Comprehensive studies of the stratigraphic relationships in the Athabasca Valles region have been performed on imagery from the Viking Orbiters and the Mars Orbiter Camera (narrow and wide angle) on board Mars Global Surveyor as well as by making use of topographic data derived from the Mars Global Surveyor Laser Altimeter (MOLA) instrument. From superimposed crater frequencies we have derived absolute surface ages by applying the Hartmann and Neukum  cratering chronology model. The main channel has been incised into the Cerberus volcanic plains with an average plains age of 3.4 Ga. The main fluvial or glacial erosion processes ended 2.6 Ga ago. This result shows that the age of the valley system itself is older than commonly believed. One major possible fluvial event occurred in the topographically lower volcanic plains southwest of the valley 1.6 Ga ago. The valley itself was covered by lava episodically and ending 0.9 Ga and in case of a few younger episodes 30 Ma ago. The surface texture south of the valley system suggests a younger, possibly fluvial overprinting of the volcanic texture 30 Ma ago. The latest volcanic activity is dated to about 3 Ma ago. With the latter ages we have been able not only to confirm earlier age estimates by Berman and Hartmann  but also to show that the valley system itself has undergone over a period of two billion years active geological cycles which were dominated by volcanic processes at the end.
 The Athabasca Valles system is believed to be an outflow channel system that cut the volcanic Cerberus Planitia, part of the Elysium Planitia. The area of interest has been mapped by Greeley and Guest  as primarily younger channel and flood-plain material (unit Achu) of Amazonian age, following Scott and Tanaka . They have interpreted it as fluvial deposits, where distinct albedo patterns probably represent channels with bars and islands. Especially in the western part mottled zones represent deposition from ponded terminus of fluvial systems. Contrary to this interpretation, Plescia  describes the Cerberus Formation to be of volcanic origin, because corresponding surface morphology, lobate edges of the unit, and the embayment relation of the unit with adjacent older units can be recognized. Low-viscosity lavas from the Elysium volcano group northwest of the plains have flooded this region and filled the topographic depression. Plescia's  interpretation is supported by Schaber , who describes the radar and thermal characteristics to be similar to those which are interpreted as flood basalt provinces, such as Syrtis Major.
 The Athabasca Valles region is located southeast of the Elysium volcano complex between 151–159°E and 0–12°N. An overview of the context of the investigated area is given in Figure 1. The main valley strikes in NE–SW direction, but also discharges to the Cerberus Plains at an elevation of about −2700 m in southeast direction following the overall topography. The possible origin of the valley correlates with the Cerberus Fossae, a set of subparallel grabens or long en echelon fissures striking NW–SE at an elevation of about −2450 m. The fissures may have developed during the rise of the Elysium volcano bulge. Burr et al. [2002a] identified relatively fresh lava extrusions from Cerberus Fossae associated with the channel origin.
 Isolated irregularly shaped remnants with an elevation above the plain of up to 1000 m are embayed by plains material and dominate the west and east of the investigated area. While the southeastern plains unit slopes very gently to a topographic low of −2750 m, the southwestern part is blocked by a sudden rise to an elevation of about −1900 m.
 For mapping purposes we combined the available data sets of MDIM-2 Viking imagery, MOLA based digital elevation models and shaded relief maps, as well as MOC wide-angle imagery of the Mars Global Surveyor spacecraft. The resolution of the different data sets was equally set to 231 m/pixel. Longitude and latitude shifts in the Viking and MOC wide-angle imagery were corrected by taking as a basis the digital elevation model using rubber sheeting methods. For small scale features and the verification of area boundaries obtained by geological and geomorphological mapping we complemented the data set by selected MOC narrow-angle shots to achieve higher precision in this highly differentiated terrain. The main mapping procedure has been performed by using Viking and MOC-WA imagery initially. In order to distinguish different units, various albedo features were extracted, while MOC-WA images were more satisfactory than Viking imagery due to its lower noise. Area boundaries have been adjusted finally using MOLA-derived data and the close views of MOC-NA imagery. For detailed studies the MOC-NA images M04/02002 and M12/01869, which cover the plains units surrounding the valley and the proposed valley rim at a resolution of 3–5 meters/pixel, were mapped to enhance the view on possibly even younger ages, and which requires higher image resolution.
