Observations of the north polar water ice annulus on Mars using THEMIS and TES

ARTICLE IN PRESS

Planetary and Space Science 56 (2008) 256–265 www.elsevier.com/locate/pss

Observations of the north polar water ice annulus on Mars using THEMIS and TES Kiri L. Wagstaffa,, Timothy N. Titusb, Anton B. Ivanova, Rebecca Castan˜oa, Joshua L. Bandfieldc a

Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA b United States Geological Survey, 2255 N. Gemini Dr., Flagstaff, AZ 86001, USA c Department of Geological Sciences, Arizona State University, Tempe, AZ 85287-6305, USA Available online 31 August 2007

Abstract The Martian seasonal CO2 ice caps advance and retreat each year. In the spring, as the CO2 cap gradually retreats, it leaves behind an extensive defrosting zone from the solid CO2 cap to the location where all CO2 frost has sublimated. We have been studying this phenomenon in the north polar region using data from the THermal EMission Imaging System (THEMIS), a visible and infra-red (IR) camera on the Mars Odyssey spacecraft, and the Thermal Emission Spectrometer (TES) on Mars Global Surveyor. Recently, we discovered that some THEMIS images of the CO2 defrosting zone contain evidence for a distinct defrosting phenomenon: some areas just south of the CO2 cap edge are too bright in visible wavelengths to be defrosted terrain, but too warm in the IR to be CO2 ice. We hypothesize that we are seeing evidence for a seasonal annulus of water ice (frost) that recedes with the seasonal CO2 cap, as predicted by previous workers. In this paper, we describe our observations with THEMIS and compare them to simultaneous observations by TES and OMEGA. All three instruments find that this phenomenon is distinct from the CO2 cap and most likely composed of water ice. We also find strong evidence that the annulus widens as it recedes. Finally, we show that this annulus can be detected in the raw THEMIS data as it is collected, enabling future long-term onboard monitoring. r 2007 Published by Elsevier Ltd. Keywords: Mars polar caps; Water ice; Seasonal ice; THEMIS; TES

1. Introduction CO2 in the Martian atmosphere undergoes an annual cycle near the poles that consists of condensation onto the polar cap in the fall and sublimation back into the atmosphere in the spring. As a result, large amounts of CO2 cycle from the north pole, during the northern winter, to the south pole, during the southern winter, and back (Kieffer, 1979). This effect is significant enough that it impacts both the atmospheric pressure (Hess et al., 1980) and the distribution of surface mass on the planet (Folkner et al., 1997; Smith and Zuber, 2002). Smith and Zuber (2002) combined topography observations from the Mars Orbiter Laser Altimeter (MOLA) with X-band Corresponding author.

E-mail address: [email protected] (K.L. Wagstaff). URL: http://www.wkiri.com (K.L. Wagstaff). 0032-0633/$ - see front matter r 2007 Published by Elsevier Ltd. doi:10.1016/j.pss.2007.08.008

radio tracking of the Mars Global Surveyor spacecraft and determined that seasonal cycling of volatiles caused a 3% change in amplitude and a 22% change in phase for the gravitational field coefficients that specify the position of the center of the core–mantle component of Mars. We have been studying the northern seasonal cap using infra-red (IR) observations from the THermal EMission Imaging System (THEMIS) on the Mars Odyssey spacecraft (Christensen et al., 2003). Most THEMIS images that contain the edge of the CO2 cap consist of temperatures that are bimodally distributed between values compatible with the presence of CO2 ice or frost and values compatible with defrosted terrain. However, we have identified several exceptions to this trend, in which the temperatures are instead divided into three distinct groups: a cold CO2 region, a warmer defrosted region, and an intermediate region that is too warm to be CO2 but too cold to be

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

regular defrosted regolith. It was previously proposed that there might be an annulus of water ice just south of the seasonal CO2 cap’s recession (Houben et al., 1997). The Thermal Emission Spectrometer (TES) on Mars Global Surveyor has seen evidence for a water ice annulus based on IR and visible observation (Kieffer and Titus, 2001), and OMEGA, the Visible and Infrared Mineralogical Mapping Spectrometer on Mars Express, has recently identified water ice in similar locations via spectral measurements (Bibring et al., 2005). The major contribution of this work is the description, analysis, and discussion of the first THEMIS observations known to contain evidence for a seasonal water ice annulus in the north polar region. The THEMIS data is particularly valuable because it has significantly higher spatial resolution (100 m per pixel) than either TES (3 km per pixel) or OMEGA (300 m to several km per pixel). In addition, THEMIS orbits Mars 12 times per sol, providing ample opportunities for studying the polar region. Most importantly, since TES is no longer collecting spectral data, it is important to find ways to continue monitoring the polar caps using THEMIS. The TES bolometer could also be used to track the polar cap, but it too has lower spatial resolution than THEMIS, and its temperature readings are more affected by dust in the atmosphere than THEMIS band 9 observations are. In this paper, we first summarize what is known about seasonal water ice in the north polar region of Mars (Section 2). In Section 3, we present our observations of the location and extent of the water ice annulus during northern winter and early spring. Next, we compare our observations with contemporary data from the TES on Mars Global Surveyor and find good agreement between the instruments (Section 4). A discussion of the interpretation of our results and their implications appears in Section 5. We conclude with a summary of our findings and some recommendations for future work that can further add to our understanding of the seasonal water ice annulus on Mars.

