Planetary and Space Science 131 (2016) 60–69 Contents lists available at ScienceDirect Planetary and Space Science http://d 0032-06 n Corr E-m ppina@ aryvinic journal homepage: www.elsevier.com/locate/pss Comparing wind directions inferred from Martian dust devil tracks analysis with those predicted by the Mars Climate Database T. Statella a,n, P. Pina b, E.A. Silva c, Ary Vinicius Nervis Frigeri a, Frederico Gallon Neto a a Instituto Federal de Educação, Ciência e Tecnologia de Mato Grosso – IFMT, 95 Zulmira Canavarro780025-200, Cuiabá, Brazil b CERENA, Instituto Superior Técnico - IST, Universidade de Lisboa, Av. Rovisco Pais1049-001, Lisboa, Portugal c Universidade Estadual Paulista, Faculdade de Ciências e Tecnologia – FCT, 305 Roberto Simonsen19060-900, Presidente Prudente, Brazil a r t i c l e i n f o Article history: Received 7 December 2015 Received in revised form 7 June 2016 Accepted 21 July 2016 Available online 25 July 2016 x.doi.org/10.1016/j.pss.2016.07.004 33/& 2016 Elsevier Ltd. All rights reserved. esponding author. ail addresses: thiago.statella@cba.ifmt.edu.br ( tecnico.ulisboa.pt (P. Pina), erivaldo@fct.unesp iusnf@gmail.com (A.V. Nervis Frigeri), fredgn a b s t r a c t We have calculated the prevailing dust devil tracks direction as a means of verifying the Mars Climate Database (MCD) predicted wind directions accuracy. For that purpose we have applied an automatic method based on morphological openings for inferring the prevailing tracks direction in a dataset comprising 200 Mars Orbiter Camera (MOC) Narrow Angle (NA) and High Resolution Imaging Science Experiment (HiRISE) images of the Martian surface, depicting regions in the Aeolis, Eridania, Noachis, Argyre and Hellas quadrangles. The prevailing local wind directions were calculated from the MCD predicted speeds for the WE and SN wind components. The results showed that the MCD may not be able to predict accurately the locally dominant wind direction near the surface. In adittion, we confirm that the surface wind stress alone cannot produce dust lifting in the studied sites, since it never exceeds the threshold value of 0.0225 Nm�2 in the MCD. & 2016 Elsevier Ltd. All rights reserved. 1. Introduction Dust devils are small whirlwinds made visible by entrained dust and sand. They are upward moving, spiraling flows caused by heating of near-surface air by insolation (Balme and Greeley, 2006). Dust devils activity is easily identified in remotely sensed images by the tracks they often leave behind. Dust devil tracks are albedo patterns on planetary surfaces that result from the removal of particles by the presence of a dust devil to expose an underlying surface with a different albedo. On Mars, dust devil tracks den- sities were shown to change with the time of the year, suggesting that dust devil activity also depends on the season of the year (Thomas et al., 2003; Whelley and Greeley, 2006). Those albedo features tend to fade with time, which is attributed to the de- position of dust (Malin and Edgett, 2001; Balme et al., 2003). In a period of a few months (2 to 4 months) old tracks can be re- covered by dust and replaced by new tracks (Statella et al., 2015). Important information that can be obtained by the analysis of the tracks is the trajectory of the vortices, which can be used to infer the prevailing wind orientation near the surface, as dust devils are typically thought to move, on average, in the direction of the dominant wind (Rennó et al., 1998). The inference of the wind T. Statella), .br (E.A. Silva), eto@gmail.com (F.G. Neto). direction based on aeolian features is one of the few procedures for verifying circulation models of the Martian atmosphere. As a plethora of orbital images of the surface of Mars has been made available, automatic methods can be developed and applied to detect (Statella et al., 2012) and measure the orientation of dust devil tracks. The prevailing direction of the wind at specific locations on the surface of Mars for a certain solar longitude can be estimated from extractions of the Mars Climate Database (MCD). Such extractions can then be compared to the prevailing direction of dust devil tracks previously detected (Statella et al., 2012) in Mars Orbiter Camera (MOC) Narrow Angle (NA) and High Resolution Imaging Science Experiment (HiRISE) images at the same locations. From the detected dust devil tracks we could estimate the prevailing track direction for each image by using the Directional Morpho- logical Openings by Linear Structuring Elements, as proposed by Statella et al. (2014). The prevailing direction of dust devil tracks for each scene has been compared to the MCD extractions in order to find out if the MCD information could be used to describe local wind patterns on the surface of Mars. 2. Image Dataset The image dataset from which we have detected the dust devil tracks was made up of 124 images: 75 MOC narrow angle pan- chromatic band and 49 HiRISE red band from Aeolis, Argyre, www.sciencedirect.com/science/journal/00320633 www.elsevier.com/locate/pss http://dx.doi.org/10.1016/j.pss.2016.07.004 http://dx.doi.org/10.1016/j.pss.2016.07.004 http://dx.doi.org/10.1016/j.pss.2016.07.004 http://crossmark.crossref.org/dialog/?doi=10.1016/j.pss.2016.07.004&domain=pdf http://crossmark.crossref.org/dialog/?doi=10.1016/j.pss.2016.07.004&domain=pdf http://crossmark.crossref.org/dialog/?doi=10.1016/j.pss.2016.07.004&domain=pdf mailto:thiago.statella@cba.ifmt.edu.br mailto:ppina@tecnico.ulisboa.pt mailto:erivaldo@fct.unesp.br mailto:aryviniciusnf@gmail.com mailto:fredgneto@gmail.com http://dx.doi.org/10.1016/j.pss.2016.07.004 Fig. 1. Distribution of the image dataset: Aeolis (MC23), Argyre (MC26), Noachis (MC27), Hellas (MC28) and Eridania (MC29). The white triangles represent the center coordinates of the original scenes. Image credits: R.K. Hayward, K.F. Mullins, L.K. Fenton, T.M. Hare, T.N. Titus, M.C. Bourke, A. Colaprete, P.R. Christensen. T. Statella et al. / Planetary and Space Science 131 (2016) 60–69 61 Noachis, Hellas and Eridania quadrangles defined by Mars Charts (MC, as defined by the USGS mapping quadrangles) 23, 26, 27, 28 and 29, respectively. By having detected the dust devil tracks we mean that the image dataset was formed by a set of binary images in which pixels belonging to tracks have been la- beled with value 1 (white) and pixels not belonging to tracks have been labeled with value 0 (black). The detection procedure has been fully described in Statella et al. (2012). The spatial resolution of HiRISE images was either 0.25 m (∼90% of the images) or 0.50 m while the MOC spatial resolution varied from 1.43 m to 8.75 m (mean ∼5 m). Some of them were cropped into several regions of interest, enlarging the image dataset to 200 images (90 MOC and 110 HiRISE). Fig. 1 shows the distribution of the initial set (before cropping procedure) of 124 images (white triangles) according to their center coordinates in the planetocentric coordinate system. The base map is the Mars Orbiter Laser Altimeter (MOLA) topo- graphy layer of the Mars Global Digital Dune Database. In order to infer the prevailing wind direction from dust devil tracks we have used one of the methods proposed by Statella et al. (2014), namely, the Directional Morphological Openings by Linear Structuring Elements method. The input for this method is a binary image showing detected dust devil tracks in white and the background in black. Dust devil tracks had been previously detected in the 200 images using the method proposed by Statella et. al (2012), with a global accuracy of 92% 75%. 