During this period, Mars is close to its coldest. NH temperature maxima are ∼ 220 K at the northerly latitudes for both scenarios, zonal and meridional winds are slower and uplifted dust quantities are low. Temperatures drop due to Mars' orbit taking the planet away from the Sun, which leads to a reduction in solar radiation and a net cooling for the atmosphere and surface. This leads to a variety of secondary effects occurring on Mars. In reality, CO2 begins to freeze more quickly than it thaws on the opposite pole, leading to a global pressure reduction as CO2 is sequestered from the atmosphere, though this is not currently simulated. Temperatures vary throughout the Martian year, but averages are lowest during aphelion months (Fig. , month 3) and highest during the dust season (Fig. , month 9). Temperature oscillation caused by Mars' eccentric orbit does not seemingly have a direct impact on atmospheric pressure in the UM, as surface pressure remains consistent across aphelion months (sols ∼ 150 to ∼ 300). Instead, the only influence on atmospheric pressure in the UM (as seen in Fig. ) is caused by the increase in dust abundance in the atmosphere during sols ∼ 360 to ∼ 660. This temperature difference leads to weaker winds during month 3 and less dust uplifting, leading to less dust MMR throughout the majority of aphelion compared to months during perihelion . Temperatures are highest in the NH, where it is summer, and then gradually decrease southward. Temperatures also decrease as height increases, as is to be expected. In terms of atmospheric circulation, Mars features strong zonal jets that alternate between hemispheres throughout the year, occurring during winter seasons to the respective hemisphere as the planet transitions between seasons (Fig. ). Meridional winds feature a weak jet near the equator surface, with an opposing jet at the upper boundary layer at ∼0.4σ (Fig. ). The UM RA dust and RI dust simulations are quite similar in month 3, with zonal wind differences of about ∼ 1–2 m s-1 and temperature differences of ∼ 1 K. This is likely due to the low levels of dust abundance in both simulations, and thus its radiative impact during this time period is minimised (Fig. , month 3).

The UM and the PCM both feature strong zonal jets in the upper atmosphere in the SH, but the UM's winds are slower at the upper boundary at the equator and in the NH zonal jet (Fig. ). The meridional jet in the UM is lower and slower than in the PCM simulations (by ∼ 4 m s-1, Fig. ). Air temperature in the UM is generally lower, with especially strong differences between model outputs at the poles, negative at the south pole and positive at the north pole. The lower atmospheric temperature is likely caused by less dust in the atmosphere. Its radiative effects are minimal in the UM, whilst it is relatively abundant in the PCM (average temperature difference omitting the poles of ≈-15 K, Fig. ). The temperature differences at the poles (exceeding -30 and 30 K at either pole, Fig. ), which are much greater than those closer to the equator, are mainly due to the absence of polar ice in the UM, with the latent heat transfer and surface optical properties of ice being simulated in the PCM (Fig. ). Lastly, dust differences between model outputs are at their highest relative to concentration comparisons between models in this month. The UM simulates far less atmospheric dust than the PCM. This is likely due to the absence of forced dust uplifting and a dust devil parameterisation in the UM, which is present in the PCM (described by ).

Here, Mars' hemispheres are transitioning seasons; this can be seen in the jet reversal in the zonal winds and by the location of the temperature maxima at the lower latitudes. Mars features a single Hadley cell which reverses twice a year, with polar cells at each pole. Dust MMR is more than month 3 due to rising temperatures increasing wind speeds, with larger concentrations near the equator (Fig. ). Despite this, dust in the RA scenario is uplifted more compared to RI by ∼ 10 %. The reason for this difference is likely uplifted RA dust scattering solar radiation close to the surface, causing more near-surface warming than there would be with RI dust, where the increased temperature causes faster near-surface winds increasing dust-uplifting rates. Temperature differences between the RA and RI scenarios (Fig. ) are between -5 and 5 K, which is more than in month 3 but lower than in later months. There is a clear difference between polar regions in opposing directions (colder NH pole and warmer SH pole), in addition to a mid-altitude band of warmer air by up to 3 K in the RA dust scenario. These indicate that dust is beginning to impact the atmosphere more actively. Zonal winds are also reflecting the increasing differences (Fig. ), with an NH polar zonal jet difference of up to 40 m s-1. Meridional wind differences are still minimal (Fig. ). Mars' seasonal cycle and the associated reversal of its Hadley cell during this month are clearly shown in the meridional wind patterns, with counter-flowing jets present in months 6 but more stable meridional winds during months 3, 9 and 12.