257

2. Seasonal water ice in the north polar region The north pole of Mars is covered by a residual ice cap that is believed to be composed primarily of water ice, due to observations of the column abundance of water vapor and thermal emission measurements of the cap temperature (Kieffer et al., 1976; Jakosky, 1985). This cap extends as far south as 80 N and is mostly circular, although not perfectly so (Tanaka and Scott, 1987), as shown in Fig. 1a. During northern fall and winter, temperatures at the north pole drop low enough that CO2 condenses out of the atmosphere and forms a seasonal ice cap (see Fig. 1b) that covers the residual water ice and grows southward to about 53 N (Kieffer and Titus, 2001). During northern spring, the CO2 cap sublimates and recedes. Smith et al. (2001) analyzed data from MOLA and found that the north seasonal cap reached a maximum of 1:5  0:25 m in thickness. The CO2 cap edge can also be detected in neutron spectrometer readings (Feldman et al., 2003; Prettyman et al., 2003; Litvak et al., 2004) and in TES observations (Kieffer and Titus, 2001; Titus, 2005). However, CO2 is not the only volatile currently involved in seasonal changes near the poles. Jakosky and Farmer (1982), using observations from the Viking Mars Atmospheric Water Detector (MAWD), observed that the water vapor column abundance above the cap area increased well before the residual cap was exposed by the CO2 cap recession. In fact, they found that at least 60% of the total increase in atmospheric water vapor appeared during this period. They identified the source of this water vapor as either ‘‘the seasonal cap or regolith’’ (Jakosky and Farmer, 1982). Smith (2002) reported on TES observations of water vapor on Mars over a full Martian year (March 1999–March 2001) and found some agreement with the MAWD results but also some significant differences, particularly in the southern hemisphere. Tamppari et al. (2003) noted that differences between MAWD and TES water vapor column abundances could be due to a seasonal coverage bias in the MAWD observations.

Fig. 1. The residual and seasonal north polar caps on Mars, as observed by the Mars Orbiter Camera (MOC). Image credits: NASA/JPL/Malin Space Science Systems. (a) Residual cap (Ls 109:18 ); (b) Seasonal cap (Ls 6–20 ).

ARTICLE IN PRESS 258

K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

3. THEMIS Observations of the water ice annulus We studied the edge of the northern seasonal cap to determine if we could find evidence for the presence of water ice in THEMIS IR data. The advantage of using THEMIS to study this phenomenon is its higher spatial resolution (100 m/pixel), which is 3–30 times higher than OMEGA and 30 times higher than TES. At this resolution, we cannot only identify transitions between different surface constituents but also determine whether the transition is sharp or gradual. Since water ice, CO2 ice, and ice-free regolith exist at very different temperatures, we examined THEMIS band 9 observations, at 12:57 mm, where the instrument has the greatest signal to noise ratio and is most sensitive to surface temperatures for the temperature ranges of surface volatiles. Fig. 2. Seasonal cap edges in the north polar region as observed in TES data at Ls 23 . The red contour is the edge present in IR data (CO2 ), and the blue contour is the edge in VIS (visible wavelength) data. The underlying colored data shows TES brightness temperature for reference. From Titus (2005).

