, and at 4.6967 μm for observations obtained on and after 11 October 1999. Data were corrected for incident sunlight where necessary and for emission angle. The resulting database contains over 1000 records of 5-μm flux alone. More than 280 nighttime hot spot spectra were identified, with an additional 32 obtained on or close to the terminator. Data processing methodology, including cosine correction, sunlight removal, and processing of data where the emitting area is resolved, is described by Davies and Ennis . This database allows thermal emission to be examined not only at local levels, extending the work begun at Prometheus (see above), but also on regional and hemispherical scales. This is of particular importance when considering the possible effects of tidal stresses on the timing and magnitude of Io's volcanic activity [e.g., Rhoden and Kite, 2011], and on magma movement through the lithosphere. We illustrate the value of the NITED database here by examining the variability of thermal emission at Loki Patera. A companion paper looks at local and regional thermal emission from other hot spots at Pele, and Janus Patera and Kanehekili Fluctus.
4. Loki Patera
 Loki Patera is Io's most powerful volcano. It's volcanic activity and styles of eruption and resurfacing are summarized by Davies . Variation in ≈3.5-μm radiant flux from mostly ground-based observations has been charted [e.g.,Rathbun and Spencer, 2006] and suggests that Loki Patera is a large lava lake [Rathbun et al., 2002]. The surface temperature distribution as seen by NIMS also suggests a cool, stable crust like that found on a quiescent lava lake [Davies, 2003]. Resurfacing by lava flow emplacement [Davies, 2003] is less likely [Matson et al., 2006]. The average modeled thermal emission from Loki Patera is ≈9600 GW [Matson et al., 2006] from an emitting area of 2.15 × 104 km2 [Veeder et al., 2011]. Previous estimates of thermal emission from a selection of NIMS nighttime observations of Loki Patera were published by Matson et al. . We present more comprehensive estimates of thermal emission from these and other nighttime observations of Loki Patera (Table 1). 35 NIMS Loki Patera observations were analyzed. 6 additional observations are not included (see Table 1 footnotes). Table 1 includes estimated values of total thermal emission (Qrad) and 3.5-μm radiant flux.
Table 1. Loki Patera NIMS Observations and Estimates of Total Thermal Emission and 3.5-μm Radiant Fluxes
Excluded observations: G8INTHRMAL04, 24INLOKIRA01 and 32INTHLOKI01 are partial observations. With E4INCOOLCV02C and E4INRTMON02R there is uncertainty as to whether the source is Loki Patera. E6INHRSPEC01 was obtained at a very high emission angle (>80°).
Given in format mm/dd/yy hh:mm.
Radiant flux corrected for distance and emission angle (see Davies et al.  for the methodology employed here).
This is the temperature at which the entire emitting area of Loki Patera (2.15 × 104 km2) has to be in order to produce the radiant flux in column 6 at the wavelength in column 5.
Total thermal emission Qradfrom the dark, low-albedo emitting area of Loki Patera (2.15 × 104 km2). Numbers in bold are from two- temperature, two-area fits to nighttime spectra (seeTable 2). Other values of Qrad are calculated using equation (3).
3.5-μm radiant fluxes in bold are from two-temperature, two-area fits to nighttime spectra (seeTable 2). Other values are calculated using equation (4).
Matson et al. fitted NIMS spectra with a two-temperature, two-area model [Davies et al., 1997], and the total thermal emission from these areas was calculated. From the shape of the thermal emission spectrum (e.g., Figure 1) it is clear that these model fits were to data at shorter wavelengths than the peak of thermal emission from Loki Patera, which meant that the position of the peak and magnitude of the longer-wavelength thermal emission were poorly constrained. The sum of the two areas derived byMatson et al.  for each observation was therefore smaller than the total emitting surface, the dark area of the patera floor (2.15 × 104 km2) measured by Veeder et al. . We use this area as an additional constraint for the two-component fit to produce more accurate estimates of integrated thermal emission (Qrad) and 2- and 3.5-μm radiant fluxes for daytime observations.
 Radiant flux as a function of wavelength is found by fitting the NIMS spectra with the function
where Iλ is radiant flux (W μm−1) at wavelength λ (μm) from a two temperature fit where areas A1 and A2 (m2) are at temperatures T1 and T2, (K), c1 = 3.7413 × 108 W μm4 m−2 and c2 = 1.4388 × 104μm K. A and T values are shown in Table 2. Emissivity ε is taken to be 1. An additional constraint is
The Loki Patera thermal emission spectrum does not greatly change in shape even as resurfacing is taking place, although the magnitude of thermal emission varies greatly [Davies et al., 2010; Matson et al., 2006]. Figure 2shows the relationship between 5-μm radiant flux and total thermal emission. This relationship allows us to estimate total thermal emission (Qrad, GW) just from a measurement of 5-μm radiant flux (Iλ,5 μm) and the best-fit slope inFigure 2, where
From a plot of 3.5-μm radiant flux versus 5-μm radiant flux (as derived from the observations in Table 2) the relationship
is used to estimate 3.5-μm radiant flux (Iλ,3.5μm), from observations of 5-μm radiant flux. The 3.5-μm radiant flux is a useful quantity for comparison of NIMS data with the extensive ground-based dataset [e.g.,Rathbun and Spencer, 2006].