3. Inferring the prevailing wind direction from dust devil tracks Dust devils are typically thought to move, on average, in the direction of the prevailing surface wind (Rennó et al.,1998). The prevailing wind direction can then be inferred by calculating the prevailing dust devil tracks direction in orbital images. For that purpose we have adopted one of the methods developed by Sta- tella et al. (2014), namely, the Directional Morphological Openings by Linear Structuring Elements. This method is based on Mathe- matical Morphology (MM), a theory developed by Georges Math- eron and Jean Serra in the 1960s, which is highly adequate to deal with the geometric features of structures. A few definitions of the theory, useful to fully explain the Directional Morphological Openings by Linear Structuring Elements method, are given below. Let us first define a digital image as follows: Let E be a none- mpty set of adjacent squares arranged in rows and columns, forming a rectangular surface. Let K be a set of gray levels. A gray level image is a mapping E- K. Usually, K belongs to the interval [0, K] in Z with E ∈ Z2. Then we can define the major mathematical morphology transformations, Erosion and Dilation: Let B be a subset from Z2, B ⊂ E. The Erosion (Soille, 2004), ε, of an image f by B is the minimum of the translation of f by the vectors –b of B. B is called Structuring Element (SE). ε ∧( ) = _ ( )∈ f f 1B b B b The Dilation (Soille, 2004), δ, of f by a SE B is the maximum of the translation of f by the vectors b of B. δ ∨( ) = ( )∈ f f 2B b B b The structuring element B is a completely defined and known (size and shape) set which is compared, in a transformation, to the image unknown set. The result of this transformation allows us to evaluate the unknown set. Based on Erosion and Dilation trans- formations we can define the morphological Opening. The Open- ing, γ, of f by a structuring element B is the erosion of f by B fol- lowed by a dilation by B transposed (Soille, 2004): T. Statella et al. / Planetary and Space Science 131 (2016) 60–6962 γ δ ε( ) = * ( ) ( )⎡⎣ ⎤⎦f f . 3B B B A morphological Opening will remove from the image all the regions of the image whose shape and size do not fully contain the structuring element. Linear Structurig Elements (LSEs) can be ar- ranged in several orientations so that a morphological Opening can provide information about the directions of binary sets in an digital image. Let us now define a set of directions α ∈ Zþ , mea- sured clockwise from North in steps of n degrees. For n ¼ 15°, we then have the set of directions α ¼ {0°, 15°,., 345°}. The direction opposed to α is denoted α′, so if α ¼ 0° then α′ ¼ 180°. The path in the direction α is defined as the union of the radii of the LSE in the directions d and d′. As an example, the path in the direction 45° is composed by the radii (with origin in a given pixel p) oriented to 45° and 225° (one can notice that the path in the direction 225° is identical to the path in the direction 45°, as both paths share the same radii). Now it is possible to define a family Lα,λ (with direc- tion α and size λ) of line SEs whose paths are oriented in the directions α ¼ {0°, 15°,., 165°}. As examples, LSEs with sizes λ¼1 in the directions 0°, 45°, 90° and 135° are: = = ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥B B 0 1 0 0 1 0 0 1 0 , 0 0 1 0 1 0 1 0 0 ,0 45 Fig. 2. Region of interest in HiRISE image PSP_006163_1345 (a); Automated tracks detec less modified by the LSE and thus the ones corresponding to the prevailing orientions) = = ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥B B 0 0 0 1 1 1 0 0 0 , 1 0 0 0 1 0 0 0 1 90 135 The choice for the angular steps of 15° derives from the digital approximations performed when building linear structuring ele- ments in a square digital grid, that is, increasing the number of angular steps after this level of detail would not provide much more meaningful information. The main advantage of this method is the high capability of directional Openings to estimate the prevailing orientation of dust devil tracks as Mathematical Morphology operators are crafted on the analysis of the geometry and topology of object features. After applying Openings with LSEs of size λ in the directions α ¼ {0°, 15°,., 165°}, the prevailing direction of the tracks is assumed to be the one in which the Opening has removed less pixels than in any other directions, that is, the direction where the structuring element better fits the tracks. The size λ of the LSEs varied from scene to scene (but, for each image, the size was fixed) according to the maximum width of the tracks in each scene. The width of the tracks in each scene is calculated as described in Statella et al. (2016). When plotting the results in a direction diagram, where each wing of the diagram represents the amount of pixels removed by the Opening, one should look for the smallest wing to determine tion (b); (c) Diagram of directions (note that the less frequented classes are the ones . [Original image credits: NASA/JPL/University of Arizona]. Fig. 3. Region of interest in MOC NA image E13-00271 (a); Automated tracks detection (b); (c) Diagram of directions (note that the less frequented classes are the ones less modified by the LSE and thus the ones corresponding to the prevailing orientions). [Original image credits: NASA/JPL/MSSS]. Table 1 Prevailing dust devil tracks directions for Aeolis quadrangle. Solar LongitudeLs[deg.] Season N° of images Prevailing direction [deg.] 0-30 Autumn 1 120-300 120-150 Winter 1 105-285 240-270 Spring 2 105-285/45-225 330-360 Summer 1 120-300 Table 2 Prevailing dust devil tracks directions for Argyre quadrangle. Solar LongitudeLs[deg.] Season N° of images Prevailing direction [deg.] T. Statella et al. / Planetary and Space Science 131 (2016) 60–69 63 the prevailing direction. Consider, as an example, a region of in- terest of the image HiRISE PSP_006163_1345 from Argyre quad- rangle shown in Fig. 2a. The binary image shown in Fig. 2b is the result of the dust devil tracks detection, which has been used as input for estimating the prevailing direction of the winds in that scene. In Fig. 2c we show the direction diagram for that image. The prevailing wind direction inferred from this image by the presented method is the less frequented one, that is, 30°–210°. Fig. 3 shows another example, this time for the MOC NA image E13-00271, also from Argyre quadrangle. As we can observe in Fig. 3a, tracks do not seem to follow any preferential direction as they do in HiRISE image PSP_006163_1345. Thus, we cannot infer, visually, any preferable direction. Nevertheless, after applying the directional morphological openings we could estimate the pre- vailing direction as being ∼30°–210°. 