Houben et al. (1997) were the first to suggest that an observable water ice annulus might be present just south of the receding seasonal cap, based on predictions from the best-fit Mars climate model. They suggested that water ice could sublimate along with the CO2 frost and recondense further north, thus gradually stepping towards the residual cap. James and Cantor (2001) studied the northern seasonal cap in images taken by the Mars Orbiter Camera (MOC), in which both water and CO2 ice are indistinguishable. A water ice annulus, if present in the form of snow or frost, would extend the observable ‘‘edge’’ of the seasonal cap beyond that observed in IR observations, which captures the CO2 ice only. Indeed, the cap edge they observed was consistently farther south than that seen in TES IR data (Kieffer and Titus, 2001). Fig. 2 shows the two cap edges as observed in IR and visible (VIS) data from TES during northern spring. In addition, observations of a small amount of water ice, probably mixed with dust, that follows the seasonal CO2 cap recession have been confirmed by the OMEGA instrument on Mars Express (Bibring et al., 2005). This study examined spectral information (0.35 to 5:1 mm) to determine the nature of the region just south of the edge of the seasonal cap. OMEGA has a spatial resolution that varies from about 300 m to a few kilometers, depending on where the spacecraft is in its elliptical orbit. A more recent analysis of TES data shows that water ice can persist at a given location for 10–45 sols (Titus, 2005). Modeling by Forget (2005) and Montmessin et al. (2004) provides additional theoretical foundation for the presence of seasonal water ice near the pole.

3.1. THEMIS data set We examined a total of 197 images from a THEMIS imaging campaign recently mounted specifically to target the edge of the seasonal cap during northern winter and spring. These images range from Ls 350 (late winter) to 70 (late spring), with center latitudes tracking the expected position of the CO2 cap. The longitude coverage is fairly even, though there is a slight bias toward longitudes 300–30 (see Fig. 3). Each IR image is accompanied by another image taken at visible wavelengths and a higher spatial resolution (18 m per pixel). However, full-resolution visible images are restricted to 65.6 km in length, while the IR images can cover anywhere from 27.2 to 6530 km. As a result, the VIS images do not always overlap with the actual cap edge, where it would be most useful for our purposes. In this study, we chose to work with the raw, uncalibrated (EDR) THEMIS data. The calibrated (RDR and BTR) data use a transformation that provides accurate 90 120

15

60

10

150

30

5 5 180

0

210

330

240

300 270

Fig. 3. Distribution of the longitudes of the 197 THEMIS images used in this study. A slight bias toward longitudes 300–30 is present.

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

x 105

259

x 105

7

6

6

5

5 4

Counts

Counts

4

3

3

2 2 1

1 0 120

140

160

180

200

220

240

260

280

Temperature (K)

0 120

140

160

180

200

220

240

260

280

Temperature (K)

Fig. 4. Temperature histograms for two example images, one from each group of THEMIS north polar images. The trimodal images in part (b) are the subject of this paper. (a) Two modes (155 images); (b) three modes (42 images).

temperatures for median Mars temperatures but less accurate values for north polar images. These inaccuracies are due to the specific observing conditions under which the north polar images were collected. First, there is significant temperature drift of the focal plane array (FPA) due to the extreme length of these images (14; 400 lines). Mars Odyssey is in a near-polar orbit, and these images are collected as the spacecraft sweeps up over the north pole and down the daytime side of the planet, crossing the cap edge and finally encountering defrosted terrain. The UDDW (‘‘undrift and de-wobble’’) compensation (Bandfield et al., 2004) can correct for some, but not all, of the instrument drift using a model of the atmosphere derived from the first year of TES observations. Second, the reference image that is collected after the shutter closes on the IR image, and used to inform the RDR calibration, is often an inaccurate reference for polar observations. There is a significant delay (60–90 s) between the close of the shutter and the start of the reference image. If temperatures are approximately constant during this period (as they are for much of the planet), then the calibration is not affected. However, the particular geometry of north polar images is problematic. The FPA is inevitably warmer during reference image acquisition than during image acquisition for these particular images. This causes an over-estimate of the observed temperatures. For example, we obtain RDR temperatures of about 175 K for CO2 ice, which is about 25 K warmer than it should be. The UDDW compensation provides some correction; CO2 ice is closer to 160 K, but this is still too warm. Therefore, we adopted an empirical calibration that is instead tailored to produce more accurate results for cold temperatures. We convert a raw digital number DN i to temperature T i using the instrument’s current offset (o) and

gain (g) settings as follows: x ¼ ðDN i  o  gÞ  g=16,

(1)

T i ¼ 101:85  lnðxÞ  223:3.

