Water Content in Different Regions of Mars According to HENDOdyssey Data

Estimation of the water content in the subsurface cannot be done using the neutron data alone. The estimation is model-dependent and one has to use the procedure of neutron data deconvolution. In this section we present the first results of HEND data analysis based on the testing of homogeneous and double-layered models of the subsurface (see [19] for details).

5.5.1 Testing Models for Regions with Brightest Emission of Neutrons

For the sake of simplicity, we may start the neutron data analysis by using the depth-homogeneous model of the subsurface layer. According to this model, the subsurface is homogeneous along the depth and each unit of mass contains a part Zhom of water and a part (1 - Zhom) of a soil with 50:50 fraction of sand and stones, which have the abundance of major elements according to the Pathfinder composition [10]. The signals SSD, SMD, SLD and SSC(IINN from 4 sensors of HEND are used for testing this model (see Section 5.3). The Pearson criterion is used to test the model by observational data where Ci are counts from different sensors i = 1-4, Mi are the model predictions of these counts, , are the errors of measurements and Pj are the set of model parameters (only one parameter, Zhom, for the homogeneous model).

According to direct measurements from Viking 1, Viking 2 and Mars Pathfinder, the content of water in the soil is about 1-3 wt % (e.g. [10]). The composition of soil in Solis Planum is thought to be similar to the composition at the landing sites of these missions because there is no considerable difference between them in the brightness of major gamma-ray lines [20]. One may use Solis Planum as the "reference region" for neutron data deconvolution, assuming that it has equal mass fractions (49 wt %) of sand and stones with Pathfinder composition and 2 wt % of water. Variation of neutron flux over this region could only result either from seasonal changes of the atmosphere or from long-term variations of cosmic rays. Changes in the atmosphere were described by the general circulation model (GCM) [18], and the variation of cosmic rays has been evaluated as a free fitting parameter by direct comparison between the observational data at Solis Planum and the model predictions for fixed composition of the soil.

Using the estimated flux of cosmic rays according to data from Solis Planum as one fixed parameter, the thickness of the atmosphere above the selected region as another fixed parameter, and parameters Pj of the structure of the soil as variable fitting parameters for testing the model, the model counts Mi(Pj) were calculated by standard code MCNPX [21]. Using the homogeneous model of the soil, the best fitting value of parameter Zhom(*) could be found by minimization of the function S(P) in eq (1). This value could be used as the estimation of water content, provided the best fitting model could be accepted according to the statistics of X3. However, if the minimum of function S(P) is too large for consistency with the statistics of X3, the tested model should be rejected, and the best fitting parameter Zhom(*) for this model can not be used as an estimation of the water content supported by observations.

For verifying the algorithm of the neutron data deconvolution, we used the data for North Hellas and for the Argyre basin, which have the same fluxes of neutrons as Solis Planum (see Fig. 5.8). One may expect that the estimated content of water should be similar to the value of 2 wt % postulated for the "reference region" of Solis Planum. Indeed, the water content was found to be 1.6 wt % for Argyre and 2.2 wt % for North Hellas.

5.5.2 Testing Models for High-Latitude Regions of Permafrost

The homogeneous subsurface model was tested for the northern region of depression of neutron emission at high latitudes (Fig. 5.8). The HEND data for season intervals LS=(120-150)° was used for the north because it corresponds to the summer surface free of seasonal deposit of CO2. It was found that the homogeneous model has rather high acceptance probabilities for the latitude belts of the northern region (Table 5.1). We believe that a more complex model of the subsurface is not necessary for the northern permafrost region. The estimated contents of water ice correspond to 44, 25 and 13 wt % for the Northern Polar Region (> 80°N), for the Northern High Latitude Belt (70-80°N) and for the Northern Boundary Belt (60-70°N), respectively. For the spot of northern absolute minimum of neutron emission, the best fitting parameter of water content is about 53 wt % (Table 5.1).