 In order to contribute to the determination of the stratigraphic relationships and the origin of the valley we measured the ages in terms of applying the Hartmann/Neukum cratering chronology model [Hartmann and Neukum, 2001] for different geological units in the region of Athabasca Valles. We have analyzed Athabasca Valles comprehensively for understanding the chronostratigraphic relationships and traced different episodes of resurfacing. As it has been discussed by Berman and Hartmann , the Cerberus plains were cited by various authors to be of Middle or Late Amazonian age (around 0.6 Ga) and show signs of even younger resurfacing episodes. Here we used the polynomial expression by Neukum and Ivanov  (and described most recently by Hartmann and Neukum ) to represent the isochrones in form of the cumulative crater size frequency distribution given for a certain age (so-called crater production function). Uncertainties in absolute ages of young surfaces (ages less than 2 Ga) are roughly a factor of two in terms of possible systematic errors in the model.
3. Geological Units of Athabasca Valles
 In the following the geological units of the investigated area as they have been mapped (Figure 2) and dated (Table 1) will be described:
Table 1. Ages Found for the Athabasca Valles Areaa
The ages are listed for each unit as they are indicated in the map shown in Figure 2. The unit “Athabasca D” includes the outflow channel itself.
N(1) is the cumulative number of craters with diameters equal to or larger than 1 km per square kilometer. The N(1) values have not always been derived by measurements at or around 1 km crater diameter but at smaller or larger crater sizes and subsequently have been recalculated for sizes ≥1 km through application of the Neukum crater SFD.
 The knobby remnant terrain unit (C) is represented in the investigated area by numerous small remnants penetrating the Athabasca Valles flow unit (D) and the platy lava unit (E). The unit is characterized by isolated knobs rising a few hundred meters above the surrounding plains. The knobs and craters are veiled by erosional debris aprons. Due to erosional processes, the overall surface texture is very smooth. The boundary of the mapped unit to the adjacent units is very distinct and characterized by a change in relative brightness. The knobby remnant unit has been embedded partly by volcanic plain material. Crater counts yielded an age of about 3.9 Ga (Figure 4, top) which is consistent with age estimates by Tanaka  and Tanaka and Scott .
 The cratered and dissected volcanic highland terrain unit (G) covers the northern part of the investigated area at an elevation of −2600 to −2000 m. It represents what Tanaka  as well as Plescia  described as Elysium volcanic units (ef, respectively Ael).
 The most prominent features in this terrain are the Cerberus Fossae intersecting the region in the northeast. The unit contacts the Cerberus Fossae lava flow unit (B) in the north, the main Athabasca Valles flow unit (D) and the knobby remnant terrain unit (C). In MOC-WA and Viking imagery the unit is characterized by a low relative brightness and encloses several isolated darker units distributed in irregularly shaped patches. MOC-NA imagery reveals a heavily cratered and structured surface, which consists of at least three units on top of each other (M11/00142). The upper, heavily eroded unit is darker and appears more heavily cratered than the underlying units. Crater-like depressions are heavily dissected and degraded and present a coalescing pattern. The crater rims of most of the craters are opened to the northwest and filled with bright material (Figure 3).
 The underlying units are exposed in arcuate, irregularly shaped “windows.” The assignment of this unit to old degraded volcanic material can be judged by the characteristic surface texture and a low albedo. Bright streaks of sediment blown out of several craters in northwestern direction and NE–SW facing dark dunes present evidence for major eolian processes. The elliptically shaped chain-like distribution of small coalescing craters indicate possible secondary craters, which were not used for surface age determination. Age determinations result in an overall age of the heavily cratered highland terrain of 3.4 Ga (Figure 4, bottom).