Table 2. Nighttime Observations of Loki Patera and Results of Two-Temperature, Two-Area Fits to NIMS Spectraa
 From the two-temperature, two-area fits (seeTable 2), total thermal emission (Qrad) varies between a low of 7599 GW (12 August 1999) and a high of 16159 GW (28 June 1997). An interesting point is that even though the E16 data (20 July 1998) were obtained at a high emission angle (77.8°) and are cosine corrected, the resulting radiant flux is the middle of the range of estimated Qrad. In other words, the assumption that thermal emission is from a flat plate tilted away from the observer works very well at Loki Patera, adding to the argument that the patera floor is very smooth and increasing the likelihood that this is indeed the crust on a low-viscosity silicate lava lake. The relatively quiescent resurfacing process and its effect on the shape and variability of the integrated thermal emission spectrum of Loki Patera are discussed in detail byMatson et al.  and Davies et al. . The resulting constancy of Loki Patera's spectral shape is what allows estimation of thermal emission at other wavelengths, having determined 5-μm radiant flux and Qrad.
 Eleven observations of Loki Patera were obtained at ranges of ∼9 × 105 km between 15:56 UT and 16:23 UT on 19 December 1996 during Galileo orbit E4. Emission angles were between 25.5° and 24.1°. Data were obtained at 10 wavelengths (out of a possible 408). Thermal emission at 5.2101 μm was steady, with an average radiant flux of 194 ± 15 GW μm−1.
 Another interesting sequence of observations was obtained between 26 June and 28 June 1997 (Galileoorbit C9). Six observations were obtained on 26 June at high emission angles of ∼77°. Thermal emission was again steady with an average 4.9951-μm radiant flux of 440 ± 39 GW μm−1. On 28 June another observation was obtained at an emission angle of 30° and yielded a 5-μm radiant flux of 620 GW μm−1. Either there was an eruption exposing enough material to increase the 5-μm thermal emission, or the high emission angle during the earlier observation led to topographic shielding of part of the patera floor. The latter scenario is more likely as the 5-μm radiant fluxes in May 1997 (Galileo orbit G8), from observations at low emission angles, were between 594 and 688 GW μm−1.
 The values of Qrad in Table 1 average 11500 ± 2900 GW, and range from a minimum of 7600 GW (12 August 1999) to a maximum of 17000 GW (7 May 1997). Figure 3shows the variability in 3.5-μm radiant flux, using the values calculated using equation (4)for daytime observations. Also plotted are ≈3.5-μm radiant fluxes from ground-based observations [Rathbun and Spencer, 2006].
5. Discussion and Conclusions
 We have quantified the 4.7 or 5-μm thermal emission for every volcanic source in every observation of Io obtained by Galileo NIMS. The resulting NIMS Io Thermal Emission Database (NITED) is a valuable resource that allows volcanic activity at different locations and different regions to be plotted as a function of time. Individual eruption episodes can be identified, and the mode of eruption can often be constrained. If additional information is available for the area that is thermally active [i.e., Veeder et al., 2009, 2011, submitted manuscript, 2011], then estimates of total thermal emission can be made. In this way, the contribution to Io's volcanic thermal emission heat flow from individual volcanoes can now be quantified. The development of NITED continues with plans for testing and validation, the expansion of entries taken from published sources, and the addition of measurements of thermal emission from ground-based instruments [e.g., seede Pater et al., 2004; Marchis et al., 2005, and references therein] and from other spacecraft (Voyager, Cassini, New Horizons) to enable the fullest possible pictures of volcanic thermal emission and variability. NITED data can now be selected to quantify thermal emission from a group of volcanoes or a specific region of Io and will be used to test models of internal heating and resulting energy transport to the surface. Rhoden and Kite  have used the NITED Loki Patera data to test a model of tidal control on volcanic activity, concluding that this is unlikely (at least at Loki Patera).
 This work was conducted at the Jet Propulsion Laboratory, California Institute of Technology, under contract with NASA, and is supported by the NASA OPR Program.
 The editor thanks the two anonymous reviewers for their assistance in evaluating this paper.