To build the map of water contents, the northern permafrost region was divided into 74 overlapping surface elements (pixels). The pixel sizes approximately correspond to the horizontal resolution of HEND mapping from the orbit. We found the sample of the best fitting parameters of the water ice content Zhom(*) for these pixels according to the homogeneous model. The distribution of minimum values Smln(Zhom(*)) of eq (1) for these pixels agrees with the distribution curve for random variable Xi (Fig. 5.9). The map of water content of the northern region was created by smoothing values Zhom(*) for 74 pixels by a filter of 5° degrees (Fig. 5.10). As expected, the largest content of water ice of about 53 wt % is observed in the spot of the absolute minimum of neutron emission, which is at the beginning of the ice canyon Chasma Boreale.

Table 5.1 Estimations of average water ice content in the polar regions of Mars.

Name of latitude belts

Longitude latitude

Average water

Acceptance probability

Average water ice

Acceptance probability

and seasons (LS) of data accumulation

content (homogeneo us model)

(homogeneous model)

content in the bottom layer and thickness of the dry upper layer (double-layered model)

(double-

layered model)

Northern

0-360°E

44.1 %

0.6

--

--

polar region

> 80°N

120-150°

Northern

0-360°E

24.8 %

0.4

--

--

high latitude belt

70-80°N

120-150°

Northern

0-360°E

12.5 %

0.7

--

--

boundary belt

60-70°N

120-150°

North. spot

24-28°E

53 %

0.8

--

--

of abs. min.

86-90°N

of neutron

emission

120-150°

Southern

0-360°E

22.7 %

0.004

55.0 %

0.26

polar region

> 80°S

16 g cm-2

330-360°

Southern

0-360°E

19.5 %

0.03

54.0 %

0.41

high latitude belt

70-80°S

19.2 g cm-2

330-360°

Southern

0-360°E

11.2%

0.07

25.3 %

0.72

boundary belt

60-70°S

22.4 g cm-2

330-360°

South. spot

220-

25 %

0.02

55.0 %

0.08

of abs. min.

224°E

16.0 g cm-2

of neutron

78-82°S

emission

330-360°

One may conclude [19] that the northern region of neutron depression could be identified as the northern permafrost region (NPR) with a high content of water ice.

We may suggest the level of 10 wt % of water ice as the boundary of NPR. In the eastern hemisphere (0-180°E) the level of 10 wt % water ice goes along the latitude of about 60°N (Figs. 5.8 and 5.10). In the sector 210-220°E the boundary of NPR is at the latitude of 45°N (Fig. 5.8). In this sector a large area with a very high content of water ice (about 35-40 wt %) was found, which lies outward from the north pole at latitudes 65-75°N.

is o

tn u

Fig. 5.9 Distribution of minimization values of eq (1) for surface elements of NPR (thin line) and SPR (thick line). The dashed line corresponds to the distribution of random variable X3.

In the second part (270-360°E) of the western hemisphere the boundary level of NPR goes up, and it touches the latitude of 75°N in the sector 310-330°E (Figs. 5.8 and 5.10). The estimated content of water ice in the surroundings (320°E, 70°N) is less than 10 wt %. The reason for the difference of about 25-20 wt % of water ice content between two areas in the western part of the northern permafrost at longitudes 210-220°E and 310-330°E along the latitude of 70°N is not clear. These two areas have rather different elevations: the first area with a higher content of water ice has a higher elevation. However, there is no clear correspondence between the content of water ice and the elevation for the western segment of NPR of Mars (Fig. 5.11).

We believe that the map (Fig. 5.10) represents the real distribution of subsurface water ice. It is very similar to the map of water content based on independent measurements by the Neutron Spectrometer [22]. This map could be used for a comparison with other sets of data about surface relief, about surface geochemistry and with future data of radio sounding of the shallow subsurface.

Fig. 5.10 Map of water ice content (in wt %) for the NPR northward of 60°N according to HEND data deconvolution for the homogeneous model [19]. The coordinates shown are east longitudes.