(2)

The parameters in this calibration were derived by performing a fit from DN observations to fully calibrated, undrifted, and dewobbled observations at a variety of temperatures, including cold regions on the planet that do not suffer from the calibration problems for north polar observations. We obtain far more reasonable temperatures, with CO2 ice at approximately 160 K. 3.2. Evidence for a water ice annulus We divided the 197 images into two groups, based on a manual examination of their temperature histograms. Examples of each group are shown in Fig. 4. The images had two or three distinct components (modes). We found that 155 images had two modes, and 42 images had three modes. The images with bimodal temperature histograms signal the presence of two distinct components, such as CO2 frost and exposed regolith. The images with trimodal histograms, since they suggest that a third component is present, are where we believe that we see strong evidence for a water ice annulus just south of the seasonal CO2 cap. Fig. 4b shows the temperature histogram for one of the trimodal images (I09626013). The three modes are centered at 161, 180, and 199 K. Fig. 5 shows the corresponding image temperature profile, which was generated by averaging all pixels in each image line to produce a single cross-track average temperature for that line. While this figure was generated using the averaging approach, all other analyses presented in this paper use all of the original

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

260

220 North

210

South

210

200

200

190

190

180

180

170

170

160

160

150

Temperature (K)

220

150 0

2000

4000

6000

8000

10000

12000

14000

Fig. 5. Temperature profile for THEMIS image I09626013, with cross-track averaging (blue line).

majority of observed temperatures. We interpret the three components as follows:

18 16

Number of images

14 12

Component 1 Component 2 Component 3

10 8 6 4 2 0 150

160

170

180

190

200

210

220

Temperature (K) Fig. 6. Histogram of observed temperatures for each of the three components in trimodal THEMIS images. The component with intermediate temperatures is the annulus between cold (CO2 ) and warm (defrosted) regions.

image pixels (no averaging). The observed temperature increases from north to south, going through two transition points (roughly at lines 5000 and 11,000) where the temperature sharply increases. Although there is undoubtedly some subpixel mixing in the transition areas, the relatively flat temperature profile in the between areas suggests that they are fairly homogeneous. We analyzed all 42 such trimodal images to identify the temperatures of each of their components using the k-means clustering algorithm (MacQueen, 1967). This approach automatically divides all image pixels into k clusters, based on their similarity, and computes the mean for each cluster.1 In this case, k ¼ 3. Fig. 6 shows a histogram of the mean temperatures we observed for each component. Although there is some overlap between the components, there is a clear separation in terms of the 1

While identifying the three temperature clusters was done automatically, the decision about whether two or three modes were present was made based on a manual examination.

Temperature range (K)

Mean temp. (K)

Probable major constituent

157–175 167–206 189–216

166 182 201

CO2 ice/frost Water ice Defrosted terrain

These temperatures are slightly warmer than we would expect each of the components to be, due to the low signalto-noise ratio for observations of cold temperatures and to the intervening atmosphere. There is also likely to be some mixing in each pixel (and therefore each component) that alters the observed mean temperatures. However, overall the temperatures are more realistic than those we obtain from the RDR data. Fig. 7 illustrates the difference concretely for image I09626013. CO2 ice appears to be about 10 K warmer in the RDR version than in our EDR calibration. Similar shifts are observed for the annulus (water ice) and defrosted terrain components. So far, we have inferred the presence of the water ice annulus by examining thermal IR data and determining that this region is too warm to support CO2 frost or ice. To distinguish between water ice and defrosted, dry terrain, it is necessary to examine images at visible wavelengths of the same area at the same time of year. As previously noted, each THEMIS IR image is accompanied by a VIS image that covers a subset of the IR image. In many cases, the VIS image did not cover the region where we found thermal evidence for water ice. However, we did identify several VIS images that do overlap with the annulus. Fig. 8 contains one such image pair, V11622003 and I11622002, that provides compelling evidence for the presence of water ice. The water ice annulus we detected in the IR image extends from 71:56 to 77:51 N (366 km). The accompanying VIS image has a center latitude of 73:17 N and almost entirely covers the annulus region, plus some of the defrosted terrain to the south. The bright regions in the VIS image signify the presence of some kind of ice, and the IR data indicates that this region is far too warm to be CO2 ice but realistic for water ice (182 K).

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

7

EDR RDR/UDDW

x 105

6

261

CO2 Ice Annulus

5 Counts

Defrosted Terrain 4 3 2 1 0 140

160

180

200

220

240

Temperature (K) Fig. 7. Comparison of temperature histograms obtained using RDR calibration (with UDDW) and with empirical EDR pseudo-calibration, on image I09626013.