Differences between the UM RA dust scenario and PCM here are varied in their magnitude but are present for each variable. For air temperatures (Fig. ), the UM is comparable at the surface near the equator, but the temperature in the UM decreases more quickly with height up to the upper atmosphere. The UM, however, features a region with higher temperatures at ∼ 60∘ S latitude, which reaches ∼ 24 K at the surface but gradually decreases with height. There are still differences between the models at the poles, but these differences are less substantial than they were during month 3 (now down to a difference of the UM being -16 K compared to the PCM at the surface). The UM RA features a band of warmer air at ∼ 55∘ S by ∼ 20 K (Fig. ), but temperatures near the SH pole are colder than in the PCM. Zonal winds feature a variety of differences between models, with faster and slower zonally averaged wind speeds distributed across the atmosphere. Both models feature polar jets in both hemispheres, but the UM zonal jets are slower than the PCMs by up to 40 m s-1 at the centre of the SH jet, which gradually becomes more comparable further from the centre of the jet towards the surface. There is an area of faster winds by ∼ 10 m s-1 in the UM at the SH pole, which is at the middle of the atmosphere and lessens towards the surface where differences again become 0. This is likely due to large polar temperature differences, leading to a weaker temperature gradient and causing the SH polar jet in the UM to be more stretched than the PCM. The polar jet in the NH is also slower in the UM by up to 30 m s-1. Dust differences between models are still considerable throughout the entire atmosphere as in month 3, with the PCM having up to 1.2×10-5 kg kg1 (∼ 700 %) more dust than the UM near the surface, while differences lessen with height. Here, the UM dust quantities are much less than the PCM, but uplifting is increasing as Mars approaches the dust season. In all three scenarios dust concentration is increasing, this is promising as it shows the ability to dynamically simulate a substantial change in atmospheric dust abundance during a transition of seasons, as occurs on Mars.

Month 9 is the peak of the dust season, and this is where differences between the RA and RI scenarios are greatest, as higher abundances of dust affect radiation fluxes more severely. Firstly, for both UM scenarios, dust (Fig. , month 9) is mainly concentrated in the SH, forming a large “plume” that extends vertically. Dust abundances have increased from earlier months and are now 2 orders of magnitude more than in month 3. The dust in the RA simulation is more concentrated at ∼ 30∘ S latitude, while RI dust is distributed more widely across latitudes but with a minor difference towards ∼ 30∘ S latitude. This difference can clearly be seen with higher dust abundance at the equator at all altitudes for RA dust compared to less dust nearer both poles than RI dust. During colder months the temperatures are similar between UM scenarios, but during the dust season, the temperature differences are much more pronounced (Fig. , month 9). This change in magnitude of differences shows the effects of RA dust on temperature, highlighting how dust influences radiative transfer in Mars' atmosphere – particularly at mid altitudes, where dust can remain suspended, influencing incoming solar radiation and outgoing thermal radiation from the surface. Temperatures (Fig. , month 9) near the surface are similar between scenarios, both being ∼ 20 K warmer than month 3, but at lower pressures (σ between ∼ 0.2 and 0.1), the RA dust scenario is considerably warmer (exceeding 30 K). The RA dust scenario also features a NH pole which is ∼ 40 K colder than the PCM but warmer by up to ∼ 80 K at the SH pole. This highlights the radiative effects of suspended atmospheric dust and how the climate might be different in its absence. Suspended atmospheric dust causes the upper atmosphere to be warmer than if there were no effects from dust (Fig. ). In the scenario with RI dust, this dust layer does not affect incoming radiation, and solar radiation can reach the surface; however, in the RA dust scenario, the suspended dust layer scatters incoming solar radiation, transferring energy to the suspended dust layer instead of the surface. This “band” of warmer air stretches from the mid altitudes up to the top of the atmosphere, across almost all latitudes (with an exception at the NH pole). Faster wind speeds are a consequence of sharper temperature gradients due to the thermal wind balance relationship and so are affected by the disparity of temperature maxima between hemispheres. This can be seen by the higher wind speeds present in month 9 compared to month 3 (first columns in Figs.  and ). Zonal winds (Fig. , month 9) feature more extreme differences between RA and RI simulations, ranging from -40 to 40 m s-1 in some places. These are mainly at the higher altitudes where temperature differences were at their greatest, above the dust layer. The polar jet in the RA dust scenario is considerably quicker but lacks an opposing jet near the top of the atmosphere in the SH (σ=0.1, 50∘ S latitude), though this jet is small in the RI dust scenario (-20 m s-1). RI dust does not feature an equatorial jet in the uppermost part of the simulated atmosphere (σ≤0.1). Meridional wind differences for this month (Fig. , month 9) are at their highest compared to other months, ranging from -3 to 3 m s-1 in the lower/middle atmosphere (σ=1 to 0.5, 15∘ N latitude). A region near the top of the atmosphere is faster in the RA dust scenario (up to 15 m s-1, σ=0.1, 15∘ N latitude). The dust content in the RA scenario is more centralised around the equator, while dust content in the RI scenario is more spread across the planet, leading to more dust at the poles (Figs.  and ). This highlights the thermal feedback effects of atmospheric dust, as the dust content in the RA scenario causes more localised warming, driving increased vertical uplifting. This thermal influence is not present in the RI scenario, causing vertical uplifting to higher levels to be reduced, leading to a lower, more latitudinally dispersed, atmospheric dust layer. This shows the ability to simulate strong vertical wind-driven dust uplifting in the UM.

