By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
 The Mars limited area model (MLAM) has been used to simulate Martian northern polar atmospheric circulation phenomena during the planned season of the Phoenix Lander touching down on the surface (Ls ≈ 76°). Initial and boundary conditions are from Thermal Emission Spectrometer observations assimilated via the United Kingdom Mars General Circulation Model. The higher-resolution north pole-centric nestings (grid lengths about 17–30 km and 8.5–15 km, respectively) resolve phenomena such as shallow mesoscale katabatic drainage flows spiraling out of the cold polar dome, strong valley winds out of Chasma Borealis, and diurnal slope winds embedded on the easterly basic flow, e.g., at Phoenix landing area D. There conditions appear to be mild (for Mars) and baroclinic activity is low. Low-level jets with wind speeds exceeding 20 m/s are, however, possible, but they are most likely nocturnal and would hence not endanger the early afternoon landing of Phoenix. Dust lifting is predicted for strong wind areas near the landing area B (where baroclinic activity appears to be high) and in the general area of the landing area D, primarily east of it.
 Wind conditions in a given landing site area are a combination of flows in a variety of spatial and temporal scales. The spatial range are from global and large down to regional and even smaller. For the landed missions, regional-scale (of the order of tens to several hundreds of km) geography, such as topography and variations in surface optical and thermal characteristics, can drive local circulations and flows [Savijärvi and Siili, 1993; Siili, 1996; Siili et al., 1999] which may surpass the large-scale flows. The typical resolution of GCMs has been inadequate for incorporating many regionally and locally important processes (although some Mars GCMs have experimented with embedded high-resolution subgrids [see, e.g., Forget et al., 1999; Moudden and McConnell, 2005]). Consequently, such smaller-scale circulations cannot be realistically simulated with large-scale models and regional mesoscale models are required. This handicap of a typical GCM, due to overall coarse grid and longitudinal spacing of grid converging to zero near the pole, is particularly acute in the polar regions, where the presence of H2O and CO2 ices and, at least in the northern polar region, pronounced topography presumably drive ice-regolith contrast flows as well as locally strong slope and valley winds. Those circulations in turn may play a significant role in, e.g., dust and water vapor transport.
 The Phoenix Lander is designed to study the Martian northern polar region. The mission is part of the National Aeronautics and Space Administration's (NASA's) Scout program and is in many ways a resurrection of the ill-fated Mars Polar Lander (MPL) mission to the planet's southern polar region. The seminal questions for the Phoenix mission are: (1) can the Martian arctic support life, (2) what is the history of water at the landing site, and (3) how is the Martian climate affected by polar dynamics. These translate into practical science goals and tasks of characterizing the surface, analyzing samples of the soil and ice, and to observing and monitoring the atmospheric conditions and phenomena.
 The atmospheric contribution is made primarily by the Phoenix' Meteorological Station (Phoenix MET) which includes [Michelangeli et al., 2006] (1) a dual-wavelength light detection and ranging (lidar) instrument to observe distributions of ice and dust particles as well as boundary layer properties, (2) three atmospheric temperature sensors of the thin-wire thermocouple type attached to a 1.2 m tall deployable mast, (3) a combination of three capacitive pressure sensors inside the Lander, and (4) a telltale hanging from the mast, observed by the Lander's Surface Stereo Imager Camera and providing some information on wind speed and direction. A thermal and electrical conductivity probe attached to the robotic arm will also provide information on soil temperature and heat transfer characteristics, thus constraining estimates of heat exchange between the surface and the atmosphere.
 A recent description of the landing site selection process is given by Arvidson et al. . Four longitude ranges (labeled A through D) satisfying the latitude range criterion (65°N ≤ ϕ ≤ 72°N) were originally selected for more detailed studies. Region B (longitude range 120°E ≤ λ ≤ 140°E) was the priority region, until the Mars Reconnaissance Orbiter's (MRO) High Resolution Imaging Science Experiment (HiRISE) images showed that region to be unsuitable. The current highest priority landing site is D (230°E ≤ λ ≤ 250°E). The Phoenix local time (PLT) in this region is thus approximately 7–9 h behind the local time at the Martian zero meridian (Mars time, coordinated or MTC). The landing is planned to occur slightly before 0030 MTC or 1630 PLT.