Fig. 8 also illustrates the benefits of using high spatial resolution THEMIS data for distinguishing between surface components. We are able to study the transitions between components and determine whether they are sharp or gradual. Fig. 8 shows a transition from CO2 ice to water ice that is more gradual than that of the edge between water ice and the defrosted terrain. Further study is needed to determine whether this is genuinely a gradual transition, or if the indistinct edge is due to the lower signal-to-noise ratio for cold temperatures and therefore more uncertainty about where the boundary is. Given these observations, we define the water ice annulus as a contiguous region south of the seasonal CO2 cap edge in which lines of constant latitude are composed of at least 50% water ice. A compositional threshold is necessary because the CO2 frost and the water ice are interleaved rather than being sharply separated. Increasing or decreasing this threshold would correspondingly shrink or lengthen the defined annulus. In the context of THEMIS observations, we interpret this condition as a requirement that each image line must contain at least 50% pixels from Component 2 (water ice temperatures). Given this definition, the recession of the seasonal polar cap (CO2 and water ice) is shown in Fig. 9 as a function of time. This plot shows the north-to-south extent of the annulus as a vertical bar. This figure also shows longitudinally averaged plots of the best-fit modeled IR and VIS cap edges from TES data (Titus, 2005). We see that the observed THEMIS annulus falls mostly between the TES IR (CO2 ) and visible (water ice) cap edges. Some deviation is unsurprising, since the TES values are averaged over all longitudes, while the THEMIS values are observations at specific longitudes. We will compare our detections to the TES models quantitatively in the next section.

Fig. 8. THEMIS IR (band 9 radiance) image I11622002 (Ls 66:38 , longitude 345:72 E), with skewed aspect ratio to permit display of the full image. Left: our classification of this image into three components (CO2 , water ice, and defrosted terrain) based on the temperature histogram analysis. Right: the accompanying visible image, V11622003 (band 3), where the presence of ice can be confirmed visually.

This figure rules out the possibility that any of the water ice we have observed is simply a part of the residual cap, which is known to be composed largely of water ice. The residual cap extends southward only as far as about 80 N,

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

262

85

85 Latitude according to THEMIS observations

IR cap edge (TES) Annulus (THEMIS) Visible cap edge (TES)

Latitude (degrees north)

80

Seasonal CO2 frost

75

70

65 Defrosted terrain

80

75

70

65 Top of the annulus Bottom of the annulus 60

60 10

20

30

40

50

60

70

60

65

70

75

80

85

Latitude according to TES observations

Time of year (Ls) Fig. 9. Latitude and extent of the northern water ice annulus for the 2004 northern cap recession observed in THEMIS data as a function of time of year. Vertical lines show the width of the annulus; observations cover a variety of longitudes. For reference, the longitudinally averaged IR and VIS cap edges observed by TES (Titus, 2005) are also shown.

Fig. 10. Comparison of TES and THEMIS observations in terms of identifying the top (diamonds) and bottom (stars) of the water ice annulus. Points falling on the dashed line indicate perfect agreement between the two instruments.

and all of our annulus observations are further south than this latitude. In addition, we observe an important trend in terms of the width of the annulus. THEMIS and TES observations both suggest that, as the spring advances and the seasonal cap retreats, the annulus becomes wider. We will show this more precisely in the next section.

tendency for the THEMIS observations to identify the top of the annulus as being farther north than the TES observations indicate. This suggests that the transition between water ice and defrosted terrain is a more easily identified feature than the transition from CO2 to water ice. There are two images for which we identify the top of the annulus from THEMIS data to be significantly farther north than the edge found by TES (outlier diamonds in Fig. 10; THEMIS ids I10457015 and I10568009). Investigation of these images reveals that the region identified as the CO2 cap narrows significantly in the northern part of the image, triggering the ‘‘50%’’ criterion prematurely. It is possible that a lower threshold would be more appropriate. If we exclude these images from the evaluation, the mean deviation drops to 1:30 (78 km). By combining the TES models of the IR and VIS cap edges, we can compute the width of the annulus expected from TES observations. Fig. 11 shows this width, indicated by color, as a function of Ls and longitude. Even at this coarse spatial resolution, we can draw two important conclusions. First, the annulus width varies significantly depending on longitude. Second, the width tends to increase as spring advances and the cap recedes. Both of these observations are consistent with what we have reported from our THEMIS observations. Although TES detections of the annulus have previously been reported (Kieffer and Titus, 2001), the increase in annulus width has not before been illustrated in this detail.

4. Comparison of THEMIS and TES observations We analyzed observations made by the TES instrument at the same time our THEMIS data was collected to determine whether the TES observations were consistent with our findings about the northern water ice annulus. We find good agreement between the two instruments in terms of the location and extent of the water ice annulus. We compared each of our detections of the water ice annulus to TES observations at the same location and time of year. As already discussed, the IR edge corresponds to the end of the CO2 seasonal cap, while the VIS edge extends further south because it also includes the water ice annulus. Therefore, we compared the modeled TES IR edge to the top of the annulus we detected in THEMIS images, and the modeled TES VIS edge to the bottom of the annulus we detected. Fig. 10 plots results from the two instruments against each other. Points falling on the dashed line indicate perfect agreement. We find that there is excellent agreement between TES and THEMIS in terms of detecting the bottom of the annulus (VIS edge); the mean absolute deviation is 0:88 of latitude (about 53 km). Since the TESbased model has a 1-sigma error estimate of 0:7 for the spring cap recession, this level of agreement is satisfactory. Agreement for the top of the annulus (IR edge) is lower (mean deviation of 1:58 , or 95 km), and there is a slight

5. Discussion In this section, we first address some important issues involved in comparing our detections with other work.