Differences between the UM RA dust scenario and PCM during month 9 are at their least compared to other months (when compared at the relative ranges of the different months), suggesting the importance of radiatively active dust for reproducing the salient features of atmospheric dynamics on Mars. Dust MMR in both models is now much more comparable, with the UM having uplifted more dust than the PCM during prior months. Spatially, dust in the UM is concentrated in a large central plume at ∼ 30∘ S latitude, with dust in the PCM being more spread out across the atmosphere. The reasons for this are uncertain but could potentially be the parameterisation of dust uplifting: the UM dynamically calculates dust reservoirs and horizontal flux, whereas PCM uses “forced” dust injection to more closely match observations . This high vertical uplifting in the UM is responsible for a high input of dust into the atmosphere past the near surface, a process described in detail by . The UM features comparable near-surface dust to the PCM across the ∼ 30∘ S latitudinal band (Fig. ), but near-surface dust levels in the PCM are higher above and below this region (above 0∘ N and below ∼ 50∘ S). Particular regions of higher near-surface dust in the UM are the Hellas Basin and the Tharsis region. Temperature differences between models exist throughout the atmosphere, with the largest differences occurring close to the surface at the poles and reaching -40 K in the NH pole and 40 K in the SH pole. These large differences are due to the current absence of ice at the poles in the UM, which will impact temperatures through emissivity and latent heat effects. The rest of the atmosphere is colder in the UM, with differences exceeding 20 K above the SH pole and above the equator at the upper edge of the simulated atmosphere. There is an agreement between models at the surface across the equator as in months 3 and 6, with another small area of agreement between models above the surface at σ=0.9 to 0.2 and 50∘ N latitude. Zonal mean differences here are concentrated in the NH polar jet, with the rest of the atmosphere in agreement. The zonal NH polar jet in the UM is more spread out meridionally than the PCM, so the difference of 40 m s-1 at σ=0.2, ∼ 60∘ N latitude is a result of the jet expanding horizontally in the UM rather than vertically as in the PCM. This is further highlighted by the -20 m s-1 difference at σ=0.3, ∼ 45∘ N latitude, where the wind speeds in the UM are slower as a result of being more dispersed horizontally. Lastly, meridional wind differences between the UM and PCM in month 9 are not concentrated in a single jet or in a single direction but instead occur sporadically throughout the atmosphere. There are some differences of -3 m s-1 near the surface and below the top of the simulated atmosphere, with 3 m s-1 faster wind speeds around the middle atmosphere (σ=0.4, 0 and 20∘ S latitude) and top of the simulated atmosphere (σ≤0.1, 0 to 40∘ N latitude).

When discussing dust, month 9 is the most relevant of the four selected months, as it features the height of the dust season when the dust is at its most abundant in both models. Our results show that we are able to simulate a dust cycle with intra-annual fluctuations in the UM without dust forcing. This month highlights the key takeaway from this study, that we have intra-annual dust quantity oscillation that is entirely reproduced dynamically by the GCM, with dust quantities rising during earlier months, peaking during the dust storm season and subsiding in later months.

In month 12, the dust season ends and the atmosphere of Mars cools. The winds are weaker and dust MMR subsides while the polar jets transition between hemispheres. The uplifted dust quantities are considerably less than those of the peak dust season, with RA scenarios still centred around the equator and RI dust shifting northward. Dust at the equator is more abundant in the RA scenario by ∼4.0×10-7 kg kg-1 (∼ 10 % more dust in the RA scenario).

