close to the observed value 29sin(, 1 241°) mas.23. We thank the Mars Pathfinder project team for their

enthusiasm and assistance in acquiring and under-standing the tracking measurements; R. Wimberlyfor recovery of the Viking lander Doppler data; and J.Williams and an anonymous referee for helpful sug-

gestions. The research described in this paper wascarried out by the Jet Propulsion Laboratory, Califor-nia Institute of Technology, under a contract withNASA.

11 September 1997; accepted 5 November 1997

The Mars Pathfinder Atmospheric StructureInvestigation/Meteorology (ASI/MET)

ExperimentJ. T. Schofield, J. R. Barnes, D. Crisp, R. M. Haberle,

S. Larsen, J. A. Magalhaes, J. R. Murphy, A. Seiff, G. Wilson

The Mars Pathfinder atmospheric structure investigation/meteorology (ASI/MET) ex-periment measured the vertical density, pressure, and temperature structure of themartian atmosphere from the surface to 160 km, and monitored surface meteorology andclimate for 83 sols (1 sol 5 1 martian day 5 24.7 hours). The atmospheric structure andthe weather record are similar to those observed by the Viking 1 lander ( VL-1) at the samelatitude, altitude, and season 21 years ago, but there are differences related to diurnaleffects and the surface properties of the landing site. These include a cold nighttimeupper atmosphere; atmospheric temperatures that are 10 to 12 degrees kelvin warmernear the surface; light slope-controlled winds; and dust devils, identified by their pres-sure, wind, and temperature signatures. The results are consistent with the warm,moderately dusty atmosphere seen by VL-1.

The ASI/MET experiment consists of asuite of sensors designed to measure the ver-tical structure of the atmosphere during en-try, descent, and landing (EDL) and to studymartian surface meteorology and climate forthe duration of the Pathfinder mission (1, 2).In situ vertical structure measurements weremade only twice by the Viking entry vehi-cles (3), both during the daytime. In addi-tion to adding a third profile, ASI/METprovides the first nighttime observation, giv-ing information about the diurnal variationof vertical structure, particularly in the upperatmosphere, which is inaccessible to existingremote-sensing techniques. Both Vikinglanders obtained records of atmospheric pres-sure, temperature, and wind velocity at thesurface that extended over several Marsyears. More recent Earth-based, disk-aver-aged microwave observations have been in-terpreted to indicate episodic cooling of the

martian lower atmosphere by about 20 Krelative to the conditions observed duringthe Viking missions (4). By continuing theViking record after 21 years, ASI/MET re-sults are able to determine whether martianmeteorology and climate have changed orremained stable in the late northern sum-mer. Improved measurement sensitivity andtemporal resolution (2) also reveal phenom-ena not seen by Viking and, together withtemperature measurements at three levels,give better information on the exchange ofheat and momentum between the atmo-sphere and the surface.

The ASI/MET experiment combinedaccelerometer and MET instruments (2).The accelerometer instrument containedscience and engineering accelerometersthat each monitored accelerations alongthree orthogonal axes. In each axis, themaximum sensitivity was 20 mm/s2 [2 310–6 Earth gravities (g)], and the accelera-tions expected during EDL were covered bycommandable measurement ranges of 16mg,800mg, and 40g full-scale. The MET instru-ment consisted of pressure, temperature,and wind sensors. Pressure was measuredthrough a 1-m inlet tube that was exposedto the atmosphere during parachute descentas well as after landing (1, 2). The pressuremeasurements have a maximum sensitivityof 0.25 mbar, which is more than a factor of100 better than that available to the Vikinglanders (5). All the MET temperature andwind sensors are mounted on a mast 1.1 m

high, deployed at the end of a lander petalto isolate it from spacecraft thermal con-tamination (1, 2). Atmospheric tempera-ture was measured by four thermocouples:one designed to measure temperature dur-ing parachute descent and three designedfor surface boundary layer measurements 25,50, and 100 cm above the base of the mast.All four thermocouples have time constantsof 1 to 2 s and sensitivities of 0.01 K. Windwas measured by a six-segment hot-wiresensor at the top of the mast, 1.1 m abovethe mast base. The wires are heated by acurrent passed in series through all six seg-ments, and the temperature differences be-tween low and high current modes for eachsegment are used to determine wind speedand direction.

The accelerometer and MET instrumentsrecorded data continuously throughout EDLuntil about 1 min after impact at about03:00 local solar time (LST). Regular surfacepressure, temperature, and wind measure-ments by the MET instrument began about 4hours after impact at 07:00 LST on sol 1 (1sol 5 1 martian day 5 24.7 hours), and theMET mast was deployed at 13:30 LST.