On the contrary, testing of the homogeneous model for latitude belts of the southern permafrost region has not been successful (Table 5.1). We used HEND data for season interval LS=(330-360)° because it corresponds to the summer surface free of seasonal deposit of CO2. It was found that for all 4 tested cases the minima of eq (1) were too large for acceptance of the model: the probabilities of acceptance are rather small. However, one could guess that the homogeneous model does not work quite well for the latitude belts as a whole because there is a large spread of the water ice contents at different surface elements along the belts. The subsurface could be homogeneous from the depth up to the surface at each surface element, but it could be non-homogeneous over the entire belt. To study this possibility, the homogeneous model was tested for each of the 98 individual surface elements of the SPR (southern permafrost region). The distribution of minimum values Smin(Zhom<*)) of testing function (1) is shown in Fig. 5.9. It should be observed that this distribution is very different from the reference curve for distribution of X, provided the differences between observed counts Ci and predicted values Mi arise due to statistical fluctuations. Thus, the model of the homogeneous subsurface should be rejected for the southern permafrost region both for latitude-averaged data and for the sample of 74 surface elements over the entire region.

Longituda.dagr«

Longituda.dagr«

Ijcnglluda.dEgrBE

Fig. 5.11 The observation of epithermal neutron flux and surface elevation within the 65-75°N latitude belt. The upper graph contains the dependency of neutron flux from longitude for late northern winter (black dots) and for early northern summer (triangles). The bottom graph presents the dependency of elevation versus east longitude.

Ijcnglluda.dEgrBE

Fig. 5.11 The observation of epithermal neutron flux and surface elevation within the 65-75°N latitude belt. The upper graph contains the dependency of neutron flux from longitude for late northern winter (black dots) and for early northern summer (triangles). The bottom graph presents the dependency of elevation versus east longitude.

The next level of complexity of the model of the subsurface corresponds to the model of the double-layered structure of the subsurface [19]. According to this model, the "dry" layer at the top is assumed to have a variable column density hup but fixed composition of 98 wt % of soil (49 wt % of sand and 49 wt % of stones with the Pathfinder composition [10]) and 2 wt % of water. The bottom "wet" layer has variable fractions of water ice Zdown % and soil (1 - Zdown) % . Again, the soil has 50 % sand and 50 % stones with the Pathfinder composition. Therefore, the double-layered model has two free parameters, hup(g cm-2) and Zdown (wt %).

Five observational signals from HEND SSD, SMD, SLD, Ssc/in/n(0.85-2.5 MeV) and Ssc/in/n(>2.5 MeV) were used to test the double-layered model: three signals from 3He counters in SD, MD and LD, and two signals from scintillator SC/IN sensor: counts at the energy ranges 0.85-2.5 MeV and above 2.5 MeV. Using five independent signals for testing the model with two free parameters, one may compare statistics of the minimal values Smin(hup(*), Z^") of eq (1) with the same distribution of random variable X3 as for the testing of the homogeneous model (Fig. 5.9).

The average data for southern latitude belts support the double-layered model of the subsurface (Table 5.1). The best fitting values of free parameters correspond to rather high acceptance probability for this model. The column density of the top dry layer is about 16-25 g cm-2 for all three belts. The average content of water ice is about 55 wt % south of 70°S and decreases down to 25 wt % at the Southern Boundary Belt (60-70°S). There is no noticeable difference between the water ice content at the southern spot of absolute minimum (about 55 wt %) and the average content south of 70°S.

The statistics of the best fitting parameters h^*1 (g cm-2) and Zdow," (wt %) was estimated for 98 surface elements of SPR. The double-layered model works rather well, the distribution of minima ^„(h^*1, Zd,,»/*') of eq (1) agrees with the statistics for the random variable %23 (Fig. 5.9). The smoothed surface distributions of the best fitting values hup0 (g cm-2) and Zdow/* (wt %) represent two maps for the SPR for water ice content and for the surface density of the dry covering layer, respectively (Figs. 5.12 and 5.13).