Overall, there is more dust uplifted in the RI scenario compared to the RA scenario, though this varies spatially (Fig. , month 12). Dust content in the RA scenario at the NH pole is less than that in the RI scenario by up to 1.6×10-6 kg kg-1 (∼ 187 %), with a similar but less extreme difference at the SH pole. Temperatures begin to decrease in this season, and the warmest region is again in the lower latitudes in both scenarios (Fig. , month 12). The RI case is warmer at the equator in the upper atmosphere and colder near the poles in the lower atmosphere than that in the RA dust case. Zonal wind patterns change between hemispheres, with higher polar jet speeds in both hemispheres for RA dust (Fig. , month 12). Differences in months 6 and 12 do switch hemispheres, but the magnitude of these differences is larger for month 12. This is likely caused by more residual dust in the atmosphere from month 9, causing temperature differences to be higher in month 12 than in month 6, in turn affecting the thermal wind relationship. This hemispheric reversal in the zonal wind mirrors what occurs in month 6 but is stronger in month 12 and stronger in the RA dust scenario, with maximum differences reaching now ∼ 20 m s-1 (compared to ∼ 10 m s-1 in month 6). Meridional winds for both scenarios show the Hadley cell direction reversal with the seasonal cycle (detailed by ); wind speeds in both scenarios are once again comparable with the largest differences being between -3 and 1 m s-1.

Comparison between the UM RA dust scenario and PCM shows similar trends to month 6 in each variable but inverted with respect to latitude (i.e. warmer temperature plume occurs in the NH instead of the SH, as in month 6). Temperatures are once again colder in the UM but are generally colder than the differences in month 6 (i.e. month 6 differences ranged from -24 to 24 K, but month 12 ranges from -30 to 18 K). The UM again features a patch of warmer air at ∼ 60∘ N latitude up to ∼0.3σ but in the opposite hemisphere to month 6. There is also a small patch of colder air at the NH pole. These temperature discrepancies are likely to be the result of no polar ice in the UM which is present in the PCM (emissivity and thermal effects of polar ice are shown by ). Throughout the rest of the atmosphere, as in month 6, the UM features lower air temperatures than those in the PCM, becoming lower as the height increases up to the top of the atmosphere (Fig. , month 12). Surface temperatures are also more comparable than month 9, with the only differences in temperatures >20 K being at the NH pole, peaking at >85∘ N by ∼ 40 K colder in the UM. Zonal wind differences between models are quite varied during month 12, with varied wind speed differences across the upper atmosphere in both hemispheres and a weaker and wider polar jet in the NH (Fig. , month 12). The winds near the surface are quite comparable between models, but at the NH pole, the wind speeds are faster in the UM, with this difference increasing with altitude up to ∼0.2σ where the difference is ∼ 30 m s-1. The entire NH polar jet in the UM is larger than the PCM, with the centre of the jet being spread out across more of the upper atmosphere, this is in contrast to the PCM where the NH polar jet is smaller but faster above ∼ 50∘ N latitude. The SH polar jet is also weaker in the UM, but this difference is much less than that of the NH (up to -10 K). Meridional wind differences are also similar to month 6 in locations, but differences, where they do occur, are less than in month 3. At ∼ 15∘ N latitude at ∼0.9σ, there is now no difference between models, as opposed to during month 6, where wind speed predicted by the UM was -2 m s-1 slower.

Dust MMR fluctuates in both models throughout the year, becoming more abundant during month 9 (peak dust storm season, Fig. ) then dissipating during the colder perihelion (Fig. , month 3). Both RA and RI outputs feature strong columns at the mid latitudes spanning up into the mid altitudes during dust season, with dust MMR being on average 2 orders of magnitude higher compared to that during colder months. Although there is an increase in dust MMR between seasons in both model outputs, the intensity of the change varies regionally, with poleward regions increasing less severely than equatorial regions. Despite similarities between the RA and RI scenarios, there are still some differences. Dust MMR is concentrated around the mid-latitudes for RA dust and is more dispersed with RI dust. Dust abundances are higher northward of the equator in the RI dust scenario during peak dust season (Fig. , month 9), but as the dust season subsides (approaching NH spring equinox) there are higher abundances of dust MMR in the RI scenarios towards the poles (Fig. , month 12). Because of this, months 6 and 12 are essential for monitoring dust-uplifting rates.

