2.1. Temperature Measurements
 Air temperatures and temperature differences between levels will be monitored, almost continuously, by three temperature sensors based on fine wire, butt-welded thermocouples (75 μm diameter, Constantan-Chromel) mounted in C frames on a 1 m mast, coupled with a reference platinum resistance thermometer (PRT) in an isothermal block containing the “cold” junctions of the thermocouples. Levels on the mast are 0.25, 0.5, and 1.0 m above the lander deck, which itself is ∼1 m above the ground. There will be three thermocouple junctions in parallel in each of the air temperature sensors, providing a degree of redundancy. Sample thermocouples are shown in Figure 1.
 Fine wire thermocouples were selected in part because of a desire for fast response and a modest effect from solar radiation, since no radiation shielding is being used, and in part because of space heritage based on their successful utilization on Viking [Chamberlain et al., 1976; Larsen et al., 2002]. Initially, we had hoped to include three-dimensional turbulent velocity measurements in order to measure fluxes via eddy correlation. As noted in section 1 this has not been possible, but fast response will still be valuable in order to determine turbulent temperature fluctuations, σT, and to detect the rapid temperature fluctuations that may be associated with features such as dust devils or sharp frontal passages. A more robust thermistor with a slower time constant may have been an alternative; but heritage is an important factor, and on Viking this type of thermocouple performed very well.
 The microvoltages generated by the thermocouple units are measured at 2-s intervals and converted to digital signals with a 16-bit analog to digital converter. Drifts in the readout electronics are calibrated and corrected for at the same time. A total of 256 data records (8.53 min) are buffered and can either be stored at full resolution in the flash memory of MET or processed to provide mean, standard deviation, maximum, and minimum values within the 8.53 min block for storage on the MET unit. The unit can switch autonomously to the full resolution mode, for both temperature and pressure data, on the basis of the magnitude of temperature or pressure fluctuations within the 8.53 min block. The MET unit has its own memory, but data must be transferred to the main lander memory prior to transfer back to Earth via one of the Mars orbiters.
2.2. Pressure Measurements
 A Finnish Meteorological Institute (FMI) sensor, based on a silicon diaphragm sensor head manufactured by Vaisala Inc., combined with MacDonald Dettwiler Associates (MDA) data processing electronics will measure pressure. The FMI unit has three pressure sensor heads. One of these is the primary sensor head, and the other two will be used for monitoring the condition of the primary sensor head during the mission. During the mission, the primary sensor will be read with a sampling interval of 2 s and the other two will be read less frequently as a check of instrument health.
 The pressure sensor system has a sophisticated real-time data processing and calibration algorithm that allows the removal of temperature-dependent calibration effects. In the same manner as the temperature sensor, a total of 256 data records (8.53 min) are buffered and can either be stored at full resolution or processed to provide mean, standard deviation, maximum, and minimum values for storage on the MET unit.
 The time constant of the pressure sensor had not been specified in the formal requirements, but it was originally hoped that it would be short compared to the sampling interval of 2 s. In fact, it is a little longer than that, ∼3 s, because of locational constraints and dust-filtering requirements. However, the temporal resolution should still be good enough to detect pressure drops associated with the passage of a nearby dust devil.
2.3. Telltale Wind Measurements
 Measuring wind on the surface of Mars is important to understanding passing weather systems and water transport issues. It will also be important in assessing the potential for local dust transport and erosion. Wind information will also aid in choosing the best time of day for sample delivery from the scoop of the robotic arm into Microscopy, Electrochemistry and Conductivity Analyzer (MECA) and TEGA; loss of fine dust blown away by the wind from the sample during this process could affect the validity of the analysis.
 As noted in section 1 we had hoped to include an anemometer in the MET package. Faced with a lack of resources to achieve this and with a real desire to have some wind information we decided to make use of the SSI camera and have designed a novel Telltale to achieve this. The Telltale assembly (Figure 2) is made from several parts. The active (deflecting) part of the Telltale consists of a hollow tube of 8 μm Kapton foil deposited with 2 nm Au before assembly. The Kevlar fibers are bonded to the mounting screw and to the Kapton tube. Attached to the mounting screw is an orientation marker, which is a cross-shaped, sandblasted aluminum piece with holes that will serve as a background for mirror images to help determine Telltale deflection and orientation. The mounting screw is attached to the gallows, which in turn is attached to the top of the MET mast. The gallows frame also holds a mirror directly below the Telltale and is inclined in such a way as to allow the SSI camera in its deployed position to see a vertical image of the reflected Telltale.
 Small wind velocities will deflect the Telltale in the wind direction with a deflection proportional to the magnitude of the horizontal velocity. At higher velocities the Telltale tends to oscillate because of oscillations in the flow field. The resonance frequency of the Telltale is at about 3 Hz on Mars, and turbulence in this frequency range will be observed as smearing of images. At winds >10 m s−1 on Mars the Telltale will start to reach its maximum deflection, laying horizontally, losing its wind speed/deflection correlation ability. Wind directions should, however, still be determined.