240-270 Spring 7 45-225 270-300 Summer 7 0-180 300-330 Summer 15 30-210 330-360 Summer 42 45-225 Table 3 Prevailing dust devil tracks directions for Noachis quadrangle. Solar LongitudeLs[deg.] Season N° of images Prevailing direction [deg.] 180-210 Spring 3 120-300/0-180/60-240 210-240 Spring 1 90-270 240-270 Spring 7 165-345 270-300 Summer 15 165-345 330-360 Summer 13 15-195 4. Results Morphological Openings with LSEs oriented in the directions α ¼ {0°, 15°,., 165°} have been applied to the image dataset in order to automatically infer the prevailing dust devil tracks direction. The size λ of the LSEs varied from scene to scene (but, for each image, the size was fixed) according to the maximum width of the tracks in each scene (this procedure has been adopted in order to avoid a possible influence of the width of the tracks in the results). Tracks width had been calculated as described in Statella et al. (2016). We show the results per region and per solar longitude Ls (in angular intervals of 30°) in the Tables 1–5, where it can be seen the varation in dust devil tracks directon with the season of the year. The solar longitude is the Mars-Sun angle, measured from the Northern Hemisphere spring equinox where Ls ¼ 0°. Intervals of 30° in solar longitude represent the passage of a martian month. Therefore, for the Northern Hemisphere we have: from Ls ¼ 0° to 90° the spring season (autumn for the Southern Hemisphere), from Ls ¼ 90° to 180° the summer season (winter for the Southern Hemisphere), from Ls ¼ 180° to 270° the autumn season (spring for the Southern Hemisphere) and from Ls ¼ 270° to 360° the winter season (summer for the Southern Hemisphere). The pre- vailing direction for a given interval in solar longitude, for example 0°-30°, has been adopted as being the modal value among all di- rections measured in all scenes whose solar longitudes fall in that interval. Table 4 Prevailing dust devil tracks directions for Hellas quadrangle. Solar LongitudeLs[deg.] Season N° of images Prevailing direction [deg.] 240-270 Spring 3 165-345/90-270/135-193 270-300 Summer 11 90-270 300-330 Summer 25 30-210 330-360 Summer 1 135-315 Table 5 Prevailing dust devil tracks directions for Eridania quadrangle. Solar LongitudeLs[deg.] Season N° of images Prevailing direction [deg.] 240-270 Spring 14 120-300 270-300 Summer 18 0-180 300-330 Summer 11 75-255/90-270/105-285 330-360 Summer 2 0-180 T. Statella et al. / Planetary and Space Science 131 (2016) 60–6964 In Table 1 we can see that the prevailing tracks directions fol- low a pattern SE-NW for solar longitudes intervals 0°–30° (au- tumn), 120°–270° (winter/spring) and 120°–300° (winter/spring/ summer). For solar longitudes ranging from 240° to 270° (spring) we found two dominant directions, which are 105°–285° and 45°– 225°. The large difference between those two directions is be- lieved to have been caused by the fact that we had only two images from Aeolis quadrangle. For Argyre quadrangle, whose results are shown in Table 2, we have found a pattern NE-SW for the orientation of the tracks. For Noachis quadrangle, as shown in Table 3, we have found different results for the prevailing tracks directions among the solar long- itude intervals. For solar longitudes ranging from 240° to 270° (spring) and from 270° to 300° (summer) a pattern SE-NW can be seen. From 180° to 210° (spring) the dominant tracks directions do not show a preferable direction and from 210° to 240° (spring), tracks directions inferred from the one image in that interval fol- lowed the E-W orientation. In Table 4, the prevailing dust devil tracks directions calculated for Hellas quadrangle were 90°–270° (E-W) for images acquired during the solar longitudes from 270° to 300° (summer), 30°–310° for images whose solar longitudes fall in the interval 300° to 330° (summer) and 135°–315° for the one image whose solar longitude falls in the interval 330° to 360°(summer). For the 240°–270° (spring) solar longitude interval the images showed no unique modal value. Instead, the tracks are mainly oriented in three directions. In Table 5 we show the results for Eridania quadrangle. For solar longitudes ranging from 270° to 300° (summer) and from 330° to 360° (summer) the tracks follow a E-W pattern. The pre- vailing directon changes for 120°–300° for images in the interval 240°–270° (spring) and we found three modal values for images with solar longitudes ranging from 300° to 330° (summer). In Tables 6–10 we show the directions calculated for each im- age of the quadrangles Aeolis, Argyre, Noachis, Hellas and Eridania, Table 6 Comparison between calculated prevailing wind direction and predicted wind direction ID Ls[deg.] Local Time Calculated [de E16-01962 18.77 13:39 120-300 R02-00357 134.17 14:40 105-285 PSP_003834_1650 242.6 15:14 120-300 R08-02402 249.25 14:27 105-285 R13-01467 331.03 13:31 120-300 the CPBL height, the solar longitude and the local time acquisition for each image and the directions of the winds predicted by the MCD. The MCD is a database of atmospheric statistics compiled from the Global Climate Model (GCM) and respective numerical simulations of the Martian atmosphere (Forget et al., 1999; Lewis et al., 1999). It was developed by the Laboratoire de Météorologie Dynamique (LMD, Paris), Atmospheric, Oceanic and Planetary Physics group (AOPP, Oxford), Department of Physics and Astron- omy (The Open University) and Instituto de Astrofísica de Anda- lucía (IAA, Granada) with the support of the European Space Agency and the Centre National d’Etudes Spatiales and the outputs provided by the MCD are based on large-scale climatological means. The extractions have been done according to the solar long- itude, acquisition date and time, latitude and longitude of each scene, at three different altitudes from surface: 50 m, 1000 m and at the Convective Planetary