 Mesoscale circulation models have been used before Phoenix for planning the Mars Exploration Rover (MER) missions [Toigo and Richardson, 2003; Kass et al., 2003; Rafkin and Michaels, 2003]. Of greatest interest from that perspective are the estimated local wind fields (including extreme values and vertical variations) during the entry, descent and landing (EDL) phases in the prospective landing regions [Tamppari et al., 2007]. In this work localized analyses of the simulation results are presented close to the midpoint of region D, i.e., (68°N, 240°E). We focus on quantities most relevant to the landing constraints and on the Phoenix MET observables during the first weeks after the targeted landing at Ls ≈ 76°, in the local spring.
 Our motivation for this work is comparison of the results with available data, preparation for eventual analyses of the meteorological data returned by the Phoenix Lander, and testing our mesoscale model for possible uses in design and support of future landed Mars missions, such as the confirmed Mars Science Laboratory (MSL) and ExoMars mission as well as the planned or proposed missions such as the MetNet network [Harri et al., 2006] and sample return missions.
3. MLAM, the Model
 The model used in this work is the Mars limited area model (MLAM). The MLAM has been jointly developed by the University of Helsinki (UH) and the Finnish Meteorological Institute and is based on the hydrostatic dynamical core of the High Resolution Limited Area Model (HIRLAM), version 5.0.0 (P. Undén et al., HIRLAM-5 scientific documentation, 2002, available at http://hirlam.org/open/publications/SciDoc_Dec2002.pdf). HIRLAM is an operational weather prediction model-analysis system, developed and used by several European countries. The present MLAM dynamical core is an Eulerian semi-implicit formulation.
3.1. The Grid
 The horizontal grid is an equispaced latitude-longitude grid of the Arakawa C type. The longitudinal spacing of grid points is hence a function of latitude, normally resulting in very short grid spacings near the poles. These short spacings may cause computational stability problems at high latitudes. An “equatorial” grid, equidistant both in longitudinal and latitudinal direction, would be the best solution for the best accuracy and stability. This issue is addressed in the HIRLAM system by allowing the N/S pole of the computational grid to be shifted or rotated to a different location on the sphere. The computational grid is placed close to the equator (to achieve close-to-equidistant grid spacing) and this chosen grid is then shifted to overlay the area of interest by specifying a suitable new location for the southern pole. This feature of pole-shifted grids is of most use when high latitudes are studied or when (terrestrial) forecasts are computed for such latitudes. It is used in this work for the Martian northern polar region (see section 5).
 Horizontal grids can be nested or embedded in MLAM. Typically two–three nesting levels are provided, thus getting progressively higher-resolution results. The coarsest grid takes its boundary and initial conditions from, e.g., a GCM, and the following higher resolution nested grids in turn take those values from the preceding MLAM levels. The terrestrial HIRLAM takes its initial and boundary conditions also from GCMs and consequently the HIRLAM system is adapted to convert data in standard GCM coordinates to rotated-pole coordinates.
 The vertical grid is Lorenz-staggered at hybrid (p − σ) or pure σ levels. The present model operates on 32 σ-levels. The lowest main points are at about 1.5, 6.5, 30, 75, 125,. meters from the ground, and presently the highest level (model top) is at about 47 km.
 The MLAM turbulence and surface schemes are basically the same as in Earth–HIRLAM. Turbulence is based on the prediction of turbulent kinetic energy (Cuxart, Bougeault, Redelsberger (CBR) scheme [Cuxart et al., 2000]).
 Radiation is taken from the UH 1-D Mars model [Savijärvi et al., 2004]. The longwave scheme uses an emissivity approximation with grey dust. The shortwave scheme (including an improved two-stream code for dust) is described and validated by Savijärvi et al. .
 The soil temperature scheme is presently a force-restore scheme, which takes the values of the soil thermal inertia derived from Mars Global Surveyor's Thermal Emission Spectrometer (MGS/TES) observations.
 The present MLAM version is dry, but future versions may accommodate water vapor as an additional field for prediction.
 The model constants have been changed from Earth values to those relevant for present-day Mars. Changing of the constants for an operational model has been a formidable task, made even more so by the same hardwired and repeatedly defined constants in numerous locations in the code.