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

85

263

IR cap edge (TES) Bimodal (THEMIS) Annulus (THEMIS) Visible cap edge (TES)

Latitude (degrees N)

80

75

Seasonal CO2 frost

70

65 Defrosted terrain 60 20

30

40

50

60

70

Time of year (Ls) Fig. 11. Width of the water ice annulus, as observed by TES, as a function of season and longitude. Width is indicated by color, ranging from violet (1 of latitude, 60 km) to white (7 ). Black curves show the location (latitude) of the seasonal CO2 cap edge.

Next, we provide further supporting evidence for our water ice annulus observations by performing a combined analysis of bimodal and trimodal THEMIS IR images. As a final piece of confirmatory evidence, we show agreement between TES and OMEGA observations of the annulus region. There are some important caveats to note when comparing detections of the polar cap edge using different methods. First, in the body of existing work that seeks to study the seasonal polar caps on Mars, several different definitions of the polar cap ‘‘edge’’ have been used. One reason for the different definitions is that the cap edge tends not to be a well defined, crisp feature. Instead, the CO2 ice thins to frost that becomes increasingly patchier and thinner before finally disappearing completely. Prior work with THEMIS (Wagstaff et al., 2005) and MOC (James and Cantor, 2001) identified the edge spatially, while work with TES (Kieffer and Titus, 2001) instead defined the edge temporally as the date when a certain location’s temperature exceeded 165 K. These differences have led to systematic offsets when we compare CO2 cap edge detections side-by-side, and the same is likely to be true for water ice detections. Our criteria for the water ice annulus was that a given THEMIS IR image line had to be more than 50% composed of pixels with temperatures from the second temperature component (compatible with water ice). As mentioned before, changes in this threshold would result in changes to our determination of the width of the annulus. It is also important to note the wavelengths at which the cap edge is being observed. As discussed, observations at visible wavelengths (MOC and TES) are sensitive to all bright ices, while thermal IR (THEMIS and TES) observations generally detect only the CO2 cap, and near

Fig. 12. Position and time of year for bimodal and trimodal THEMIS images used in this study. Trimodal images are represented by the extent of the water ice annulus they contain. Bimodal images are represented by the point at which they transition between two regions (CO2 and water ice or water ice and defrosted terrain). The longitudinally averaged IR and VIS cap edges observed by TES (Titus, 2005) are also shown.

IR observations from OMEGA are sensitive to compositional differences. These various data sets complement each other by providing a range of spatial scales and temporal coverage of the same regions. Bimodal and trimodal images confirm water ice annulus edges: In Section 3.1, we noted that our data set of THEMIS observations of the north polar defrosting zone contained 155 images that were bimodal and just 42 images that were trimodal, which are the images that contain the water ice annulus. It is reasonable to wonder why we did not see evidence for water ice in all of these images. Fig. 12 plots the trimodal images from Fig. 9 as well as the 53 bimodal images that overlap with the seasonal cap edge. Trimodal images are represented by the extent of the water ice annulus they contain, as shown in Fig. 9. Bimodal images are represented by the point at which they transition between their two components, as determined from a manual examination of their temperature profiles (Wagstaff et al., 2005). These images contain only two components because they do not span the entire water ice annulus. They show a transition between either the CO2 cap and the water ice annulus, or the annulus and defrosted terrain. This is demonstrated by the location of the detected transition, which tends to track either the TES IR edge or the TES VIS edge. While the bimodal images also provide information about the location of the annulus, they are incomplete in that they do not cover the annulus fully. To analyze the width and extent of the annulus, we must examine the trimodal images. Agreement about water ice between TES and OMEGA: Recent observations by OMEGA of the north seasonal cap edge have demonstrated evidence for the presence of

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265

264

Fig. 13. Water ice index (1:5 mm band depth) computed from OMEGA observations of the north polar seasonal cap around Ls 40 . The north pole is in the middle left of this image. The two curved lines indicate the best fit estimate of the IR (dashed line) and VIS (solid line) cap edges based on TES data (Titus, 2005).

transient water ice (Bibring et al., 2005). We computed a water ice index value, I w , from OMEGA data as the 1:5 mm band depth, using a continuum calculated from OMEGA’s observations at 1.35 and 1:80 mm ðI cont Þ: I cont ¼ 0:677  O1:35 þ 0:323  O1:80 , Iw ¼