 The boundary between the “cratered and dissected volcanic highland terrain” and the “valley unit” (D) is well defined by a sudden change in elevation visible in the MOLA data and in relative brightness on Viking and MOC-WA imagery. The stratigraphic relation we extracted through the cratering statistics is well supported by geological evidence at the boundary to the knobby remnant terrain (C) where craters have been filled with material of the younger material of unit G. In the MOC wide angle mosaic we could observe a sudden decrease in relative brightness of the Cerberus Fossae lava flows (B) in the vicinity of the Cerberus Fossae.
 The platy lava unit (E) appears to be typical of the Cerberus plains. It spreads toward the eastern part homogeneously and remains patchy and lobate toward the western part of the investigated area. It consists of extensive lava plains which are broken up into numerous spots of different relative surface brightness. MOC narrow-angle imagery reveal the typical platy lava pattern. The area is well separable from the Athabasca Valles flow unit (D) by an increase in brightness and the absence of distinct lobate flow termini visible in MOC wide-angle imagery. The transition to the undivided plateau unit (F) is not marked by a distinct boundary, as this unit thins away over the lava unit in irregularly shaped wedges. The lava flow unit (E) together with the Athabasca Valles flow unit (D) has been mapped as channeled plains material (cp) by Tanaka  and has been placed into the Upper Amazonian. Plescia  did not differentiate this unit, mapped as Cerberus volcanic plains (Cp), either. According to our crater counts, unit (E) has an age of approximately 3.4 Ga but has undergone major resurfacing processes ending 1.6 Ga ago (Figure 4, bottom).
 The isolated lobate volcanic unit (H) at 158°E, 6°N consists of a small patch of dark material with a smooth surface texture which resembles a lava flow unit. The isolated unit has a lobate shape and fingers onto the volcanic plain material of the platy lava unit (E) in the northern part. The contact to the adjacent unit is very diffuse. Boundaries were mapped according to changes in relative brightness and topographic elevation. This implies that unit H either lies on top of the platy lava unit (E) or it breaks through it. Age determinations for unit H yield an age of about 3.8 Ga. As MOC-NA imagery for a closer look is not available, unit H might as well be linked to lava flows of the Athabasca Valles flows unit (D). We also have to take into consideration, that the age determination of unit H is unreliable due to the small size of the area and nonrepresentative cratering statistics. Neither Tanaka  nor Plescia  has characterized this unit.
 The undivided plateau unit (F) at the southwestern part of the investigated area is characterized by a homogeneous surface texture with sets of subparallel southwest-northeast striking grooves in MOC wide-angle imagery. In MOC narrow angle imagery we observe large fields of numerous east-west and northwest-southeast striking yardangs and associated dune fields on a relatively uncratered unit. Yardangs become more densely spaced and larger toward the southern part of the unit which is most probably connected to a larger thickness of fine-grained material in the south reflected in an increase in elevation. Craters are commonly filled with bright eolian material. The contact to the platy lava unit (E) in the north is rather diffuse (M03/00471). Platy lava is covered by small yardang fields of bright eolian material in the north. To the south, the yardangs change their direction from east-west to northwest-southeast. A distinct boundary can not be determined exactly. Crater counts yield a young surface age of ≈0.6 Ga. This unit has been mapped by Tanaka  and Plescia  as undivided plains material, respectively undifferentiated terrain and dated as Hesperian to Lower Amazonian of age [Tanaka, 1986].