3.3. Boundaries and Climate Files
 The initial and horizontal boundary conditions for MLAM are presently derived either the European Mars Climate Database (EMCD) or from the United Kingdom Mars GCM (UKMGCM) output [Montabone et al., 2006]. UKMGCM data taken from runs with either prescribed dust scenarios or with assimilation of MGS/TES data can be used. For the Phoenix simulations we use the UKMGCM data with assimilation of MGS/TES observations.
 The climate files (which include the physiographic data) describe the surface characteristics. These data have been collected from different sources. The albedo and the thermal inertia originate from MGS/TES, the topography from MGS Mars Observer Laser Altimeter (MOLA) data, and the seasonal cap prescription and values of the different surface materials inertia (CO2 ice, water ice) presently from Oregon State University [Tyler and Barnes, 2005; D. Tyler, personal communication, 2006].
4. Model Testing
 During its development and implementation the MLAM model has been compared with and tested against both observational data and other models. A comprehensive description of the model and its testing will be published in a separate article in the near future. In the absence of such work we give here a brief outline of the key comparisons with observations and with other independently validated models. Since the scheme for the surface temperature effects of CO2 ice has been recently implemented into the MLAM, we present the high-latitude cases forming part of this work as well as some low-latitude test cases not directly related to this work.
 The Viking Lander 1 observations and the University of Helsinki 1-D column model results have been compared by Savijärvi . In Figure 1 we present the MLAM temperature profile against UH 1-D, EMCD and observations. The column model fits nicely with the VL-1 afternoon (1600 local time) entry sounding. The MLAM is slightly warmer near the ground but a little colder above ca. 3 km. The EMCD, which has been used as the MLAM boundaries, is the coldest of the three models below about three kilometers. The mixed layer depth predicted by the MLAM and VL1 data has reasonable agreement. The diurnal surface temperature comparisons between the MLAM, EMCD, and UH 1-D model have been presented by Järvenoja et al.  as well as a comparison of the diurnal cycle of the v-component of wind at VL-2. The MLAM agreed fairly well with the EMCD and UH 1-D column model. High-latitude comparisons with the UKMGCM results and TES observations are shown in section 5.
5. Polar Simulations and Results
 As described in section 3.1, MLAM has the ability to use pole-shifted computational grids. After initial tests with non-shifted grids it was soon discovered that for studies in the high northern latitudes of the Phoenix landing areas the pole-shifting was a natural (grid symmetric about the pole) and in fact a necessary approach to avoid stability problems (range of grid spacing in length units much less) and to cover reasonable (contiguous instead of a longitudinal sector) simulation domains over the North Pole. This placement and shape of the grid also replaces the most problematic region in GCM simulations. We selected the computational South Pole to be at (0°N,0°E). With this selection the actual North Pole is located in the middle of our simulation area. All grids are square-shaped and symmetrical about the equator in the computational coordinates (the north pole in real coordinates). This grid shape and placement puts all the suggested Phoenix landing areas (A to D) inside the simulation grids and the polar-circling circulations are also well simulated. The nestings and grid characteristics used are described in Table 1 and Figure 2.
All three grids have a shifted south pole and all the coordinates and directions refer to the pole-shifted computational grids. ID is the identifier and also the number of the first simulation sol of the set; nest is the nesting level; nEW, nNS are the numbers of grid points in east–west and north–south directions; Δα is the angular grid spacing; Δx is the range of zonal grid spacing.
 We have made what is essentially one simulation set comprising three runs using three pole-centered concentric nested grids or domains as described in Table 1 and Figure 2. The nesting is one-way, i.e., the geographically larger domain forces the smaller domain through the initial and boundary conditions, but not in the opposite direction. The durations of the model runs are 19, 18, and 17 sols for the sets 1, 2, and 3, respectively. This gives the model up to two and a half weeks' time to stabilize and create the local weather phenomena. The spin-up time for the MLAM is 3 to 5 sols by our experience. After that the local effects due the better resolution can be clearly seen. In both nested runs (sets 2 and 3) the first day of the preceding model run is cut off to reduce the starting noise, originating from the model adjusting to the new higher-resolution geographics (the starting noise will amplify if the noisy first sol is applied several times). For inter-comparison, the simulation sols common to all three runs are hence labeled from 3 to 19. We started the simulation from Ls = 74.7°, three sols before the estimated landing at Ls ≈ 76°, and mainly concentrated on the results from the last simulation week (sols 12–19).