One may conclude [18] that the southern region of neutron depression could be identified as the southern permafrost region (SPR) with a high content of water ice. This region has a double-layered structure of the subsurface: the dry layer at the top covers the layer with a high content of water ice at the bottom. The column density of the top layer is about 16-22 g cm-2. Assuming a soil bulk density of about 1.6 g cm-3, one may estimate that the thickness of the dry layer of SPR is about 14 cm. For SPR the boundary level of 10 wt % water ice approximately follows the latitude of 65°S. In the longitude sector (60-150°E) the boundary goes down to moderate latitudes below 60°S and follows the southern edge of the Hellas basin, but surprisingly the permafrost does not come into the basin itself (Figs. 5.8 and 5.12).

In the longitude sector (300-360°E) the 50 wt % level of high water ice content goes up to the latitude 82°S. On the contrary, in the symmetrically opposite longitude sector (120-180°E) the level of 50 wt % goes down to the latitude of about 68°S. One may note that the 50 wt % contour of water ice and the contour of the southern residual polar cap are shifted in respect to each other along the meridians 150°/330°E (Fig. 5.12). Generally speaking, the double-layered model of the subsurface is not applicable to the area of the residual polar cap because the substance of the cap is a mixture of carbon dioxide and water ice instead of the mixture of soil and water ice assumed in the model. The analysis of the composition of the southern residual polar cap will be presented elsewhere. However, the present model allows us to make some qualitative conclusions about the southern residual cap.

Fig. 5.12 Map of water ice content (in wt %) for the SPR southward of 70°S according to HEND data deconvolution for the double-layered model [19]. The coordinates shown are east longitudes.

According to the map based on the double-layered model (Figs. 5.12 and 5.13), the southern residual cap could be divided into two parts: one part lies north of the latitude 80°S and the other one lies south of it. The poleward part of the permanent cap has about the same content of water ice (> 50 wt %) as the surrounding frost-free area of SPR, and it is covered by a top layer of dry substance (CO2?) with a thickness of about 17 g cm-2. This top dry layer of the residual polar cap has about the same thickness as the top layer of dry soil at high latitudes outside the cap (Fig. 5.13). A part of the residual cap south of 80°S also has a high but somewhat smaller content of water ice (about 30-40 wt %) and this layer is also covered by a layer of some dry material (CO2?) with a smaller thickness of about 10 g cm-2 (Fig. 5.13).

Fig. 5.13 Map of the column density (in g cm-2) of the top dry layer above the water-rich layer in the southern permafrost region according to HEND data deconvolution for the double-layered model [19].

The high content of water ice in the southern residual cap is also well-proved by the seasonal variation of neutrons. Indeed, during winter the residual cap is covered by a thick seasonal deposition layer of atmospheric CO2. The winter layer contains a small fraction of water, and its accumulation leads to a significant increase of epithermal neutron flux from the surface [15] (Fig. 5.14). Therefore, the southern residual cap as well as the northern residual cap should contain a large fraction of water ice of about 50 wt % according to the present estimations [19]. (See also an independent discussion of polar caps in Chap. 6 by Hvidberg.)

Both permafrost regions NPR and SPR have about the same maximal water ice content of about 50-60 wt %. The ice is lighter than the soil, and this fraction of mass corresponds to a volume fraction of about 60-70 %. In the case of terrestrial permafrost, water ice fills the pore volumes between the grains of the soil, and this volume is about 30 %. The estimations for Martian permafrost indicate that water ice is the dominating material of the subsurface. The regions on Mars with high content of water (> 50 wt %) would be more appropriately named the "frozen oceans of dirty water ice". One has to explain the origin of "frozen oceans" around poles down to latitudes of about 60°. Direct deposition of water ice could be the main process of formation of the subsurface. Sand and stones are the components of the subsurface of these regions, and were delivered either by dust storms or by excavation by asteroids.