 The Athabasca Valles flows unit (D) extends from the Cerberus Fossae in the north toward the southwest and continuous to the southeast where it overflows the platy lava unit (E) in a patchy way. Toward the southeast the unit presents distinct flow features like ridges and flow lobes at the lava fronts. MOC-NA image M12/01869 (Figure 5) shows the main valley floor located in the northern part of the image. The dark plain (plf) is flooded with volcanic lava with a measured cratering model ages of 3 Ma and 30 Ma (Figure 6, M12/01869 Area 12 and 1, with a crater retention age for N(1km) of 1.33E-06 and 1.35E-05, respectively). The transition to the mesa remnant is characterized by a steep escarpment and the exposure of several remnant layers (white arrows). The flat-topped mesa slopes gently toward brighter plains in the south. In both images of high- and low-resolution cratering model ages of 0.9 Ga were found (Figure 6, low-resolution Athabasca D, high- resolution M12/01869 Area 5) with a subsequent resurfacing 30 Ma ago (Figure 6, M12/01869 Area 5). The slope walls (sw) are cut by a set of very narrow subparallel ridges, which resemble small dunes. At the base of the southern slope we have observed a streamlined island (si), which is clearly defined at the eastern part and broken up at its western margin. There are sets of small valleys branching at the front of the island. The channels spread in southern and western direction onto older lava plains units. A network of polygons might indicate surface desiccation or cooling fractures of basaltic lava. The unit has not been differentiated in detail by Tanaka  and Plescia , who described the flows unit as channeled plains (cp), respectively Cerberus Plains Volcanic plains (Cp). Crater counts yield an age of unit D of about 2.6 Ga and a resurfacing age of 0.9 Ga on the basis of MOC wide-angle imagery (Figure 6, Athabasca D). The counts on MOC narrow-angle imagery support the 0.9 Ga resurfacing event, but also reveal that subsequently the valley area has undergone more resurfacing where the latest event has occurred as recent as about 3 Ma ago (Figure 6).
 The Athabasca Valles buried plains unit (A): The buried plains unit within the Athabasca Valles has been observed by, e.g., Plescia  and described as an old, more densely cratered unit which has been buried by younger smooth material (i.e., the Athabasca Valles flows unit (D)). Several channels are cut into the exposed plains material but they are terminated at the contact with the Cerberus Formation. Plescia  observed possible “pressure ridges or festoons as the lava encroached on the topographically higher area.” The buried plains have an elliptical to teardrop-like shape and are lined up in row parallel to the main Athabasca Valles direction. The plains have been dissected by small N–S trending channels. The plains surface is characterized by a high relative brightness and consists of numerous layers. The plains unit slopes gently under the surrounding material to the south while the northern boundary is well-defined by a steeper escarpment. The surface of the plains material appears to be more degraded toward the north. The contact to adjacent flows units (D) is characterized by ridges parallel to the exposed plains unit as observed and described in detail by Burr et al. [2002a]. The north-south trending channels cut into unit A do not always show a distinct contact with the plains material but are fretted at their rims. Crater counts on MOC-WA and NA imagery yield a plains age of 3.6 Ga but major resurfacing has taken place as recent as 0.03 Ga ago.
 The diagram in Figure 6 of cumulative crater frequencies per km2 versus crater diameter combines a subset of the measurements, which show that the determined ages of the low- and high-resolution imagery fit together and allow an insight into very recent activity. “Kinks” in the size-frequency distribution for small crater diameters, e.g., crater counts of Area 5 of the MOC-NA image M1201869 (Figure 5), indicate two distinct ages due to resurfacing processes. The corresponding ages are given in Table 2.
Table 2. Summary of All Surface Ages Found for Certain Unitsa
Age(s) of the Certain Unit, Ga
The ages correspond to the different isochrones as shown in Figure 6 (solid lines). The measured surface age of 0.9 Ga was found in units of all imageries and establishes a good connection between high- and low-resolution images.
Viking - Area 1
MOC-WA - Area D
MOC - m1201869 Area 1
MOC - m1201869 Area 12
MOC - m1201869 Area 5
MOC - m0402002 Area 1
 The Cerberus Fossae lava flow unit (B) extends north and south of the Cerberus Fossae system and spreads into the main Athabasca Valles flows unit (D) where it thins away. The unit boundaries have been mapped on MOC-WA imagery on the basis of albedo signatures only, as it could not be defined on MOC-NA. Due to its presence northwest of the main Athabasca Valles flows unit (D) at a different topographic height, we had to separate it from unit D, although surface texture is quite similar to unit D. Crater counts yield an age of 3.6 Ga for the Cerberus Fossae lava flow unit.