 The dust visible optical depth τ is taken from the UKMGCM as an area and time average over the simulation domain and is set to a constant τ = 0.36. The roughness length is set to a constant z0 = 0.01 m.
5.1. Surface and Near-Surface Atmospheric Temperatures
 Accurate surface temperatures are essential in a high-resolution model since local temperature differences modify the pressure distributions and drive local near-surface circulations and turbulence. Comparison with the TES observations (Figure 3) shows that the hourly averaged MLAM surface temperatures are rather good (especially in the afternoon) south of about 67°N, where the thermal inertia values are more reliable. North of 67°N the difference is partly explained by the regolith thermal inertia values being nearly constant and less reliable; constant values are also used for the water ice and CO2 ice covers, even though the boundaries are smoothed along 5 grid points. Moreover, the model's climatological distribution of water ice (here assumed north of 75°N) and CO2 ice (north of 85°N) may well differ from the actual ice covers seen by TES. This difference (water ice cap being somewhat too large) also partly explains the large temperature difference between TES and MLAM. Cold temperatures over the CO2 polar cap (north of 85°N) are due to defining those temperatures with the CO2 equilibrium temperature. Cold katabatic drainage flows from the polar dome (to 75°N longitude) may also cool temperatures down the dome. The cold polar temperatures may force baroclinic storms to be stronger than in the real atmosphere.
Figure 4 shows the predicted temperatures and wind speeds at 1.5 m level from the first nesting and from a location representing the Phoenix-D and -B midpoints. In the Phoenix area D for Ls = 76° − 85°, the MLAM-predicted 1.5 m (level 32) air temperatures reach 230–235 K in the afternoons, dropping to about 195 K each night. The ground temperature cycle is very regular and is hence not shown in Figure 4. The results indicate that the temperatures and winds correlate fairly well. This should facilitate the thermal control of the Lander [Tamppari et al., 2007], since heat dissipation increases with wind speed. Within the diurnal cycle, increased dissipation (caused by stronger winds) would typically be helpful during daytime due to a combination of higher ambient temperatures and higher consumption of power; during nights and at colder temperatures weaker winds and weaker heat dissipation are an advantage.
5.2. Surface Pressure
 In the MLAM simulations there is a decreasing trend in the area average surface pressure, which is also seen in the UKMGCM data. Local surface pressures in the middle of the Phoenix B and D landing areas are shown in Figure 5 from the second nesting and from UKMGCM.
 Occasionally there are also moving baroclinic disturbances (low-pressure systems) in these large simulation areas. At Phoenix area D there is, however, relatively high pressure nearly all the time, excepting a passing low at sols 8, 11, and 17 (Figure 5). These transient lows are located mainly east and west/northwest of area D. They can be seen in the animations as moving negative pressure anomalies associated with cyclonic surface wind vortexes.
Figure 6 shows the temporal variance of surface pressure from the second nesting. According to Figure 6, area D is in a low variance region, while area B is close to a high variance region. The variance is caused mainly by transient lows but also by diurnal and semidiurnal patterns (especially at area B, see Figure 5).
 Near-surface (1.5 m) winds are shown in Figure 7 from the second nesting. They are averaged over the whole 17 sol simulation period. The main feature is the katabatic drainage flow out of the high and cold CO2-ice covered north pole area. These shallow (lowest 300 m) flows are turned to the right by the strong Coriolis force. This leads to a belt of outward–spiraling near–surface north–easterlies along the lowlands between 70° and 80°N, strongly resembling the left-spiraling katabatic outflows from Antarctica on the south polar dome of the Earth [King and Turner, 1997].
 Combined with the above are diurnal across-slope and along-valley winds. Figure 8 displays the time-dependent wind anomalies at 0400 (Figure 8, top) and 1600 PLT (Figure 8, bottom) with the respective near-surface temperatures (1.5 m). The wind anomalies are 17-sol means for hours 0000 and 1200 MTC with the mean wind (displayed in Figure 7) subtracted. The multisol anomaly means for 0400 and 1600 PLT filter out the day-to-day variation but display the dominant time specific flows rather clearly. The near–surface temperatures are 17–sol means for 0000 and 1200 MTC.