Fig. 5.14 Increase of the flux of epithermal neutrons in the south during the transition from summer (Ls = 330-360°) to winter (Ls= 90-180°). The flux of neutrons is normalized by the maximal flux for Solis Planum. Fluxes for southern latitude belts 85-90°, 80-85°, 75-80°, 70-75°, 65-70° and 60-65° are shown by dark blue, bright blue, green, yellow, orange and red squares, respectively.

Fig. 5.14 Increase of the flux of epithermal neutrons in the south during the transition from summer (Ls = 330-360°) to winter (Ls= 90-180°). The flux of neutrons is normalized by the maximal flux for Solis Planum. Fluxes for southern latitude belts 85-90°, 80-85°, 75-80°, 70-75°, 65-70° and 60-65° are shown by dark blue, bright blue, green, yellow, orange and red squares, respectively.

5.5.3 Testing Models for Neutron Depression Regions at Moderate Latitudes

There are two regions at low latitudes on Mars with decreased emission of epithermal neutrons (Fig. 5.8): Arabia and Memnonia. However, one may note that this depression anomaly is not observed on the map for high energy neutrons (Fig. 5.8). The Arabia region is sufficiently large, and it could be divided into the North Arabia and South Arabia regions (Table 5.2). Also, we may select one more particular region, South Hellas, which lies at the boundary of SPR and which could be the best reservoir for accumulation of shallow water at moderate latitudes. All four regions have much smaller emissivity of epithermal neutrons in comparison with Solis Planum. Therefore, the soil in these regions has to contain more water than the 2 wt % assumed for the Solis Planum soil. We used HEND data for a large seasonal interval LS = (0-150)° because conditions of the atmosphere remain stable during this interval.

To estimate the content of water in the four regions we may use both models [19]: the homogeneous model of the subsurface with variable content of water Zhom (wt %) along the depth and the double-layered model with two variable parameters: the thickness of the upper dry layer hup(g cm-2) and the content of water in the bottom wet layer Zdown (wt %). For both models we use 5 independent signals of HEND measurements (SSD, SMD, SLD, SSC/IN/N (0.85-2.5 MeV) and SSC/IN/N (> 2.5 MeV)). (The signal SSC/IN/N from scintillator SC/IN was separated into two signals at energy ranges (0.85-2.5) MeV and above 2.5 MeV.) The best fitting values of parameters for both models are presented in Table 5.2.

The homogeneous model gives similar estimations of water content (5.0-5.2 wt %) for all four regions. For the double-layered model the best fitting values of water content in the bottom layer Zdowne*) are 9-10 wt %, which is about twice as high as the estimation for the homogeneous model. The thickness of the top dry layer is similar for all four regions studied (26-32 g cm-2).

The acceptance probabilites for the homogeneous model are rather small (< 0.15), which means that this model is not well supported by the observations. On the other hand, the observational data clearly support the double-layered model for all four tested regions at moderate latitudes (Table 5.2). The best fitting parameters for the layered model may be used as reliable estimations of parameters of the subsurface h^*1 = 26-32 g cm-2 and Zdowf = 9-10 wt % for all four tested regions. For a soil density of 1.6 g cm-3 this column density corresponds to a thickness of 16-20 cm. The double-layered structure of the subsurface of Arabia and Mamnonia is illustrated by the maps of epithermal and high energy neutrons. The first map (Fig. 5.8 top) for epithermal neutrons represents a content of water in the layer of about 1-2 m, and this map has well-pronounced depression. The second map (Fig. 5.8 bottom) for high energy neutrons represents the top layer of the subsurface of about 20 cm thickness, and it may not manifest any depression of neutron emission.

Table 5.2 Estimated content of water in regions with depression of epithermal neutrons at moderate latitudes.