The science accelerometer detected theupper atmosphere 160 km above the land-ing site when the entry vehicle had a ve-locity of 7.4 km/s relative to the atmosphereand a flight path angle 14.8° below thelocal horizontal. 1.5 min later, the entryvehicle experienced a peak deceleration of15.9g at an altitude of 33 km. After 3 min(9 km) the parachute deployed, and at 3.4min (7.4 km) the heat shield separated fromthe lander, allowing the pressure sensor tobegin unobstructed measurements of the at-mosphere. The inflation of shock-absorbingairbags at 5.1 min (0.3 km) terminated theunobstructed pressure measurements, anddescent rocket firing at 5.2 min (0.1 km)ended the direct measurement of aerody-namic decelerations. The first impact of theprobe with the martian surface occurred 5.3min after it entered the atmosphere. In thefirst minute after impact, the landerbounced 15 times and pressure sensor dataindicated that it rolled 10 m verticallydownhill. It came to rest about a minutelater at a site 3389.7 km from the center ofmass of Mars (6). Surface acceleration mea-surements of 3.716 m/s2 agree with values of3.717 m/s2 calculated for the lander loca-tion and height (7), providing a verificationof accelerometer gain calibration.

Because the engineering accelerometerswere used to control parachute deploymentand remained in their least sensitive 40gscale, atmospheric profiles were derivedfrom science accelerometer data only,which were logged at 32 Hz throughoutEDL. MET pressure and temperature datawere collected at 2 Hz during the parachute

J. T. Schofield and D. Crisp, Jet Propulsion Laboratory,California Institute of Technology, 4800 Oak Grove Drive,Pasadena, CA 91109, USA.J. R. Barnes, College of Oceanic and Atmospheric Sci-ences, Oregon State University, Corvallis, OR 97331,USA.R. M. Haberle, Ames Research Center, Moffett Field, CA94035–1000, USA.S. Larsen, Wind Energy and Atmospheric Physics, RisoeNational Laboratory, AMV-125, Post Office Box 49,Roskilde, Denmark DK-4000.J. A. Magalhaes, J. R. Murphy, A. Seiff, San Jose StateUniversity Foundation, San Jose, CA, and Ames Re-search Center, Moffett Field, CA 94035–1000, USA.G. Wilson, Arizona State University, Tempe, AZ, andAmes Research Center, Moffett Field, CA 94035–1000,USA.

SCIENCE z VOL. 278 z 5 DECEMBER 1997 z www.sciencemag.org1752

on February 3, 2021

http://science.sciencem

ag.org/D

ownloaded from

descent and landing phases of EDL. How-ever, spacecraft design constraints did notallow the descent temperature sensor to beexposed to the free flow around the lander,so that direct atmospheric temperaturemeasurements were not possible (2, 8).

For the 30-day primary landed mission,51 equally spaced MET measurement ses-sions were made each day to monitor theatmospheric diurnal cycle and synoptic (day-to-day) variations. These 3-min sessionssampled the atmosphere at 4-s intervals andwere interspersed with 15-min and 1-hoursessions of 1-s sampling to monitor the sur-face boundary layer. The boundary layer ses-sions were repeated as frequently as allowedby data volume, in a pattern that sampled alllocal times of day as rapidly as possible.Finally, on sol 25, a 24-hour session wasexecuted that sampled science data contin-uously at 4-s intervals for a complete dailycycle. The continuous MET observations de-sired during the primary mission were inter-rupted by spacecraft computer resets. One ofthese resets was associated with errors inMET software, and the measurements werelimited to the nighttime from sols 12 to 17,until the errors were corrected.

Entry, descent, and landing. Atmospher-ic density, pressure, and temperature profileswere derived from ASI/MET decelerationmeasurements during Pathfinder EDL andcompared with VL-1 results (Figs. 1 and 2).The mean solar forcing of the martian atmo-sphere at the time of the Pathfinder andVL-1 profiles was similar. Solar longitudes(Ls) of 98° (VL-1) and 142° (Pathfinder)correspond to a seasonal difference frommid- to late summer and to a smaller Mars-sun distance at the time of the Pathfinderprofile. Solar activity, as measured by sun-spot counts, was near-minimum for both pro-files, the lower atmosphere had comparableamounts of dust (9), and the two landingswere at similar latitudes and longitudes onMars. However, the different time of day(03:00 LST for Pathfinder versus 16 :15 LSTfor VL-1) and the 21 years separating thetwo entry profiles suggest that diurnal andsecular effects could be important in under-standing their differences.

For an entry vehicle, atmospheric den-sity is directly related to aerodynamic de-celeration, velocity relative to the atmo-sphere, and aerodynamic characteristics(2). Deceleration was measured directly;velocity and position can be reconstructedby integrating the equations of motionusing the observed decelerations and aninitial velocity and position; and Path-finder entry vehicle aerodynamic charac-teristics are known from computationalaerodynamic simulations and laboratoryexperiments.

Atmospheric densities measured by Path-

finder varied from ;5 3 10–11 kg/m3 at thethreshold of detection to 8 3 1023 kg/m3

immediately before parachute release at 9km (Fig. 1). From 160 to 90 km, densitiesrange from a factor of 5 lower than VL-1values near and above 120 km to a factor of2.5 lower near 90 km. The increase in den-sity between 90 and 80 km, which corre-sponds to a deep temperature minimum (Fig.2), raises the Pathfinder densities at loweraltitudes to values slightly lower than VL-1densities. The lower values of density, andtherefore pressure, encountered by Pathfind-er below 30 km are generally consistent withthe lower overall mass and surface pressure ofthe martian atmosphere at the time of thePathfinder landing (10). The decreased sur-face pressure results from the annual varia-tion in atmospheric mass caused by conden-sation and sublimation from the polar caps(11).