4. Geological History of Athabasca Valles
 The evaluation of the cratering statistics for the areas mapped in both low-resolution (Viking and MOC-WA) imaging data resulted in similar ages (3.6 and 0.9 Ga, respectively). We have been able to filter out at least three episodes. The average plains age counted on MOC-WA imagery is 3.4 Ga, Late Hesperian (Figure 2) (represented by the platy lava unit (E), which underwent a resurfacing process at about 1.6 Ga ago, Early Amazonian, and by the cratered and dissected volcanic highland terrain unit (G)).
 The oldest areas are remnants (knobby remnant terrain unit (C) and with some restrictions the isolated lobate volcanic unit (H)) breaking through plains units E and G. Their age has been determined at 3.9 Ga, Noachian.
 Areas surrounding the origin of the valley (Cerberus Fossae lava flow unit (B)) and buried plains within the main valley system and along its southern rim (Athabasca Valles buried plains unit (A)) are about 3.6 Ga old with major resurfacing processes as recent as 0.03 Ga. For the valley itself (Athabasca Valles flows unit (D)) we could determine two even younger ages of about 2.6 and 0.9 Ga. We were able to measure one of the youngest ages of 0.9 Ga ago in the low-resolution images and in the most densely cratered units in both MOC-NA images. Another age of 0.6 Ga was found in the images of both low and high resolution, marked as undivided plateau unit (F) in the MOC-WA map. Due to its topographic height this area does not seem to be connected to the valley, despite the fact that areas of the age of 0.6 Ga can be seen in spots in both MOC-NA images, and originated in an independent process. All ages are summarized in Table 1.
 In order to explain the crater count ages (Figure 7) and the complex surface morphologies at the Athabasca Valles region, we propose alternating phases of fluvial/glacial and volcanic activity which ended with the last fluvial event 1.6 Ga ago. The Noachian remnant terrain unit C (and the lobate unit H) have been overflowed partly by the volcanic units E and G from north to northeast direction 3.4 Ga ago during a first major volcanic event. The volcanic units A and B, which might be linked together have been buried by those younger lava flows. The main Athabasca outflow channel has been incised into the pre-existing lava units before 2.6 Ga ago as the topography and morphology of the valley and its morphological inventory suggests. No evidences which would favor a fluvial or glacial process can be found. Later volcanic events have overflowed the valley system in phases of resurfacing events ending at 0.9 Ga and 30 Ma ago. The surface texture south of the buried plains unit A suggests a younger possibly fluvial event overprinting the volcanic texture at a major resurfacing event around 30 Ma ago. The latest volcanic activity we dated back to about 3 Ma ago.
5. Athabasca Valles as MER 2003 Landing Site
 The results of our crater statistics investigation contradict the assumption that the Athabasca Valles were excavated by one or a few possibly still ongoing catastrophic outflow events. More likely is a continual volcanic activity accompanied by fluvial activity. We could record in the MOC-NA imagery the end of several resurfacing events in stratigraphically differing areas. Even more erosional episodes might be found if one studies the “terraces” of the streamlined islands (as shown by Burr et al. [2002b]). The age determination favors the hypothesis that the streamlined features are erosional remnants. The interpretation of recent fluvial activity could not be verified within the geological units we could identify on the imagery described above. However, the presence of water might be visible in the ejecta pattern of a few larger craters superimposing the lava blankets covering the valley floor.
 Independently, the interpretation of our crater statistics data in terms of determining absolute surface ages applying the Martian cratering chronology model presented by Hartmann and Neukum  led to results similar to those Berman and Hartmann  found for comparable regions in the Athabasca system. This supports the reliability of the different styles in the approach to determining surface ages using crater statistics [Hartmann and Neukum, 2001].
 In the case of choosing the Athabasca Valles landing site ellipse as one of the two final destinations of the MER 2003 experiment, it has to be kept in mind that in terms of surface age and its subsequent geologic history the rover would operate in a very complex area. For an understanding of any measurements of the rover it is crucial to recognize the complex geological settings.
 Special thanks to Ursula Wolf for her assistance in the crater counts and their visualization. Our work is supported by a German Research Foundation (DFG) grant.