Names of regions and seasons (LS) of data accumulation

Longitude latitude

Estimation of water content (homogeneous model)

Acceptance probability (homogeneous model)

Estimation of water content and thickness of dry layer (double-layered model)

Acceptance probability

(double-

layered model)

South Hellas

50-98°E

5.0%

0.1

9.0%

0.2

0-150°

47-53°S

26.0 g cm-2

North Arabia

0-45 °E

5.1%

0.15

9.0%

0.7

0-150°

0-30°S

26.0 g cm-2

South Arabia

0-45°E

5.2%

0.02

10.0%

0.25

0-150°

0-20°S

32.0 g cm-2

Memnonia

180-

5.1%

0.1

9.0%

0.7

0-150°

200°E 0-25°S

29.0 g cm-2

The area of Arabia is sufficiently large in comparison with the scale of horizontal resolution for orbital measurements. It was divided into a large number of surface elements with a size of 2°x2° (Fig. 5.15), whose data were individually tested by the double-layered model. For all surface elements the double-layered model is found to agree with the observational data, and this model provides a pair of best fitting parameters hup(*' and Zdown<*) of the subsurface composition. One particular surface element with coordinates (30°E, 10°N) has the smallest emission of epithermal neutrons (Fig. 5.15), and the best fitting parameters of the subsurface for this element correspond to a content of water of 16 wt % under the dry layer with a thickness of 29 g cm-2. The estimation of the dry layer is consistent with the average value found for the entire North Arabia (Table 5.2). Therefore, the high content of water at this surface element is not produced by uncertainties of model-dependent data deconvolution. The value of 16 wt % could correspond to the real minimum of epithermal neutron flux in Arabia. We may name this spot the "Arabian water-rich spot" (AWRS) [19]. It is located around the old eroded crater between the famous craters Cassini and Schiaparelli (Fig. 5.15).

One has take to into account that the water content of 16 wt % in AWRS has been estimated using the simplest numerical model of Martian neutron emission, when the same composition of subsurface is attributed to the entire planet. This approach provides the lowest limit of water content in AWRS because the data accumulated in these surface elements are also contributed by surrounding surface elements with higher fluxes of neutrons. One has to use more sophisticated methods of inverse projection of orbital data to the surface elements to get a more accurate estimation of water contents of AWRS. However, even in the present initial stage of the neutron data analysis one may draw reliable conclusions that the spot at (30°E, 10°N) corresponds to the maximal water content in Arabia, and the corresponding value of 16 wt % is the lowest limit for the real amount in this spot.

This value of 16 wt % could be compared with the theory of water forms in the Martian soil (see Chaps. 4, 7 and 8 for a review). One has to check the first alternative as to whether this content could be associated with the chemically bound water in the minerals. If so, predictions should be drawn about the abundances of elements, which constitute the water-bearing minerals in AWRS. These predictions could be tested by data from the gamma-ray spectrometer. A scenario of water enrichment for these minerals should be developed to explain the condition of the past history of Mars, when 16 wt % of water molecules were implemented into the minerals of shallow subsurface. If this content would be too high to be associated with the chemically bound water in minerals, a second, alternative model of water ice condensation and preservation should be developed. The value of 16 wt % in AWRS just above the equator is challenging for both alternatives, and this conflict between theory and observation is probably the most important contribution of neutron measurements for present Mars science.

005 010 015 020 0.25

005 010 015 020 0.25

Fig. 5.15 Map of epithermal neutrons (normalized neutron flux) in the Arabia region (10°S-30°N, 350°E-50°E) with a surface resolution of 2°x2° [19]. The large crater centered at 2.7°S, 16.7°E is the Schiaparelli crater.

Fig. 5.15 Map of epithermal neutrons (normalized neutron flux) in the Arabia region (10°S-30°N, 350°E-50°E) with a surface resolution of 2°x2° [19]. The large crater centered at 2.7°S, 16.7°E is the Schiaparelli crater.

0 0

Post a comment