The martian thermosphere, where tem-perature increases rapidly with altitude be-

cause of heating by solar extremeultraviolet radiation, is evidentabove 125 km in the Pathfinderprofile (Fig. 2). Although the uncertaintiesin the derived temperatures at these altitudesare large, it appears that Pathfinder temper-atures are close to or slightly higher thanthose measured by VL-1.

From 65 to 125 km, observed tempera-tures were, on average, 20 K lower thanthose observed by VL-1 (Fig. 2). This con-trast is responsible for the lower Pathfinderdensities above 90 km. Because radiativetime constants reach a minimum in thisaltitude range, a large response to diurnalforcing is a likely explanation for this dif-ference (12). The temperature minimum of92 K at 80 km is the lowest temperatureever measured in the martian atmosphereand may result from the superposition ofwaves, such as thermal tides, on the overallnighttime cooling. Tides propagate fromthe lower atmosphere with amplitudes that

Fig. 1 (left). The atmospheric density profile derived from ASI/MET. The solid lines give the meanatmospheric density profile derived from the accelerometer data and profiles reflecting 62s uncertain-ties in density based on uncertainties in the entry velocity and the finite digital resolution of theinstrument. Errors in aerodynamic characteristics are not included but are not expected to change theerror envelopes substantially. Further work with the accelerometer and pressure data will allow us toextend the density profile down to the surface, where the MET observations of pressure and tempera-ture indicate an atmospheric density of 1.76 3 10–2 kg/m3, marked by an oval on the x axis of the figure.Results from the VL-1 atmospheric structure instrument (ASI) (2, 3) and the Viking 1 upper atmospheremass spectrometer (UAMS) are also plotted for comparison (2, 3). Fig. 2 (right). The atmospherictemperature profile derived from the Pathfinder atmospheric density profile. Temperature profiles cor-responding to the nominal and 62s density profiles of Fig. 1 are represented by solid lines. Thehydrostatic equation is integrated to derive a pressure profile from the density profile, and temperatureis calculated from density and pressure with the use of the ideal gas law. To begin the integration, anupper boundary temperature is determined from the density scale height at the upper boundary.Uncertainties in this temperature affect the derived profiles significantly only above 125 km. In order toderive temperatures, we have constructed a molecular weight versus atmospheric density model basedon the results of the Viking UAMS (29). At present, we have no way of quantifying the accuracy of thismodel, which influences temperature above 120 km. At lower altitudes, the martian atmosphere is wellmixed, with a constant molecular weight of 43.49. Temperature profiles from the VL-1 ASI and UAMSexperiments (2, 3), the CO2 condensation temperature profile, and the surface temperature measuredby the Pathfinder MET instrument (circle) are also shown for comparison.

REPORTS

www.sciencemag.org z SCIENCE z VOL. 278 z 5 DECEMBER 1997 1753

on February 3, 2021

http://science.sciencem

ag.org/D

ownloaded from

increase with altitude.At 80 km, the Pathfinder temperature

profile is lower than the CO2 condensationtemperature (Fig. 2). The lower tempera-ture may result from supercooling or frominaccuracies in the vapor pressure curve forCO2 (13), but it is possible that CO2 couldbe condensing at these levels to form high-

altitude clouds. Observations by the imagerfor the Mars Pathfinder (IMP) of sky bright-ening well before the expected start of dawnmay indicate the presence of such clouds,although the substantial morning cloudsidentified in images of the eastern horizonbefore sunrise are almost certainly waterclouds formed at much lower levels (9).

Below 60 km, temperatures measured byPathfinder are higher than those measuredby VL-1 down to 35 km and are similar orslightly lower at lower altitudes down to16.5 km (Fig. 2). These differences are con-sistent with minor variability in dust con-tent and are also within the amplitudes ex-pected from vertically propagating atmo-spheric waves such as tides (10). In view ofthe similar solar forcing, it is not surprisingthat the Pathfinder and VL-1 profiles arecomparable in the lower atmosphere. How-ever, Earth-based, disk-averaged microwaveobservations over the past 10 to 15 yearshave been used to suggest that the martianlower atmosphere undergoes episodes ofcooling, which are characterized by reducedsolar-absorbing dust content and 20 K lowertemperatures relative to the conditions ob-served during the Viking mission (4). ThePathfinder entry profile shows no evidencefor substantial cooling of the lower atmo-sphere (Fig. 2), although microwave profilesobtained before and after landing suggestedthat a cool atmosphere would be found (14).Furthermore, IMP observations yield atmo-spheric dust opacities of 0.5 (9), which arecomparable to those found by VL-1 at thesame season. These opacities are consistentwith our temperature profile observations.Finally, it is unlikely that Pathfinder entereda locally dusty and anomalously warm re-gion, as dust opacities remained stablethroughout the 30-day primary mission (9).