Boundary Layer (CPBL) height. The Martian Planetary Boundary Layer (PBL) is the portion of the at- mosphere closest to the surface, within which interactions be- tween the atmosphere and the surface itself are active. In practice, this represents the lowest portion of the atmosphere, within which surface-driven intense convection may take place, forming convective plumes and vortices during the day (Haberle et al., 1993; Hinson et al., 2008; Petrosyan et al., 2011). The height of the PBL and, in particular, the height of daytime CPBL, is a key quantity describing the vigor of convective activity, that changes over the surface according to the local thickness of the atmosphere layer. Examples of estimated PBL heights inferred from lander measurements, from heights of convective clouds, from temperature profiles or radio occultation are given by Fenton and Lorenz (2015) and a more extensive study is given by Hinson et al. (2008). We have chosen the altitudes in our extractions from the MCD so that they match the lower CPBL, which is 1000 m from the surface, and the upper CPBL limit, which has been obtained from the MCD for each image center coordinates. Fenton and Lorenz (2015) have performed a study on Amazonis Planitia and have found out that the CPBL thickness can be estimated as being ∼5 times the median of the active dust devil plumes, after analyzing MRO CTX images. As we are rather working with dust devil tracks instead of plumes, we cannot infer such relationship and, there- fore, we have decided to consider the upper limit of the CPBL in the experments. In addition, we have run the predictions at 50 m from the surface in order to see if the wind directions could differ significantly from those in the CPBL and if the portion of the at- mosphere closer to the surface could play a more dominant role in the directions of the dust devils. For the experiments we have used the MCD version 5.2 and we have adopted the climatology average solar dust scenario for it is the most representative of a standard condition. Running the extractions with Mars Years (MY) scenarios corresponding to the acquisition date of the images in a sample of the dataset did not show significant (i. e., were not larger than the 15° of the precision of the method used to calculate directions in the images) differences when compared with the climatology for Aeolis. g.] CPBL [m] Predicted [deg.] 50 m 1000 m CPBL 4500 45-225 45-225 30-210 5500 30-210 15-195 0-180 6300 90-270 90-270 75-255 6300 105-285 90-270 90-270 5600 60-240 60-240 45-225 Table 7 Comparison between calculated prevailing wind direction and predicted wind direction for Argyre. ID Ls[deg.] Local Time Calculated [deg.] CPBL [m] Predicted [deg.] 50 m 1000 m CPBL M10-01206 260.85 14:09 165-345 4800 120-300 120-300 0-180 S08-02952 257.68 14:24 90-270 6400 30-210 15-195 30-210 R08-02621_P1 250.90 14:40 0-180 6300 45-225 30-210 30-210 R08-02621_P2 250.90 14:40 120-300 6300 45-225 30-210 30-210 S08-03151 259.03 14:18 45-225 5000 0-180 0-180 0-180 ESP_012927_1245 256.30 15:24 60-240 4500 75-255 75-255 60-240 ESP_013204_1260 270.00 15:11 45-225 4600 90-270 75-255 75-255 S10-01582 297.27 13:52 75-255 6200 165-345 165-345 0-180 S10-01598 297.32 13:46 15-195 4600 30-210 30-210 30-210 ESP_013310_1200_P1 275.10 15:07 165-345 6200 0-180 0-180 165-345 ESP_013310_1200_P2 275.10 15:07 75-255 6200 0-180 0-180 165-345 ESP_013626_1245 290.40 15:02 120-300 4700 120-300 105-285 105-345 ESP_013520_1180_P1 285.30 14:58 120-300 6400 120-300 120-300 165-345 ESP_013520_1180_P2 285.30 14:58 105-285 6400 120-300 120-300 165-345 E13-00271_P1 320.78 13:28 30-210 5600 120-300 120-300 150-330 E13-00271 _P2 320.78 13:28 75-255 5600 120-300 120-300 150-330 M14-00175_P1 329.22 13:22 0-180 4200 30-210 30-210 75-255 M14-00175_P2 329.22 13:22 105-285 4200 30-210 30-210 75-255 M12-02214 305.76 13:30 15-198 4300 60-240 60-240 60-240 ESP_013996_1155 307.60 14:42 30-210 5700 90-270 90-270 135-315 ESP_014049_1200 310.10 14:38 45-225 5700 45-225 45-225 15-195 ESP_014259_1230_P1 319.50 14:32 0-180 4600 135-315 120-300 90-270 ESP_014259_1230_P2 319.50 14:32 30-210 4600 135-315 120-300 90-270 PSP_005596_1245 326.20 14:28 90-270 4200 90-270 90-270 90-270 PSP_005397_1270_P1 317.40 14:29 60-240 6000 105-285 105-285 135-315 PSP_005397_1270_P2 317.40 14:29 105-285 6000 105-285 105-285 135-315 PSP_005397_1270_P3 317.40 14:29 75-255 6000 105-285 105-285 135-315 PSP_005397_1270_P4 317.40 14:29 75-255 6000 105-285 105-285 135-315 PSP_005397_1270_P5 317.40 14:29 90-270 6000 105-285 105-285 135-315 E14-00400_P1 339.47 13:26 0-180 4100 30-210 30-210 75-255 E14-00400_P2 339.47 13:26 135-315 4100 30-210 30-210 75-255 E14-00400_P3 339.47 13:26 0-180 4100 30-210 30-210 75-255 R13-02691 334.68 13:45 0-180 5300 105-285 105-285 120-300 PSP_005820_1320_P1 335.70 14:25 30-210 4200 30-210 30-210 60-240 PSP_005820_1320_P2 335.70 14:25 45-225 4200 30-210 30-210 60-240 PSP_005820_1320_P3 335.70 14:25 45-225 4200 30-210 30-210 60-240 PSP_005820_1320_P4 335.70 14:25 45-225 4200 30-210 30-210 60-240 PSP_005820_1320_P5 335.70 14:25 45-225 4200 30-210 30-210 60-240 PSP_005820_1320_P6 335.70 14:25 30-210 4200 30-210 30-210 60-240 PSP_005820_1320_P7 335.70 14:25 165-345 4200 30-210 30-210 60-240 PSP_005820_1320_P8 335.70 14:25 45-225 4200 30-210 30-210 60-240 PSP_005846_1235_P1 336.80 14:29 75-255 5400 105-285 105-285 150-330 PSP_005846_1235_P2 336.80 14:29 75-255 5400 105-285 105-285 150-330 PSP_005846_1235_P3 336.80 14:29 105-285 5400 105-285 105-285 150-330 PSP_005846_1235_P4 336.80 14:29 105-285 5400 105-285 105-285 150-330 PSP_005846_1235_P5 336.80 14:29 90-270 5400 105-285 105-285 150-330 PSP_005846_1235_P6 336.80 14:29 105-285 5400 105-285 105-285 150-330 PSP_005846_1235_P7 336.80 14:29 105-285 5400 105-285 105-285 150-330 PSP_005846_1235_P8 336.80 14:29 0-180 5400 105-285 105-285 150-330 PSP_005846_1235_P9 336.80 14:29 45-225 5400 105-285 105-285 150-330 PSP_005846_1235_P10 336.80 14:29 105-285 5400 105-285 105-285 150-330 PSP_005846_1235_P11 336.80 14:29 75-255 5400 105-285 105-285 150-330 PSP_005846_1235_P12 336.80 14:29 60-240 5400 105-285 105-285 150-330 PSP_005846_1235_P13 336.80 14:29 90-270 5400 105-285 105-285 150-330 PSP_005846_1235_P14 336.80 14:29 120-300 5400 105-285 105-285 150-330 PSP_005846_1235_P15 336.80 14:29 165-345 5400 105-285 105-285 150-330 PSP_005846_1235_P16 336.80 14:29 30-210 5400 105-285 105-285 150-330 PSP_005846_1235_P17 336.80 14:29 120-300 5400 105-285 105-285 150-330 PSP_005846_1235_P18 336.80 14:29 45-225 5400 105-285 105-285 150-330 PSP_005846_1235_P19 336.80 14:29 60-240 5400 105-285 105-285 150-330 PSP_005846_1235_P20 336.80 14:29 15-195 5400 105-285 105-285 150-330 PSP_005846_1235_P21 336.80 14:29 90-270 5400 105-285 105-285 150-330 