2.4. Atmospheric Structure Experiment
 In addition to the MET experiment on Phoenix, the atmospheric structure on the day of landing will be derived from measurements of the deceleration of the spacecraft during atmospheric entry. Deceleration is caused by frictional drag that depends on the atmospheric density, pressure, and temperature along the flight path. Previous data sets collected during atmospheric entry were from Viking 1, Viking 2, Mars Pathfinder, and the Mars Exploration Rovers. Phoenix will provide the first in situ atmospheric structure in the high-latitude atmosphere. Vertical atmospheric temperature will provide us with a basis for estimating the saturated vapor holding capacity of the atmospheric column. By examining the temperature profile and comparing with general circulation model (GCM) predictions, it will also help verify radiative calculations and understand net poleward heat transport. The degree of stability of the atmospheric profile bears upon whether convection is likely to be important in transporting dust, water vapor, or other tracers aloft. Mars GCMs predict quiescent large-scale winds in the northern polar region during the northern summer [Haberle et al., 1993]. However, the strong meridional temperature gradient between bare ground and the polar cap generates cyclonic systems [Hunt and James, 1979]. The extent to which such weather systems affect atmospheric profiles is unknown. Consequently, a polar atmospheric profile will allow us to better understand the heat budget, dust and water cycles, and dynamical meteorology of the polar environment.
 The effects of aerodynamic drag during Phoenix entry are monitored by two inertial measurement units (IMUs) that are located beneath the lander deck on opposite sides of the spacecraft. Each IMU uses three accelerometers for linear acceleration measurement in three Cartesian axes and three ring-laser gyroscopes to measure the three-dimensional angular orientation of the entry vehicle. The IMUs are quasi-commercial products manufactured by Honeywell (Clearwater, Florida,) with part number YG9666BC. The IMUs were selected and configured as part of the active engineering sensing system needed for the critical entry, descent, and landing (EDL) phase of the mission. Consequently, the IMU operational characteristics were wholly driven by EDL engineering demands; use of the IMU data for atmospheric structure reconstruction was a secondary consideration.
 Each IMU is internally configured so that it outputs three-axis accumulated linear velocity changes (the time integral of acceleration or “delta-v”) and three-axis accumulated angle changes (the time integral of angular rate). Unlike previous Mars lander missions, data will be saved onboard at the maximum device output rate of 200 Hz. The IMUs were delivered with a factory calibration based upon individual unit testing after assembly. Outputs of delta-v from accelerometers and angle change from gyros are internally compensated for with biases, scale factors, and alignments on the basis of coefficients determined by manufacturer calibration tests. The accelerometers have two stages of digitization that result in pulses and then counts. The final raw digital output is in delta-v units, where one output count is equivalent to 0.0753 mm s−1. However, the final output moves up and down in pulses, which are leaps of many counts, with each pulse equivalent to 2.7 mm s−1. Thus, the amplitude of one pulse is equivalent to 2.7/0.0753 = 35.86 counts, on average. Added to this is noise typically on the order of one count, so that jumps in the raw output are typically 36 or 37 counts. This determines the digital resolution of the raw 200 Hz velocity data to be a delta-v of 2.7 mm s−1 and typical noise level to 0.0753 mm s−1. Individual gyros measure the angular rotation in one axis up to a full-scale rate of 375° s−1 with a digitization of 0.01144° s−1. Three-sigma noise on the raw gyro output is estimated at 45 microradians.
 The noise in the raw digital data can be lowered by integration over selected time intervals, but this is a trade because time integration reduces the vertical resolution in the atmospheric reconstruction. Another consideration is that the IMUs are located neither at the entry vehicle center of mass nor on the symmetry axis. Consequently, consideration of a center of mass frame is needed to take account of spurious components of acceleration measured away from the center of mass. The temperature structure of the atmosphere along the flight path can be calculated from Phoenix IMU data using a four-step process [see, e.g., Magalhaes et al., 1999; Withers et al., 2003] as follows:
 1. The time history of the spacecraft velocity relative to the atmosphere is calculated as a function of altitude by integrating forward the measured velocity changes using the equations of motion and the initial velocity and path angle at entry as a boundary condition. Also, gyroscope data can be integrated to derive the orientation of the spacecraft along the flight path, provided an orientation boundary condition is available at atmospheric entry.
 2. The local atmospheric density is calculated from acceleration along the flight path from the drag equation applied to the spacecraft. This step must be iterated because drag coefficients depend on atmospheric density.
 3. The vertical pressure profile is calculated from integrating the hydrostatic equation for the atmosphere.
 4. Given the density and pressure profiles, the final step is to calculate atmospheric temperature from the ideal gas law.