Below 16.5 km a strong thermal inver-sion is present in the Pathfinder data, wheretemperature decreases from 200 to 181 K at10 km (Fig. 2). This inversion is at too highan altitude to be the strong thermal inver-sion in the lowest few kilometers of thenighttime diurnal thermal boundary layeron Mars predicted by radiative-convectivemodels and one-dimensional dynamicalboundary layer models (15, 16). At the baseof the Pathfinder profile, temperature ap-pears to increase with decreasing altitudeagain and can easily achieve the observedsurface temperature, including predictednear-surface thermal inversions, withoutexceeding the adiabatic lapse rate.

The temperature minimum in the 10-kminversion is well below the condensationtemperature of water vapor in the martianatmosphere, assuming that the 10 precipi-table micrometers of water derived fromIMP measurements are uniformly mixed(9). This inversion may mark the altitude of

clouds seen in IMP images before sunrise(9) and near the low-latitude morning ter-minator in Hubble Space Telescope images(17), although the height of these clouds isnot well known. After sunrise, the cloudsburned off rapidly, and the inversion mayhave disappeared. What triggered the for-mation of a strong temperature inversion isnot known, but possible mechanisms in-clude the horizontal advection of cooler airat 10 km and vertically propagating finite-amplitude gravity waves excited by surfacetopography in the strongly stratified near-surface nighttime boundary layer (18).Once clouds formed, thermal emission fromthe cloud tops could have enhanced theintensity of the inversion at night.

Measurements made after landing. Thepressure, temperature, and wind velocitydata acquired by ASI/MET during the land-ed mission allow the variability of the mar-tian atmosphere at the Ares Vallis landingsite to be studied during the midsummerseason on short, daily, synoptic (day-to-day),and seasonal time scales. This reveals notonly the local properties of the atmosphereand its interaction with the surface but alsomore global information on atmosphericdust loading, circulation, and the seasonalCO2 cycle. Comparisons with VL-1 data tak-en at the same time of year can be used toidentify longer-term climate changes.

Pressure data. During sols 1 through 30,surface pressure at the landing site under-went substantial daily variations of 0.2 to0.3 mbar, which were associated primarilywith the large thermal tides in the thinmartian atmosphere (11) (Fig. 3A). Dailypressure cycles were characterized by astrong semidiurnal oscillation, with twominima and two maxima per sol, togetherwith diurnal and higher-order components,although there was considerable day-to-dayvariability (Fig. 3B). The presence of a largesemidiurnal tidal oscillation is indicative ofatmospheric dustiness over broad regions ofMars and over an altitude range of at least10 to 20 km (19).

A long-term trend in daily mean pres-sure was also seen. A third-order polynomi-al fit to the data shows that mean pressurefell slowly at the beginning of the periodand rose at the end, with a minimum justunder 6.7 mbar near sol 20 (Ls ; 153°) (Fig.3A). This time corresponds to the annualdeep minimum in the seasonal pressure cy-cle associated with CO2 condensation andsublimation in the polar regions of Marsand was seen previously by the Viking land-ers (11).

The ASI/MET pressure sensor detecteda variety of pressure variations on relativelyshort time scales. These ranged from sec-onds to hours and had magnitudes of 1 to 50mbar. The shorter time-scale variations

Fig. 3. (A) Time-averaged surface pressures mea-sured by the MET instrument over the first 30 solsof the Pathfinder landed mission. The averagesare primarily over the 3-min default measurementsessions, of which there are nominally 51 per sol;and the resulting points have been connectedwith straight lines, except for sols 12 through 15,where cubic spline interpolation has been used tofill data gaps of about 8 hours in length. METoperation was restricted to nighttime observa-tions during this period to prevent spacecraft re-sets associated with MET data collection. Themajor gaps in the data set at sols 1, 8, 11, and 17are caused by various spacecraft software resetand downlink problems. After sol 17, the resetproblems associated with MET were corrected,and continuous sampling was resumed. The long-term trend in pressure is represented by a third-order polynomial fit to the data (solid curve). (B)Diurnal pressure cycles for sols 9 (solid line) and19 (dashed line), illustrating the observed day-to-day changes in the diurnal pressure cycle andallowing details of the daily pressure variation tobe seen more clearly.

SCIENCE z VOL. 278 z 5 DECEMBER 1997 z www.sciencemag.org1754

on February 3, 2021

http://science.sciencem

ag.org/D

ownloaded from

(,10 to 15 min) appear to be correlatedwith wind and temperature fluctuations andtend to be largest during late morning andearly afternoon, when the boundary layer ismost turbulent. The most dramatic pressurefeatures were minima of 10 to 50 mbar,usually less than a minute in duration, as-sociated with vortices (dust devils) passingover the lander. A particularly good exam-ple was seen during the continuous sam-pling of sol 25.