PSP_005846_1235_P22 336.80 14:29 15-195 5400 105-285 105-285 150-330 PSP_006163_1345_P1 349.80 14:28 30-210 4200 0-180 0-180 15-195 PSP_006163_1345_P2 349.80 14:28 30-210 4200 0-180 0-180 15-195 PSP_006163_1345_P3 349.80 14:28 45-225 4200 0-180 0-180 15-195 PSP_006163_1345_P4 349.80 14:28 60-240 4200 0-180 0-180 15-195 PSP_006176_1225_P1 350.40 14:34 135-315 4200 105-285 105-285 90-270 PSP_006176_1225_P2 350.40 14:34 120-300 4200 105-285 105-285 90-270 PSP_006176_1225_P3 350.40 14:34 150-330 4200 105-285 105-285 90-270 PSP_005780_1215 334.00 14:30 105-285 5500 105-285 105-285 105-285 T. Statella et al. / Planetary and Space Science 131 (2016) 60–69 65 Table 8 Comparison between calculated prevailing wind direction and predicted wind direction for Noachis. ID Ls[deg.] Local Time Calculated [deg.] CPBL [m] Predicted [deg.] 50 m 1000 m CPBL PSP_002548_1255_P1 181.80 16:00 120-300 2600 60-240 75-255 75-255 PSP_002548_1255_P2 181.80 16:00 60-240 2600 60-240 75-255 75-255 PSP_002548_1255_P3 181.80 16:00 120-300 2600 60-240 75-255 75-255 PSP_003326_1255 217.80 15:53 75-255 4900 75-255 75-255 105-285 PSP_004038_1255 257.70 15:20 75-255 6300 135-315 135-315 75-255 PSP_004249_1255 263.10 15:11 15-195 6600 135-315 135-315 60-240 R0903467 267.72 14:22 135-315 6100 165-345 165-345 150-330 S09-01660_P1 269.93 14:12 165-345 6200 165-345 165-345 150-330 S09-01660_P2 269.93 14:12 165-345 6200 165-345 165-345 150-330 S09-00929_P1 265.39 14:17 0-180 6100 165-345 165-345 150-330 S09-00929_P2 265.39 14:17 165-345 6100 165-345 165-345 150-330 ESP_013321_1175 275.70 15:08 135-315 6000 150-330 150-330 120-300 ESP_013557_1245 287.10 14:52 165-345 6400 165-345 165-345 135-315 E11-01722 287.99 13:52 30-210 5900 165-345 165-345 0-180 E11-03103 293.62 13:48 0-180 5700 165-345 165-345 0-180 E11-00747 284.82 13:55 30-210 6100 150-330 150-330 0-180 E11-01129 286.08 13:53 90-270 6400 150-330 150-330 0-180 E11-01527 287.34 13:51 60-240 6400 150-330 150-330 165-345 E11-03844 297.33 13:43 30-210 5900 150-330 150-330 165-345 E11-02963 292.98 13:46 15-195 6100 150-330 150-330 165-345 E11-00582 284.17 13:54 45-225 6400 150-330 150-330 165-345 E11-01314 286.68 13:51 0-180 6300 165-345 165-345 165-345 R10-04224 285.72 14:08 165-345 6100 150-330 150-330 165-345 R10-02844 281.23 14:08 165-345 6300 165-345 165-345 150-330 R10-04196 285.67 14:03 30-210 6400 165-345 165-345 150-330 R10-00382 272.25 14:17 150-330 6200 165-345 165-345 150-330 R11-03714 304.30 13:50 15-195 5500 165-345 150-330 135-315 PSP_005238_1255 310.30 14:33 120-300 6400 120-300 135-315 150-330 PSP_005383_1255 316.80 14:26 0-180 6100 120-300 120-300 150-330 PSP_005528_1255 323.20 14:20 105-285 5900 105-285 120-300 135-315 ESP_013992_1170_P1 307.50 14:41 0-180 5600 150-330 150-330 150-330 ESP_013992_1170_P2 307.50 14:41 15-195 5600 150-330 150-330 150-330 ESP_013992_1170_P3 307.50 14:41 30-210 5600 150-330 150-330 150-330 ESP_014020_1150 308.70 14:42 30-210 5300 165-345 165-345 165-345 ESP_014322_1215 322.30 14:38 15-195 5300 135-315 135-315 120-300 PSP_005659_1335_P1 328.90 14:24 15-195 6000 75-255 75-255 90-270 PSP_005659_1335_P2 328.90 14:24 0-180 6000 75-255 75-255 90-270 PSP_005659_1335_P3 328.90 14:24 0-180 6000 75-255 75-255 90-270 E12-01041 306.48 13:38 150-330 5400 165-345 165-345 150-330 T. Statella et al. / Planetary and Space Science 131 (2016) 60–6966 average solar dust scenario. As the MYs scenarios are produced from irregularly observations, which are subsequently gridded into a regular network by kriging, and as we had no means to assess and compare the accuracy of the MYs scenarios against the aver- age solar dust scenario, we have run the experiments with this last scenario. Regarding the time used for the extractions, we have chosen it to match each image acquisition time. As we cannot know the exact time of dust devil formation, some approximated time value should be adopted. In general, the image acquisition time is very close the peak time of the formation of the convective plumes, therefore, we have adopted that time for the extractions from the MCD. Experiments carried out on a subsample set of images showed that extractions from an interval of 71 hour centered at the time of acquisition did not result in differences larger than the precision of the method used for calculating the wind direction from the tracks. Information about winds from the MCD is based on large-scale (computed for regions of a few degrees in latitude and longitude) climatological means, therefore, the locally prevailing winds that drive the dust devil can be expected to be influenced by regional and local conditions which may not be resolved at the grid scale of the MCD. In the extractions, the MCD provides wind speeds defined by two components: the meridional SN wind (positive when oriented from south to north) and the zonal WE wind (positive when or- iented from west to east) in m/s. The resulting wind direction has been defined as: ( )α = ( )− WE SNtan / 41 for WE 4 0 and SN 4 0. If WE 4 0 and SN o 0, then α ¼ 180° - | α|; ifWE o 0 and SN o 0, then α ¼ 180° þ |α|; and ifWE o 0 and SN 4 0, then α ¼ 360° - |α|. Albeit MOC NA and HiRISE images are projected in Sinusoidal and Equirectangular systems, respectively, which are not con- formal projections, the errors involved in estimating directions from that remotely sensed data can be considered negletable in the present case. Statella (2015) showed that the maximum an- gular distortion for a typical HiRISE image is ∼0.6°. And for a ty- pical MOC NA image it is ∼0.02°, even though meridians are not straight lines in the Sinusoidal projection. For the comparison between the wind directions inferred from dust devil tracks and the predicted wind directions from the MCD we have considered a tolerance of 15° (which was the angular step used in the directions measurements) for the differences obtained. Therefore, when the difference between the calculated dust devil track prevailing direction and the MCD predicted wind direction differed only by one angular bin class (15º), we considered both directions as being the same. In Table 6, where we compare the wind directions inferred from dust devil tracks with the predicted wind directions for Aeolis quadrangle, only 20% of the values agree for the three altitude values. The results obtained for the other 4 quadrangles Table 9 Comparison between calculated prevailing wind direction and predicted wind direction for Hellas. ID Ls[deg.] Local Time Calculated [deg.] CPBL [m] Predicted [deg.] 50 m 1000 m CPBL E10-00012_C1 263.49 14:17 165-345 5400 120-300 120-300 120-300 E10-00578_C 266.02 14:18 150-330 6700 15-195 0-180 0-180 S09-01218_C 267.20 14:17 135-315 7300 165-345 0-180 0-180 E10-02468_C3 273.23 13:58 90-270 4800 60-240 60-240 60-240 