As shown in Fig. , overall patterns are similar between models, with both RA UM and PCM simulating strong eastward jets that alternate between hemispheres throughout the Martian seasons. Wind speed maxima in the PCM are generally faster than the UM, as is clearly seen in the plot for month 6, where the jets are present in both hemispheres but are ∼ 30 m s-1 slower in the UM. This is likely due to less atmospheric dust around 0.6σ in the UM, which in turn leads to lower temperatures; this reduces pressure gradients, causing slower winds . Despite these discrepancies, our results are encouraging, as they demonstrate the ability to model the major seasonal wind patterns with the UM.

As shown in Fig. , meridional wind patterns are similar between models but do feature some key differences. In month 3, the jet at the upper boundary of the model is situated lower in the UM, relative to the PCM output, in addition to stronger surface winds in the UM. Month 6 features the largest differences, with more distinctive jets in the UM, that are more fragmented and overall weaker in the PCM output. Months 9 and 12 are highly similar between outputs, with the largest difference being a slightly faster mid-latitude jet in the UM during month 12.

As shown in Fig. , the differences in temperature between the UM and PCM outputs are notable throughout the year, the highest occurring in months 6 and 12. There is a consistent difference at the poles: this is likely due to the UM not having any form of polar ice and its effect on albedo and heat transfer . These missing parameterisations will have major impacts around the near surface, as shown in Figs.  and at the poles, with the effect weakening with height. Month 6 and 12 differences both feature patches where the UM simulations are warmer, at ∼ 60∘ in the hemisphere exiting winter (∼ -60∘ during month 3, ∼ 60∘ during month 12). Interestingly, the differences in temperature are small in months 3 and 9 (except at the poles), despite the absence and presence of higher dust abundance, respectively. This occurs despite the difference in dust MMR maximum and minimum with month 3 being half the amount of month 9 (6×10-6 and 1.2×10-5, respectively). This is likely due to the effects of dust uplifting and deposition (highest rate of change during months 6 and 12) having higher horizontal flux rates in the UM, leading to non-uniform differences across the atmosphere. This is also suggested by the fairly consistent distribution of differences in temperatures during months 3 and 9, where the dust is at its lowest and highest dust MMR abundances, respectively, but dust-uplifting and deposition rates are fairly homogeneous.

Across all months, the PCM output is generally warmer than the UM, with the higher temperature differences correlating with months of higher dust differences (Fig. , month 9). This highlights the importance of atmospheric dust in thermal insulation in the Martian atmosphere, as temperatures nearer the upper boundary are cooler in the UM. This is likely caused by the lower amounts of suspended dust compared to the PCM, allowing more solar radiation to the surface, causing a cooling gradient as height increases above the missing dust layer . Surface polar temperatures in the RA dust scenario are consistently different to the PCM throughout the Martian year (Fig. ). This is most likely due to the absence of polar CO2 and H2O ice in the UM, which are included in the PCM. During month 9, there is still a residual ice cap in the PCM , but as there is no such cap in the UM, the radiation is incident directly on the polar soil. This will change the albedo and thermal inertia properties of these regions, leading to increased shortwave absorption in the UM than in the PCM. The subsequent effects of this potentially explain the stronger meridional circulation in the UM. As the hemispheric temperature difference is stronger, transport from the SH to the NH is amplified (as can be seen in Fig. ).