Temperature data. In common with thesingle-level observations made by VL-1 atthe same season, the diurnal temperaturevariations at the three Pathfinder levelsrepeat from day to day with a high degree ofconsistency. The diurnal cycle was sampledparticularly well on sol 25 (Fig. 4). For thetop mast thermocouple, a typical maximumtemperature was 263 K at 14 :15 LST and atypical minimum was 197 K at 05 :15 LST,shortly before sunrise. Because of the lowdensity of the martian atmosphere, near-surface atmospheric temperatures are influ-enced by the surface temperature cycle,which is driven by solar heating during theday and infrared cooling at night. For thisreason, diurnal temperature extremes at thebottom of the mast exceeded those at thetop (Fig. 4). Temperatures at the top of themast (1.1 m) were greater by 10 K duringthe day and 12 K at night than those seen at1.6 m by VL-1, probably because of thelower albedo and greater thermal inertia ofthe surface at the Pathfinder landing site.

The diurnal variation of the vertical tem-perature gradient was also consistent fromday to day. At sunrise, the atmosphere istypically stably stratified, and cool dense airlies near the surface (Fig. 5). As the surfacewarms, the air mass is heated from below. Byabout 06:30 LST, all three mast tempera-tures are about equal, indicating that thenear-surface atmosphere is neutrally stable,because isothermal and adiabatic lapse ratesare indistinguishable over a vertical range of75 cm. By 07:30 LST, ground heating ex-ceeds the ability of the atmosphere to trans-fer surface heating by conduction, the tem-perature gradient reverses, the atmospherebecomes unstable, and convection begins.Turbulent mixing carries heat from the sur-face and is revealed by large rapid tempera-ture fluctuations of up to 15 to 20 K observedduring the remainder of the morning andearly afternoon. Later in the afternoon, thesurface cools and turbulent mixing diminish-es (Fig. 4). By 16:45 LST, the thermal pro-file is neutral and what surface winds areobserved convect heat quickly. Shortly after-ward, surface cooling causes the temperaturegradient to invert, and the surface boundarylayer becomes stably stratified for the dura-tion of the night. Major nighttime tempera-ture fluctuations are caused by downslope

winds that disturb the surface boundary layer(Fig. 4).

Large near-surface temperature gradientsof 10 to 15 K are probably a commonfeature of the martian daytime boundarylayer. Because of low atmospheric densities,the convective heat flux is unable to coolthe surface as efficiently as on Earth, whereatmospheric fluxes typically remove 80 to90% of the net surface radiative flux underconvective conditions.

Wind data. The ASI/MET wind sensormeasured wind speed and direction 1.1 mabove the base of the mast. Accurate speeddetermination requires further calibrationof the relation between wind speed, airtemperature, and sensor hot wire overheatunder Mars surface conditions and is not yetavailable. Here we discuss wind directionand approximate speed based on azimuthalvariations of wire overheat, which are afunction of both wind speed and direction.

For sols 1 to 30, wind direction generally

rotated in a clockwise mannerthrough all the points of the com-pass during the course of 1 sol (Fig.6). This rotation was not uniform. Windswere consistently from the south in the latenight and early morning hours and thenrotated steadily through west, north, andeast during the day. Over the 30-day periodstudied, nighttime wind direction was re-markably constant, but considerably morescatter was seen during the day (Fig. 6). Themost anomalous wind variations were ob-served on sols 8 through 10, when north-easterly winds were absent from the dailycycle. A pronounced reduction in the dailyvariation of surface pressure was seen in thesame time interval (Fig. 3).

The recurrent southerly wind from lateevening through morning is consistent witha drainage flow down Ares Vallis, which risesto the south of the lander (1); and thenortherly wind seen in the afternoon is in-dicative of flow up Ares Vallis. Although theclockwise rotation of the wind vector agreeswith that expected from the westward-mi-grating classical diurnal thermal tide, thetime-phasing of wind direction throughoutthe day does not. The classical tidal drivewould generate a westerly maximum near18:00 LST and a southerly maximum atabout 12:00 LST. It therefore appears thatlocal topography, or possibly nonclassicaltides, are controlling the wind direction atthe Pathfinder location during midsummer.

Winds measured at the VL-1 site at thissame season 11 Mars years earlier (1976)were generally weak (,6 m/s), exhibited atime-averaged northwesterly direction, andwere approximately upslope during the after-noon and downslope during the night andearly morning. Increased wind speeds accom-panied the increased pressure oscillationsseen during the several sols surrounding Ls 5150 (20). If the winds are slope-driven at this

6 9 12 15 18 21 24 63

Local solar time (hours)

200

220

240

260

Tem

per

atu

re (

K)

6 9 12 15 18 21 24 3 6

Local solar time (hours)

-4

-2

0

2

4

6

8

Tem

per

atu

re (

K)

Fig. 4 (top). The diurnal variation of atmospherictemperature measured by the top (red), middle(black), and bottom (blue) mast thermocouples,from 06:00 LST on sol 25 to 06:00 LST on sol 26.These thermocouples are respectively 100, 50,and 25 cm above the plane of the lander solarpanels. Temperatures are sampled continuouslyat 4-s intervals throughout this period, but theplots use 30-point (2-min) running means for clar-ity (this smoothing reduces the amplitude and fre-quency of the fluctuations that are present in theraw data). Fig. 5 (bottom). The data of Fig. 4plotted as temperature deviations from the meanof all three thermocouples. Sampling times anddata smoothing are identical to those of Fig. 4.