E10-02820_C 274.51 13:56 60-240 4900 60-240 60-240 60-240 E11-01296_C 286.63 13:53 165-345 6300 150-330 150-330 150-330 E11-00727_C1 284.72 13:57 45-225 5800 150-330 150-330 165-345 E11-00727_C2 284.72 13:57 0-180 5800 150-330 150-330 165-345 E11-00727_C3 284.72 13:57 150-330 5800 150-330 150-330 165-345 E10-04932_C 282.14 13:57 105-285 6200 105-285 105-285 165-345 E10-01933_C 271.23 14:07 0-180 6600 120-300 120-300 150-330 E11-00173_C 282.75 14:00 60-240 7200 165-345 165-345 165-345 E10-02444_C 273.12 14:02 150-330 6600 45-225 45-225 15-195 E11-01270_C 286.53 13:51 45-225 7000 0-180 0-180 15-195 E12-01277_C1 307.63 13:27 75-255 4800 60-240 60-240 45-225 E12-01277_C2 307.63 13:27 75-255 4800 60-240 60-240 45-225 E12-01277_C3 307.63 13:27 105-285 4800 60-240 60-240 45-225 E12-01752_C 310.02 13:36 30-210 5200 150-330 150-330 150-330 E12-00190_C1 302.08 13:30 30-210 5800 165-345 165-345 120-300 E12-00190_C2 302.08 13:30 120-300 5800 165-345 165-345 120-300 E1301622_C 329.28 13:23 75-255 4200 60-240 60-240 60-240 E13-00975_C 325.28 13:23 60-240 4300 60-240 60-240 60-240 R12-01589_C 316.21 13:50 105-185 5100 150-330 150-330 150-330 R1201264_C 314.38 13:49 15-195 6300 150-330 150-330 165-345 ESP_014399_1220_C 325.60 14:34 75-255 5700 120-300 120-300 120-300 ESP_014176_1155_C1 315.80 14:34 165-345 5100 150-330 150-330 150-330 ESP_014176_1155_C2 315.80 14:34 15-195 5100 150-330 150-330 150-330 ESP_014176_1155_C3 315.80 14:34 45-225 5100 150-330 150-330 150-330 ESP_014176_1155_C4 315.80 14:34 150-330 5100 150-330 150-330 150-330 ESP_014176_1155_C5 315.80 14:34 150-330 5100 150-330 150-330 150-330 ESP_014176_1155_C6 315.80 14:34 0-180 5100 150-330 150-330 150-330 ESP_014108_1200_C1 312.70 14:37 30-210 6800 150-330 150-330 150-330 ESP_014108_1200_C2 312.70 14:37 120-300 5100 150-330 150-330 150-330 ESP_014070_1170_C 311.00 14:43 90-270 5500 120-300 120-300 135-315 ESP_014069_1180_C 311.00 14:39 30-210 6700 135-315 135-315 150-330 ESP_014056_1180_C 310.40 14:40 75-255 6800 120-300 120-300 150-330 ESP_014004_1180_C 308.00 14:41 120-300 5600 120-300 120-300 150-330 ESP_013991_1160_C 307.40 14:43 120-300 5500 135-315 135-315 150-330 ESP_013965_1165_C 306.20 14:43 150-330 5600 135-315 135-315 150-330 PSP_006264_1420_C 359.90 14:28 45-235 4000 45-225 45-225 45-225 T. Statella et al. / Planetary and Space Science 131 (2016) 60–69 67 are similar, not showing a high percentage of concordance be- tween calculated and predicted wind directions. For Argyre (Table 7), we have found that ∼39% of the predictions at 50 m and at 1000 m, and ∼20% of the predictions at the CPBL height agreed with the measurements. For Noachis (Table 8), we have found that ∼33% of the predictions at 50 m and at 1000 m, and ∼31% of the predictions at the CPBL height agreed with the measurements. For Hellas (Table 9), we have found that ∼35% of the predictions at 50 m and at 1000 m, and ∼25% of the pre- dictions at the CPBL height agreed with the measurements. Fi- naly, we have found, for Eridania (Table 10), that ∼18% of the predictions at 50 m, ∼22% at 1000 m, and ∼27% of the predic- tions at the CPBL height agreed with the measurements. The diverging results for Argyre, when compared to those from Statella et al. (2014), are due to the coarser precision of the measured directions. In Statella et al. (2014) angular steps of 45° were adopted, instead of 15° used here. As the measurement details in the latter were coarser, the prevailing wind direction from the MCD in that case was considered to be the one showing a higher wind speed between N-S and EW components. When we considered the whole set of 200 images, only 32%, 33% and 25% of the extractions at 50 m, 1000 m and at the CPBL height, respectively, agreed with the dust devil tracks prevailing direction measurements. Table 11 summarizes the results for each quadrangle and also for the whole set of images. In addition, we have extracted the surface wind stress using the MCD for the whole set of images. According to Haberle et al. (2003), a threshold stress of 0.0225 Nm-2 would be needed to lift dust through bombardment from sand saltation. This has also been found to be true by Kahre et al. (2006), which used the NASA Ames general circulation model to investigate the mechanisms responsible for the observed martian dust cycle. According to their results, dust devil and wind stress lifting should both contribute to the simulated amount of dust found on the atmosphere. In our results, in none of the cases the large scale wind surface stress was higher than that threshold. The highest value we have found was ∼ 0.0108 Nm-2, extracted from the MCD using the local time 16:00h, φ∼54°S, λ∼13°E and Ls∼182°. In such cases, the suction effect caused by the dust devils low-pressure cores must have assisted the wind stress in order to lift particles, as it has already been stated by Greeley et al. (1981), Greeley and Iversen (1985) and Greeley et al. (2003). 5. Conclusions We have applied an automated method based on directional morphological openings to infer the prevailing dust devil tracks direction to a dataset of 200 images from Aeolis, Argyre, Eridania, Noachis and Hellas quadrangles. Next, we have compared the Table 10 Comparison between calculated prevailing wind direction and predicted wind direction for Eridania. ID Ls[deg.] Local Time Calculated [deg.] CPBL [m] Predicted [deg.] 50 m 1000 m CPBL E10-00563 265.98 14:06 105-285 7200 120-300 120-300 105-285 E09-02016 259.51 14:16 165-345 6600 105-285 105-285 60-240 R09-03051 266.28 14:23 150-330 6700 15-195 15-195 15-195 R09-01707 261.74 14:28 150-330 6600 30-210 30-210 30-210 ESP_021874_1175_P1 262.80 15:13 135-315 5700 15-195 0-180 150-330 ESP_021874_1175_P2 262.80 15:13 120-300 5700 15-195 0-180 150-330 ESP_021874_1175_P3 262.80 15:13 105-285 5700 15-195 0-180 150-330 ESP_021939_1170 266.00 15:10 75-255 6200 105-285 120-300 150-330 ESP_021940_1205_P1 266.00 15:07 45-225 5900 0-180 165-345 165-345 ESP_021940_1205_P2 266.00 15:07 120-300 5900 0-180 165-345 165-345 ESP_021940_1205_P3 266.00 15:07 120-300 5900 0-180 165-345 165-345 PSP_004086_1180_P1 255.10 15:25 165-345 5200 30-210 30-210 165-345 PSP_004086_1180_P2 255.10 15:25 60-240 5200 30-210 30-210 165-345 PSP_004086_1180_P3 255.10 15:25 45-225 5200 30-210 30-210 165-345 E10-04027 278.89 14:02 15-195 7000 0-180 0-180 0-180 E11-01652 287.74 13:46 30-210 6500 165-345 0-180 30-210 E11-01664 287.79 13:53 0-180 7200 0-180 0-180 0-180 E11-02785 292.18 13:46 165-345 6800 0-180 0-180 0-180 E11-02787 292.18 13:45 0-180 6500 0-180 0-180 0-180 E11-02045 289.04 13:40 0-180 6000 165-345 165-345 120-300 R09-04267_P1 270.71 14:16 60-240 6300 105-285 105-285 45-225 R09-04267_P2 270.71 14:16 150-330 6300 105-285 105-285 45-225 R10-04120 285.52 