 In Phoenix area D, for instance, the local north facing slope is relatively warm in the afternoon in Figure 8, and the flow duly displays an up-slope anabatic wind component (Figure 8, bottom), while the early morning slope and the air above it is cold, and the surface flow has a rather weak down-slope trend (Figure 8, top). In the middle point of Phoenix area D the typical wind speed at 1.5 m is about 3 m/s at night and 9 m/s in the afternoon (Figure 4). The wind maxima near the surface can however rise up to 14 m/s.
 Along-valley winds are demonstrated vividly by the strong, cold and steady outflow from the mouth of the Chasma Borealis canyon at 80°N, 300°E (Figure 7).
 Day-to-day variation is, however, relatively strong in the near-surface winds, as there are moving high-pressure ridges and baroclinic cyclones, in which fronts may lead to strong and rapidly varying winds, as at high latitudes on the Earth. These structures have been observed by Viking Lander 2 at 48°N. Figure 4 demonstrates the day-to-day variation in wind speeds at areas B and D, the winds being stronger at B (up to 18 m/s), where also the surface pressure variance was stronger in all our model runs.
 There is a possibility for a strong nocturnal low-level jet (LLJ) as discussed by Savijärvi and Siili . When the conditions are favorable, the wind maximum can be up to 25 m/s, usually at 700–1000 m above the ground, at around local midnight. One such case at the Phoenix area D midpoint is shown in Figure 9. During sol 16 there are relatively weak winds all night at the Phoenix D middle point. However, at sols 17 and 18 just after local midnight there are strong 18–20 m/s LLJs in MLAM (Figure 9a), their maxima located between 600 and 1100 m. The LLJs are much weaker in the UKMGCM (Figure 9b).
Figure 9 also provides some insight into the expected wind speed and wind shear profiles and their diurnal variation. The LLJ wind speeds exceed the EDL phase criteria (<20 m/s below 40 km [Arvidson et al., 2007]), but as the LLJs are expected to occur during the local night and the landing is scheduled for early afternoon PLT, the likelihood of the Lander passing through a nocturnal LLJ seems remote.
 In general, the boundary layer height is a few hundred meters during each night but increases up to 4500 m every afternoon, e.g., in Phoenix area D. Local variations in the thermal inertia and albedo are the main reasons for its modest spatial variations.
5.4. Dust Lifting
 The likelihood of dust lifting has been diagnosed by computing the scalar surface stress from the fields predicted by the simulation:
where τ0 is the surface stress, ρ the air mass density, p the pressure, R the Martian gas constant, T the temperature, and u* the friction velocity. Subscript s refers to the surface and 32 to the lowest model vertical level 32, ∼1.5 m above and nearest to the ground.
 According to Greeley and Iversen , the threshold for dust lifting τ0,lift is in the range of 0.03–0.04 . We have estimated via equation (1) the surface stress fields τ0(ϕ, λ) with hourly intervals for all three simulation sets (selected examples are shown in Figure 10). These τ0(ϕ, λ, t) data cubes have been inspected via animations, time averages (Figure 11) and time series plots from selected coordinates to identify regions and times of interest (primarily occurrences of τ0(ϕ, λ) > τ0,lift). The time series τ0(t) at the Phoenix D landing site, at another active region (close to Phoenix area B) and the grid-averaged values of τ0 are shown in Figure 12.
 The grid spacing has a significant effect. For the coarsest grid (simulation set 1) τ0 approaches, but does not clearly reach or exceed the threshold value. For the higher-resolution grids (sets 2 and 3) the situation changes. Two preferential regions (at approximately the same latitude band as the Phoenix D site) and diurnal times of have been identified. One region is at the sector between the longitudes 120°E and 200°E (“area B”; Figure 10, bottom); the other is between longitudes 240°E and 360°E (bordering the landing area D in the west; Figure 10, top). Surface stress values are also often elevated in the mouth of the Chasma Borealis canyon, the valley winds being the likely cause, but mostly not sufficiently high for these steadier winds to lift dust.