Fig. 6. Time-averaged wind direction measured bythe MET instrument over the first 30 sols of thePathfinder landed mission, plotted as a function ofLST. Each point represents an average over a 3-min default measurement session. Wind directionis defined as follows: 0° and 360° (northerly), 90°(easterly), 180° (southerly), and 270° (westerly).

REPORTS

www.sciencemag.org z SCIENCE z VOL. 278 z 5 DECEMBER 1997 1755

on February 3, 2021

http://science.sciencem

ag.org/D

ownloaded from

season, the differences between VL-1 andPathfinder winds are expected to reflect dif-ferences in the magnitude and direction ofthe slope at the two sites. Finally, prelimi-nary estimates suggest that wind speeds werecomparable with or lower than those mea-sured by VL-1 at the same time of year.Speeds were generally less than 5 to 10 m/s,except during the passage of dust devils, andwere often less than 1 m/s in the morninghours. This may be consistent with the lowerslope at the Pathfinder site.

Synoptic and seasonal variations. Syn-optic and seasonal variability in the martianatmosphere at the Ares Vallis landing siteare seen most clearly in the pressure data(Fig. 3). In particular, the annual pressurecycle reached a minimum at about sol 20,corresponding to the greatest mass of thesouthern seasonal polar cap (11), althoughthe precise timing of the pressure minimumcan be up to 5 sols earlier and 1 sol later,depending on exactly how the pressure dataare analyzed. The Pathfinder minimum ap-pears to have occurred about 6 sols laterthan that seen during the first year at theVL-1 site (21), which seems to imply moreor longer lasting CO2 condensation on thesouth polar cap. Variations from year toyear of at least several sols in the timing ofthe annual pressure minimum were ob-served by the two Viking landers (22).

Small synoptic variations in amplitude ofless than 3 to 4 K in temperature and 20 to30 mbar in pressure are present in the ASI/MET data, after diurnal fluctuations andlonger term trends are removed. Variationsof comparable magnitude were found in theVL-1 data for this season; the greatest vari-ance was at relatively long (10- to 30-sol)periodicities (21, 23). Missing data and thelimited temporal extent of the available

Pathfinder observations make the estimationof periodicities difficult. On the basis of theVL-1 data and pre-landing general circula-tion model predictions for the Pathfindersite, it is expected that synoptic variations ofmuch larger amplitude will begin as the tran-sition to the fall and winter weather regimetakes place (10, 21). There is evidence thatthis is beginning to happen at the end of thePathfinder pressure record.

Thermal tides. For the first 30 days of thelanded mission, the semidiurnal tide was thelargest component of the daily variation insurface pressure (Fig. 7). Its amplitude isgenerally similar to that found by VL-1 andis consistent with a visible optical depth ofglobal dust of order 0.5 (19). This value iscomparable to that measured by the IMP(9). Given the long vertical wavelengthsassociated with the semidiurnal tide, itsmeasured amplitudes further suggest that thedust was deeply distributed in the atmo-sphere and was not confined to a shallowlayer near the surface. This inference is con-sistent with the dust scale height of about 13km deduced from IMP measurements (9)and with our entry profiles, which show arelatively warm lower atmosphere below 60km. However, as mentioned earlier, a deeplydistributed global dust haze is not consistentwith Earth-based microwave measurementsfor this same period (14).

The amplitude of the diurnal tide wassmaller and more variable. On sols 5 and 9,normalized amplitudes were down to 0.002(Fig. 7), which is more than a factor of 2smaller than anything measured by Viking.The variation seen by Pathfinder is some-what systematic in that diurnal amplitudeswere low during the first 5 to 10 sols, grad-ually rose to values of around 0.010 by sol20, then declined to rather low values again

by sol 27. Phase also varied during thisperiod, retarding during the first 15 solsthen advancing during the last 15 sols. Theamplitude of the diurnal tide is sensitive tothe calibration of the pressure sensor, whichis still preliminary. However, the time vari-ation of amplitude and phase is relativelyinsensitive to calibration.

These variations in the diurnal tide sug-gest that it was modulated by interferenceeffects between the sun-following westwardtide and the topographically induced, reso-nantly enhanced, eastward traveling, diurnalkelvin mode (19, 20, 24). One possibility isthat the interaction is between eastward andwestward modes of comparable amplitude,where the eastward mode period is firstslightly greater than and later slightly lessthan 1 sol. Alternatively, the eastward andwestward modes could have similar periods,and amplitudes may vary with time. Themodeling of this interaction may help todistinguish between these two possibilities.

Nighttime temperature inflections. ASI/MET temperature measurements show that,on average, atmospheric temperatures fallmonotonically through the evening andnight until sunrise. On most nights, howev-er, the steady decline in temperature wasinterrupted by one or more inflections in thegradient, where the temperature fell less rap-idly or even rose before resuming its down-ward trend. Striking features of this kind areseen in Fig. 4 at 18 :15, 19:00, 20:40, and03:05 LST and are particularly marked intop mast thermocouple temperatures.