14:07 135-315 7700 0-180 0-180 0-180 R11-02327 298.55 13:50 45-225 6000 30-210 30-210 0-180 R10-03758 284.16 14:10 135-315 6800 165-345 165-345 0-180 R10-02441 279.70 14:15 165-345 6800 165-345 165-345 165-345 R10-03959_P1 284.82 14:10 120-300 6600 165-345 165-345 0-180 R10-03959_P2 284.82 14:10 120-300 6600 165-345 165-345 0-180 R10-04815 287.98 14:06 15-195 6700 165-345 165-345 0-180 R10-03563_P1 283.55 14:12 45-225 6400 165-345 165-345 0-180 R10-03563_P2 283.55 14:12 135-315 6400 165-345 165-345 0-180 R11-02811 299.83 13:51 0-180 6800 135-315 135-315 0-180 E11-04510 300.81 13:36 165-345 6100 120 135-315 0-180 R12-00250 309.59 13:43 30-210 6200 105-285 120-300 0-180 R12-02283 319.06 13:40 15-195 5800 90-270 105-285 165-345 PSP_005510_1290 322.40 14:25 0-180 6200 60-240 75-255 120-300 ESP_023021_1160_P1 317.10 14:30 90-270 5300 120-300 135-315 150-330 ESP_023021_1160_P2 317.10 14:30 75-255 5300 120-300 135-315 150-330 ESP_023021_1160_P3 317.10 14:30 90-270 5300 120-300 135-315 150-330 ESP_023021_1160_P4 317.10 14:30 105-285 5300 120-300 135-315 150-330 ESP_023021_1160_P5 317.10 14:30 105-285 5300 120-300 135-315 150-330 ESP_023021_1160_P6 317.10 14:30 75-255 5300 120-300 135-315 150-330 ESP_014121_1180 313.30 14:38 150-330 6600 165-345 165-345 150-330 R13-01492 331.07 13:48 15-195 5100 45-225 45-225 120-300 PSP_006248_1235 353.20 14:34 0-180 5100 105-285 105-285 105-285 Table 11 Summary of the results for the comparison between measured and predicted wind directions for each of the quadrangles and for the whole set of images. Number of images Altitude 50 m 1000 m CPBL Aeolis 5 20% 20% 20% Argyre 71 39% 39% 20% Noachis 39 33% 33% 31% Hellas 40 35% 35% 25% Eridania 45 18% 22% 27% For the whole dataset 200 32% 33% 25% T. Statella et al. / Planetary and Space Science 131 (2016) 60–6968 inferred directions with the MCD predicted wind directions for different altitudes from the surface: 50 m, 1000 m and at the CPBL height. In all cases the accordance between measurements and predictions is low. In the best case (for 1000 m above the surface), this value reaches 33%, considering a tolerance of 15° (the angular steps adopted when measuring dust devil tracks directions). All the results are very similar, with the prediction performed using the CPBL height giving the worst agreement with the dust devil tracks prevailing direction measurements. Taken individually, none of the quadrangles showed a conformity between measured and predicted directions higher than 39%. This seems to be a good indicator that the MCD fails to predict local wind patterns, such as the ones driving dust devil vortices on the surface of Mars, per- haps due to an insufficient resolution. Therefore, for a more local scale, the dominant wind direction should be inferred preferably by methods other than the MCD simulations, namely from the analysis of the dust devil tracks prevailing orientations. For that purpose, we intend to perform a systematic collection of image data showing dust devil tracks, specifically from several re- presentative sites in several time periods, aided by recent plat- forms that facilitate checking about the availability of multi-tem- poral imagery of high resolution in order to quantify the changes of these dynamic surface processes on Mars (Sidiropoulos and Muller, 2015; Erkeling et al., 2016). In addition, the extractions for the surface wind stress at each scene center coordinates was never higher than the threshold stress of 0.0225 Nm-2 needed to lift dust even though there are dust devil tracks. It agrees with results from Greeley et al. (1981), Greeley and Iversen (1985) and Kahre et al. (2006) and reinforces T. Statella et al. / Planetary and Space Science 131 (2016) 60–69 69 that dust devil vortices, as well as other mechanisms such as slope winds (Rafkin et al., 2002) and rocket dust storms (Spiga et al., 2013), are important actors in lifting dust particles, contributing to the local atmosphere opacity and surface modification. References Balme, M.R., Greeley, R., 2006. Dust devils on Earth and Mars. ReviewsofGeophysics 44, RG3003. Balme, M.R., Whelley, P.L., Greeley, R., 2003. Mars: dust devil track survey in Argyre Planitia and Hellas Basin. J. Geophys. Res. 108 (E8). Erkeling, G., Luesebrink, D., Hiesinger, H., Reiss, D., Heyer, T., Jaumann, R., 2016. The Multi‐Temporal Database of Planetary Image Data (MUTED): A database to support the identification of surface changes and short-lived surface processes. Planet. SpaceSci. http://dx.doi.org/10.1016/j.pss.2016.03.002 Fenton, L.K., Lorenz, R., 2015. Dust devil height and spacing with relation to the martian planetary layer thickness. Icarus 260, 246–262. Forget, F., Hourdin, F., Fournier, R., Hourdin, C., Talagrand, O., Collins, M., Lewis, S.R., Read, P.L., Huot, J.-P., 1999. Improved general circulation models of the martian atmosphere from the surface to above 80 km. J. Geophys. Res. 104 (E10), 24155–24175. Greeley, R., Iversen, J., 1985. Wind as a Geologic Process on Earth, Mars, Venus and Titan. Cambridge Univ. Press, New York. Greeley, R., White, B., Pollack, J.B., Iversen, J., Leach, R.N., 1981. Dust storms on Mars: Considerations and simulations, in Desert Dust: Origin, Characteristics, and Effect on Man edited by T. Péwé. Spec. Pap. Geol. Soc. Am 186, 101–121. Greeley, R., et al., 2003. Martian dust devils: Laboratory simulations of particle threshold. J. Geophys. Res. 108, E5. Haberle, R., Houben, H., Hertenstein, R., Herdtle, T., 1993. A boundary layer model for Mars: Comparison with Viking lander and entry data. J. Atmos. Sci. 50, 1544–1559. Haberle, M.H., Murphy, J.R., Schaeffer, J., 2003. Orbital change experiments with a Mars general circulation model. Icarus 161, 66–89. Hinson, D.P., Pätzold, M., Tellmann, S., Häusler, B., Tyler, G.L., 2008. The depth of the convective boundary layer on Mars. Icarus 198, 57–66. Kahre, M.A., Murphy, J.R., Haberle, R.M., 2006. Modeling the Martian dust cycle and surface dust reservoirs with the NASA Ames general circulation model. J. Geophys. Res. Planets 111, 10.1029/2005je002588. Lewis, S.R., Collins, M., Read, P.L., Forget, F., Hourdin, F., Fournier, R., Hourdin, C., Talagrand, O., Huot, J.P., 1999. A climate database for Mars. J. Geophys. Res. Planets 104, E10. Malin, M.C., Edgett, K.S., 2001. Mars Global Surveyor Mars Orbiter Camera: inter- planetary cruise through primary mission. J. Geophys. Res. 106, E10. Petrosyan, A., Galperin, B., Larsen, S.E., Lewis, S.R., Määttänen, A., Read, P.L., Renno, N., Rogberg, L.P.H.T., Savijärvi, H., Siili, T., Spiga, A., Toigo, A., Vázquez, L., 2011. The Martian armospheric boundary layer. ReviewsofGeophysics 49, RG3005. http://dx.doi.org/10.1029/2010RG000351. Rafkin, S.C.R., Maria, M.R.V.S., Michaels, T.I., 2002. Simulation of the atmospheric thermal circulation of a martian volcano using a mesoscale numerical model. Nature 419, 697–699. Rennó, N.O., Burkett, M.L., Larkin, M.P., 1998. A simple thermodynamical theory for dust devils. J. Atmos. Sci. 55, 3244–3252. Sidiropoulos, P., Muller, J.