 Dust lifting appears to be most likely in the vicinity of area B and possibly also close to but not at area D. This is partly supported by the pattern of Figure 11, although the highest average values shown do not necessarily indicate occurrences of τ0(ϕ, λ) > τ0,lift; such values can also be explained by persistently, but moderately elevated values of τ0, although τ0(ϕ, λ) < τ0,lift.
 It is of some interest here, from the landing site selection perspective, that the second region, close to the landing area B, was abandoned on the basis of MRO imaginary. In the area B, elevated values occur very regularly in the diurnal cycle (typically from the late afternoon to early morning hours, see Figure 13), whereas in the vicinity of area D even the peak values occur in a more episodic fashion. However, the dust lifting threshold is exceeded only occasionally even in area B (Figure 12). Since the area D time series in Figure 12 is from the nominal Phoenix landing site, the dust lifting threshold is not predicted to be exceeded there, but Figure 10 (top) shows that dust lifting may occur at an area slightly north of the landing site already on simulation sol 7, a few days after the Phoenix sol of landing. The high values of τ0 tend to correlate strongly with local wind speeds (contours in Figure 10).
 The regularity combined with the threshold being exceeded only episodically at or near area B versus the much less regular time variation at area D (Figure 12) point at area B to an underlying (but insufficient by itself) diurnally varying mechanism (slope or ice edge flows), perhaps supplemented by storm events which strengthen the winds and the stress values. This is supported by the pressure variance pattern (shown in Figure 6), indicating large variations close to area B and considerably smaller at area D. Neither factor alone nor their combination appears to be sufficient at area D to lead to exceeding of the lifting threshold.
 As the MLAM cannot as yet lift, transport, and track dust, we are unable to estimate quantitatively, how likely the transport of dust lifted from elsewhere in the polar region to the landing area would be (with subsequent effects on the Lander's power generation and thermal environment, but also on the scientific return of the MET experiment).
 Our results demonstrate that there is a need for mesoscale models for present and future mission planning. With a mesoscale or a limited area model one can study local weather conditions (landing weather/operating weather) with much higher resolution, and presumably with more accuracy, than when using a GCM. With better resolution in topography, albedo and thermal inertia one may find “new” weather phenomena which are too small for GCMs to simulate. These local effects can be decisive in the selection of safe landing sites.
 In particular, the Phoenix landing area D was found to be relatively safe according to the present high-resolution (20 and 10 km grid length) MLAM nested simulations. The near-surface (1.5 m) diurnal temperature range is between 195 and 235 K and the average wind speed about 6 m/s, although nocturnal low-level jets at about 600–1000 m altitude may lead to strong nocturnal wind shear episodes. There are transient baroclinic lows with strong cyclonic winds but they tend to favor longitudes to the east and west of area D in our simulation. However, the too cold polar surface temperatures (compared to TES observations) and the single and relatively short simulation period render the storm statistics only preliminary.
 The time series of near-surface temperature and wind speed at the Phoenix area D indicate frequent coincidence of high-/low-temperature/wind speed and hence more benign conditions for the thermal control of the Phoenix Lander.
 Dust lifting during the initial few weeks after landing at the currently favored landing area D or in its immediate vicinity appears possible, although unlikely for most of that time. Two areas with more likely dust lifting have been identified in the approximate latitude band of the landing site. Although the surface stress values peak daily and regularly on one of the areas, even there the values required to lift dust occur only episodically. This seems to indicate that a combination of local and regional factors are needed for dust lifting, an interesting avenue for future work. Dust transport especially from the region east and north of the landing site is possible with the typical easterly winds near the surface, but the current MLAM lacks the capability to model the dust transport and possible dust storm evolution.
 In memoria, Simo Järvenoja, a long-term key member of the MLAM team, fought a long battle against a serious illness. On 9 October 2007 the illness finally exhausted him and extinguished the candle of his life. This article is dedicated to the memory of Simo. The UKMGCM team from Oxford and Open universities is gratefully acknowledged for the provision of the MGCM data. Dan Tyler from Oregon State University is gratefully acknowledged for providing data for thermal inertia, albedo, water ice, and CO2 ice and for sharing his experiences with MMM5. Mark Paton helped in the TES comparisons. Timo Nousiainen helped with the visualization of the tau pictures.