Temperature inflections seen in theVL-1 temperature data set were attributed toradiative effects associated with the forma-tion of ground fogs (25). At the Pathfinderlanding site, temperatures are about 10 Kwarmer than those of VL-1 at night, andcondensation is unlikely, given the 10 pre-cipitable micrometers of water measured bythe IMP (9). However, there was a positivecorrelation between increased wind speedand the observed thermocouple temperatureinflections. Increased horizontal flow mayhave disturbed the strong temperature in-version that developed at night, bringingwarmer air from aloft down to thermocoupleheights through enhanced vertical mixing.After the nighttime warming episodes, tem-peratures fell rapidly in conjunction withsubstantial reductions in wind speed. Thenighttime temperature inflections occurredin the presence of southerly winds and can-not be produced by thermal contaminationfrom the body of the spacecraft, which islocated to the northwest of the MET mast.

Dust devils. Short-term variations inmeasured surface pressure, wind velocity,and air temperature over periods of tens ofseconds to minutes suggest that small-scaleconvective vortices passed through the

Fig. 7. (A) The amplitudeof the diurnal (black dia-monds) and semidiurnal(open diamonds) surfacepressure tides for thefirst 30 sols of the Path-finder landed mission.Amplitudes are normal-ized relative to the meanpressure for each day.(B) The phase, in LST, ofthe diurnal (black dia-monds) and semidiurnal(open diamonds) surfacepressure tides. Enoughpressure data were col-lected to characterizethe thermal tide on 20sols. Pressure measure-ments from each ofthese sols were fittedwith a cubic spline, fromwhich 51 equally spaced intervals were sampled to define the tide.

SCIENCE z VOL. 278 z 5 DECEMBER 1997 z www.sciencemag.org1756

on February 3, 2021

http://science.sciencem

ag.org/D

ownloaded from

Pathfinder lander site. During the passageof one of these features, pressure fell andrecovered rapidly, accompanied by abruptshifts in wind direction. These events,which we refer to as dust devils, are proba-bly similar in character to features noted inthe VL-1 and VL-2 meteorology data (26)and may have been seen by the Vikingorbiter cameras (27). The narrow, well-de-fined pressure minimum is the most char-acteristic feature of these vortices, but it wasnot seen in the Viking lander pressurerecord because of poor resolution and coarsetemporal sampling. There are currently noPathfinder lander images to indicate wheth-er the apparent vortices entrain dust, butthe passage of a particularly large feature onsol 62 was correlated with a short-livedreduction of about 1.5% in the power gen-erated by the Pathfinder solar panels.

A dust devil event at 14 :15 LST on sol 25has the characteristics of a clockwise-rotatingvortex traveling toward the south-southeast(160°), with the vortex center passing westand south of the lander (Fig. 8). During thisevent, pressure fell 0.028 mbar in 24 s to aminimum, followed by a more rapid rise of0.027 mbars in 8 s. The pressure minimumwas accompanied by a local temperature max-imum at all three thermocouple heights.These temperature maxima are themselvesembedded in a pronounced temperature min-imum, which extended 30 to 40 s before andafter the pressure minimum. Wind directionrotated from northwesterly, which was themean flow direction before the vortex arrived,through north to northeasterly in the 40 spreceding the pressure minimum, before shift-ing abruptly through northerly to westerly atand just after the pressure minimum. Winddirection slowly returned to north-northwest-erly over the minutes following the pressureminimum. This pattern is consistent with aclockwise-rotating vortex embedded in anambient flow that has a speed less than thevortex rotational speed (26).

Boundary layer results. The near-sur-face boundary layer of the martian atmo-

sphere can be described by “universal” scal-ing laws involving dimensionless coeffi-cients determined from measurements inEarth’s atmosphere. The Viking and Path-finder measurements provide us with anopportunity to test this universality by usingmeasurements in another atmosphere. Al-though the martian atmosphere differs inmany respects from our own, most of theuniversal relations are expected to be valid,based on the analysis of Viking data (28).

Relative to Viking, the Pathfinder mea-surements have the following advantages forboundary layer studies. Temperature mea-

surements at several heights allowfor a more direct determination ofthe vertical temperature gradient,and the successful MET sampling strategyhas ensured that turbulence and profile datawith high temporal resolution (1 Hz or bet-ter) are available over a wide variety of con-ditions, with good systematic coverage ofdiurnal and interdiurnal variability.

As an example of the analysis of turbu-lence characteristics from Pathfinder, weshow an average temperature power spectrumobtained from the Fourier decomposition ofseveral temperature time series (Fig. 9). Indi-vidual spectra are scaled by the appropriatetemperature scale T

*, which, following the

Monin-Obuchov formulation, is the most im-portant scaling parameter for both verticalprofiles and spectra (28). The T

*values scal-

ing the spectra were derived from the mea-sured temperature gradients, assuming close toneutral thermal stability due to the proximityof the ground (0.25 m), although T

*ranges

from –5 K to 10.2 K. Although simplified,the scaling does a reasonable job in collapsingindividual spectra and hence in supportingthe applicability of the Monin-Obuchov scal-ing laws.