-P., 2015. On the status of orbital high-resolution repeat imaging of Mars for the observation of dynamic surface processes. Planet. SpaceSci. 117, 207–222. Soille, P., 2004. Morphological Image Analysis. Springer-Verlag, Berlin. Spiga, A., Faure, J., Madeleine, J.B., Maatanen, A., Forget, F., 2013. Rocket dust storms and detached dust layers in the martian atmosphere. J. Geophys. Res. ((Planets)) 118, 746–767. Statella, T., Pina, P., Silva, E.A., 2012. Image processing algorithm for the identifi- cation of Martian dust devil tracks in MOC and HiRISE images. Planet. Space. Sci. 70, 46–58. Statella, T., Pina, P., Silva, E.A., 2014. Automated determination of the orientation of dust devil tracks in Mars orbiter images. Adv. SpaceRes. 53, 1822–1833. Statella, T., 2015. Mapping Mars: Geodetic and cartographic aspects. Planet. SpaceSci. 108, 1–12. Statella, T., Pina, P., Silva, E.A., 2015. Extensive computation of albedo contrast be- tween Martian dust devil tracks and their neighboring regions. Icarus 250, 43–52. Statella, T., Pina, P., Silva, E.A., 2016. Automated width measurements of Martian dust devil tracks. AeolianResearch 20, 1–6. Thomas, P.C., Gierasch, P., Sullivan, R., Miller, D.S., Cartillo, E.A., Cantor, B., Mellon, M.T., 2003. Mesoscale linear streaks on Mars: environments of dust entrain- ment. Icarus 162, 242–258. Whelley, P., Greeley, R., 2006. Latitudinal dependency in dust devil activity on Mars. J. Geophys. Res. 111, E10003. http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref1 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref1 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref2 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref2 http://dx.doi.org/10.1016/j.pss.2016.03.002 http://dx.doi.org/10.1016/j.pss.2016.03.002 http://dx.doi.org/10.1016/j.pss.2016.03.002 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref4 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref4 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref4 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref5 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref5 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref5 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref5 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref5 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref6 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref6 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref7 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref7 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref7 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref7 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref8 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref8 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref9 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref9 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref9 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref9 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref10 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref10 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref10 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref11 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref11 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref11 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref12 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref12 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref12 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref13 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref13 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref13 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref14 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref14 http://dx.doi.org/10.1029/2010RG000351 http://dx.doi.org/10.1029/2010RG000351 http://dx.doi.org/10.1029/2010RG000351 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref16 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref16 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref16 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref16 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref17 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref17 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref17 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref18 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref18 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref18 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref18 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref19 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref20 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref20 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref20 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref20 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref21 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref21 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref21 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref21 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref22 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref22 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref22 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref23 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref23 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref23 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref24 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref24 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref24 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref24 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref25 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref25 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref25 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref26 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref26 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref26 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref26 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref27 http://refhub.elsevier.com/S0032-0633(15)30131-8/sbref27 Comparing wind directions inferred from Martian dust devil tracks analysis with those predicted by the Mars Climate... Introduction Image Dataset Inferring the prevailing wind direction from dust devil tracks Results Conclusions References