ðI cont  O1:50 Þ , I cont

(3) (4)

where Ol is the reflectance observed by OMEGA at l mm. Fig. 13 shows the resulting I w values for a large portion of the north polar cap during northern spring ðLs ¼ 40 Þ. Values range from red (high) to purple (low). There is a strong water ice signal that correlates with the edge of the seasonal cap as observed in TES VIS data (solid white line). The contemporary TES IR (CO2 ) edge is shown with a dashed line. One explanation for the fact that the water signal extends much farther north than the CO2 edge is that the CO2 acts as a cold trap, and water condenses onto the top of it. This water signal is apparent in the OMEGA spectral data and masks out the underlying CO2 (at 1:5 mm). However, we infer that CO2 is present underneath the water ice due to the temperatures observed by TES and THEMIS northward of the indicated CO2 cap edge. 6. Conclusions and future work We have presented an analysis of THEMIS IR images covering the north polar region during the 2004 seasonal cap recession, in which we see evidence for a water ice annulus that follows the CO2 cap edge as it retreats. We find that this annulus can range from 167 to 206 K, with a mean of 182 K. While the annulus varies in width with longitude and season, it is definitely a phenomenon distinct from the seasonal CO2 cap. We find that the annulus tends to grow wider as spring advances and the seasonal cap recedes, a claim that is supported by contemporary observations from TES. We also find good agreement between THEMIS, TES, and OMEGA observations about the nature and location of the water ice annulus. A natural next step is to determine whether there is a corresponding water ice annulus in the southern seasonal

ice cap. With the recent discoveries of water ice in the southern hemisphere (Titus et al., 2003), it is not only possible, but probable, that transient water ice also tracks the retreat of the southern seasonal cap. The MAWD instrument did not detect a significant increase in water vapor for the south polar cap, but Smith (2002) detected large amounts of water vapor coming off the cap in southern summer, using TES data. This water vapor could be a potential source for a southern water ice annulus. In the future, we plan to examine THEMIS data from the south pole of Mars to determine whether we see evidence for an annulus. If so, we will perform a thorough comparison of the annuli in terms of temperature, physical extent, and seasonal persistence. We require THEMIS data due to its high spatial resolution and good coverage both spatially and temporally. We will also compare our observations against what is revealed in OMEGA data based on spectral analysis. Acknowledgments This research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. It was funded by the New Millenium Program, the Interplanetary Network Directorate, the NASA Advanced Information Systems Research Program, and the Mars Odyssey Participating Scientist program. We wish to thank Steve Chien for his continued support, Paul Geissler and Ken Herkenhoff for their suggestions on how to improve the manuscript, and the TES, THEMIS, and OMEGA teams, without whom this work would not have been possible. References Bandfield, J.L., Rogers, D., Smith, M.D., Christensen, P.R., 2004. Atmospheric correction and surface spectral unit mapping using thermal emission imaging system data. J. Geophys. Res. 109 (E10), E10008. Bibring, J.-P., Langevin, Y., Gendrin, A., Gondet, B., Poulet, F., Berthe, M., Soufflot, A., Arvidson, R., Mangold, N., Mustard, J., Drossart,