REFERENCES AND NOTES___________________________

1. M. P. Golombek et al., Science 278, 1743 (1997 ).2. A. Seiff et al., J. Geophys. Res 102, 4045 (1997 ).3. A. Seiff and D. B. Kirk, ibid. 82, 4364 (1977 ).4. R. T. Clancy et al., Icarus 122, 36 (1996).5. S. L. Hess, R. M. Henry, C. B. Leovy, J. A. Ryan, J. E.

Tillman, J. Geophys. Res. 82, 4559 (1977 ).6. W. M. Folkner, C. F. Yoder, D. N. Yuan, E. M.

Standish, R. A. Preston, Science 278, 1749 (1997 ).7. W. M. Folkner, personal communication.8. T. Rivell et al., J. Spacecr. Rockets 34, 265 (1997 ).9. P. H. Smith et al., Science 278, 1758 (1997 ).

10. R. M. Haberle, J. R. Barnes, J. R. Murphy, M. M. Joshi,J. Schaeffer, J. Geophys. Res. 102, 13301 (1997).

11. R. W. Zurek et al., in Mars, H. H. Kieffer, B. M.Jakosky, C. W. Snyder, M. S. Matthews, Eds. (Univ.of Arizona Press, Tucson, 1992), pp. 835–933.

12. M. Lopez-Puertas and M. A. Lopez-Valverde, Icarus114, 113 (1995).

13. P. B. James, H. H. Kieffer, D. A. Paige, in Mars, H. H.Kieffer, B. M. Jakosky, C. W. Snyder, M. S. Mat-thews, Eds. (Univ. of Arizona Press, Tucson, 1992),pp. 934–968.

14. R. T. Clancy, unpublished data.15. P. J. Gierasch and R. M. Goody, Planet. Space Sci.

16, 615 (1968).16. R. M. Haberle, H. C. Houben, R. Hertenstein, T.

Herdtle, J. Atmos. Sci. 50, 1544 (1993).17. M. J. Wolff, R. T. Clancy, P. B. James, S. W. Lee,

J. F. Bell III, Bull. Am. Astron. Soc., in press.18. J. A. Magalhaes and R. E. Young, Icarus 113, 277

(1995).19. R. W. Zurek and C. B. Leovy, Science 213, 437

(1981).20. J. E. Tillman, J. Geophys. Res. 93, 9433 (1988).21. J. R. Murphy, C. B. Leovy, J. E. Tillman, ibid. 95,

14555 (1990).22. J. E. Tillman, N. C. Johnson, P. Guttorp, D. B. Per-

cival, ibid. 98, 10963 (1993).23. R. D. Sharman and J. A. Ryan, J. Atmos. Sci. 37,

1994 (1980).24. R. J. Wilson and K. Hamilton, ibid. 53, 1290 (1996).25. B. M. Jakosky, A. P. Zent, R. W. Zurek, Icarus, in

press.26. J. A. Ryan and R. D. Lucich, J. Geophys. Res. 88,

11005 (1983).

Fig. 8. Pressure, wind,and temperature chang-es associated with asmall-scale vortex, ordust devil, passingthrough the Pathfinderlanding site. The mea-surements were taken at4-s intervals.

Fig. 9. An average power spectrum S(f ) for themartian atmosphere (dashed line) is compared toa similar spectrum from Earth obtained underweakly unstable conditions (bold line). The aver-age spectrum is derived from night- and daytimepower spectra for 10-min martian atmospheretemperature time series measured by the bottommast thermocouple. The spectra are normalizedby the turbulence temperature scale T

*derived

from the temperature profiles, assuming a loga-rithmic profile of T(z) 5 T

*/k Ln(z/z0), where k is

Von Karman’s constant and z is altitude, and plot-ted versus the frequency f . The difference be-tween the Earth and Mars spectra at high frequen-cies is mainly due to sensor time constants. Theline marked 2/3 represents the ideal slope of theinertial subrange power law.

REPORTS

www.sciencemag.org z SCIENCE z VOL. 278 z 5 DECEMBER 1997 1757

on February 3, 2021

http://science.sciencem

ag.org/D

ownloaded from

The Mars Pathfinder Atmospheric Structure Investigation/Meteorology (ASI/MET) ExperimentJ. T. Schofield, J. R. Barnes, D. Crisp, R. M. Haberle, S. Larsen, J. A. Magalhães, J. R. Murphy, A. Seiff and G. Wilson

DOI: 10.1126/science.278.5344.1752 (5344), 1752-1758.278Science 

ARTICLE TOOLS http://science.sciencemag.org/content/278/5344/1752

REFERENCES

http://science.sciencemag.org/content/278/5344/1752#BIBLThis article cites 22 articles, 5 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

No claim to original U.S. Government Works.Copyright © 1997 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science.

on February 3, 2021

http://science.sciencem

ag.org/D

ownloaded from