ARTICLE IN PRESS K.L. Wagstaff et al. / Planetary and Space Science 56 (2008) 256–265 P., the OMEGA team, 2005. Surface diversity as revealed by the OMEGA/Mars Express Observations. Science 307, 1576–1581. Christensen, P.R., Bandfield, J.L., Bell, J.F., Gorelick, N., Hamilton, V.E., Ivanov, A., Jakosky, B.M., Kieffer, H.H., Lane, M.D., Malin, M.C., McConnochie, T., McEwen, A.S., McSween, H.Y., Mehall, G.L., Moersch, J.E., Nealson, K.H., Rice, J.W., Richardson, M.I., Ruff, S.W., Smith, M.D., Titus, T.N., Wyatt, M.B., 2003. Morphology and composition of the surface of Mars: Mars Odyssey THEMIS results. Science 300, 2056–2061. Feldman, W.C., Prettyman, T.H., Boynton, W.V., Murphy, J.R., Squyres, S., Karunatillake, S., Maurice, S., Tokar, R.L., McKinney, G.W., Hamara, D.K., Kelly, N., Kerry, K., 2003. CO2 frost cap thickness on Mars during northern winter and spring. J. Geophys. Res. 108 (E9), 5103. Folkner, W.M., Yoder, C.F., Yuan, D.N., Standish, E.M., Preston, R.A., 1997. Interior structure and seasonal mass redistribution of Mars from radio tracking of Mars pathfinder. Science 278, 1749–1751. Forget, F., 2005. Mars water ice and carbon dioxide seasonal polar caps: GCM modeling and comparison with Mars express OMEGA observations. In: Proceedings of the 36th Lunar and Planetary Science Conference. Abstract #1605. Hess, S., Ryan, J., Tillman, J., Henry, R., Leovy, C., 1980. The annual cycle of pressure on Mars measured by Viking-lander-1 and Vikinglander-2. Geophys. Res. Lett. 7, 197–200. Houben, H., Haberle, R.M., Young, R.E., Zent, A.P., 1997. Modeling the Martian seasonal water cycle. J. Geophys. Res. 102 (E4), 9069–9083. Jakosky, B.M., 1985. The seasonal cycle of water on Mars. Space Sci. Rev. 41, 131–200. Jakosky, B.M., Farmer, C.B., 1982. The seasonal and global behavior of water vapor in the Mars atmosphere: complete global results of the viking atmospheric water detector experiment. J. Geophys. Res. 87, 2999–3019. James, P.B., Cantor, B.A., 2001. Martian north polar cap recession: 2000 Mars orbiter camera observations. Icarus 154, 131–144. Kieffer, H.H., 1979. Mars south polar spring and summer temperatures: a residual CO2 frost. J. Geophys. Res. 84, 8263–8288. Kieffer, H.H., Titus, T.N., 2001. TES mapping of Mars’ north seasonal cap. Icarus 154, 162–180. Kieffer, H.H., Chase Jr., S.C., Martin, T.Z., Miner, E.D., Palluconi, F.D., 1976. Martian north pole summer temperatures: dirty water ice. Science 194, 1341–1344.

265

Litvak, M.L., Mitrofanov, I.G., Kozyrev, A.S., Sanin, A.B., Tret’yakov, V.I., Boynton, W.V., Shinohara, C., Hamara, D., Saunders, S., Drake, D.M., 2004. Seasonal carbon dioxide depositions on the Martian surface as revealed from neutron measurements by the HEND instrument onboard the 2001 Mars Odyssey spacecraft. Sol. Syst. Res. 38 (3), 167–177. MacQueen, J.B., 1967. Some methods for classification and analysis of multivariate observations. In: Proceedings of the Fifth Symposium on Math, Statistics, and Probability, vol. 1. University of California Press, Berkeley, CA, pp. 281–297. Montmessin, F., Forget, F., Rannou, P., Cabone, M., Haberle, R.M., 2004. Origin and role of water ice clouds in the Martian water cycle as inferred from a general circulation model. J. Geophys. Res. 109 (E10), E10004. Prettyman, T.H., Feldman, W.C., Murphy, J.R., Funsten, H.O., Lawrence, D.J., Linn, R.R., Maurice, S., Tokar, R.L., 2003. Seasonal advance and retreat of Mars’ south polar cap as measured by the Mars Odyssey neutron spectrometer. In: Proceedings of the 34th Lunar and Planetary Science Conference. Abstract #8099. Smith, M.D., 2002. The annual cycle of water vapor on Mars as observed by the thermal emission Spectrometer. J. Geophys. Res. 107, E101029. Smith, D.E., Zuber, M.T., 2002. Seasonal changes in the icecaps of Mars from laser altimetry and gravity. In: Proceedings of the 13th International Workshop on Laser Ranging: Science Session. Smith, D.E., Zuber, M.T., Neumann, G.A., 2001. Seasonal variations of snow depth on Mars. Science 294, 2141–2146. Tamppari, L.K., Smith, M.D., Hale, A.S., Bass, D.S., 2003. Interannual comparison of water vapor in the north polar region of Mars. In: Proceedings of the Third Mars Polar Science Conference. Abstract #8119. Tanaka, K.L., Scott, D.H., 1987. USGS Miscellaneous Investigations Series Map I 1802 C. Atlas of Mars, 1:15,000,000 Geologic Series. Titus, T.N., 2005. Mars polar cap edges tracked over 3 full Mars years. 36th Annual Lunar and Planetary Science Conference, March 14–18, 2005, League City, Texas, Abstract #1993. Titus, T.N., Kieffer, H.H., Christensen, P.R., 2003. Exposed water ice discovered near the south pole of Mars. Science 299, 1048–1051. Wagstaff, K.L., Castan˜o, R., Chien, S., Ivanov, A.B., Pounders, E., Titus, T.N., 2005. An onboard data analysis method to track the seasonal polar caps on Mars. In: Proceedings of the Eighth International Symposium on Artificial Intelligence, Robotics and Automation in Space.