Differences in dust distributions between the UM and PCM results are the most significant of all the variables, both throughout the year and regionally within monthly outputs. Differences between the dust amount vary greatly between models, with dust season total quantities being comparable but spatially varied and the cold season retaining vastly more atmospheric dust in the PCM output than that in the RA UM scenario (Fig. ). Whilst the PCM output during month 3 still has less uplifted dust than its dust season quantities, there is still significantly more dust in this month than the UM by ∼ 1.5 orders of magnitude. This disparity changes as both models approach the dust season (Fig. , month 6), however, with the dust-uplifting rate in the UM being higher than the PCM. This means that there are initially large differences between the two models, but the UM is beginning to rectify the disparity between atmospheric dust amounts as it approaches the dust season. This becomes apparent in month 9 (Fig. , month 9): where the dust season is at its peak, both scenarios have much more similar amounts of dust in their outputs compared to previous months. Both outputs differ in their vertical and spatial (Figs.  and ) distributions, with the UM featuring a large equatorial plume that extends into the upper atmosphere, whilst the PCM features higher abundances at the SH pole. Month 9 features the first instance of a localised higher amount of near-surface dust content in the UM compared to the PCM (mainly in the Hellas Basin). Near-surface dust in the PCM, however, is distributed more evenly across the rest of the planet. This locality of dust in the UM is likely the origin of the vertical plumes present at higher altitudes (Fig. ). Dust is being transported southward from the equator and NH (as shown by the vectors in Fig. ) and is redirected vertically at ∼ 30∘ S. As the simulations progress towards month 12 (Fig. , month 12), dust abundances in the UM reduce more quickly than the PCM, resulting in higher dust abundance in the PCM output. This is a reversal of the observed pattern as seen in month 6, where UM dust uplifting is greater than the PCM, but instead the UM dust deposition is now stronger than the PCM output. This presents an interesting dilemma in understanding how these variables are represented in simulations, as the models do not have the same parameterisations for dust uplifting, and as a result, there is a clear disparity in the amount of dust that can be uplifted between models. The usage of a “free” dust scheme has also been explored by with the Global Environmental Multiscale model (further described by ). They are able to characterise the Martian atmosphere according to Mars year 27 using a “free” dust scheme. Their work alongside this study emphasises the potential of a “free” dust scheme and also acts as a demonstration of model capabilities which we aim to explore further. Although the cause of the disparity in dust distributions between those predicted by the UM and PCM is uncertain, we can speculate on potential causes. These might be caused by dust over-sensitivity to temperature in the UM. As the average atmospheric temperature decreases after the cold season, lower temperatures could potentially indirectly affect dust-uplifting or deposition rates via slower wind speeds more severely in the UM, causing uplifted dust to decrease more quickly than anticipated. There are also consistently colder air temperatures in the UM that vary in intensity throughout the Martian months, which could be amplifying this effect. Another reason might be that both models assume an infinite availability of dust reservoirs, which start uniformly distributed across the surface but are allowed to develop and congregate in areas as the model progresses. This is currently an issue in Mars modelling, as the dust deposition in “free” GCMs does not always match observations . To rectify this, the PCM uses dust-uplifting maps to dictate where dust is being forced into the atmosphere, as described by , and . In our set-up, the UM does not prescribe dust uplifting in such a way but instead relies on the surface scheme to calculate dust reservoirs. Therefore, direct comparison to PCM output cannot be solely attributed to the difference in model parameterisation. Despite this, a comparison of averaged seasonal trends does show that both GCMs are able to capture annual dust storm seasons and non-dust seasons. Where dust is deposited in the model dictates source reservoirs for its subsequent uplifting. Therefore, understanding the deposition of dust and the formation of dust reservoirs (particularly after dust storms due to the amount of dust transported) is paramount to be able to reproduce the Martian dust cycle accurately . If certain regions are key contributors to dust uplifting, then misrepresenting the surface wind conditions over them would have a particularly noticeable effect on the global rate of dust uplifting. The third likely cause of such differences could be dust nucleation scavenging from CO2 and H2O condensation being present in the PCM but absent in the UM. This affects global dust abundances where dust is being extracted from the atmosphere in the PCM during colder conditions. In the present paper, we focus only on dry simulations of the UM, so the dust abundance is not affected by scavenging from condensation, leading to potential overestimation of dust abundance during the colder cloud season. The total dust abundance during month 9 is also impacted by dust uplifting in prior months, so where pressures vary between models (Fig. ), the ability for increased/decreased rates of dust suspension will be likely to vary between models.

Rapid vertical dust-uplifting “rocket dust storms”, as described by , play a key role in dust injection into the atmosphere. The PCM currently factors for this, but the UM dust scheme has not previously been required to simulate such intense vertical uplifting (since it does not feature on the same scale on Earth), and therefore additional changes to the dust scheme are required in the UM. Observations and model comparison investigating this uplifting rate are essential to fine-tune the UM and verify whether the UM captures this correctly, alongside determining which developments or adjustments might be required for the UM dust scheme .

Further differences may be caused by the PCM featuring a varying inter-annual dust content between Martian years , while the UM inter-annual dust content remains largely the same as that displayed in Figs.  and . This was mainly mitigated by using the average scenario for the MCD (described in Sec. ). This remains something to consider, however, were the UM output to be used to investigate the Martian climate across multiple Martian years.

Disparities between dust MMR in the RA and PCM outputs are also affected by the absence/presence of a dust devil parameterisation. Dust devils play a large role in dust vertical transport, particularly during the NH summer , enabling and sustaining dust suspension above the surface (but mainly below ∼ 8 km) during NH spring and summer . While the PCM includes parameterisation for this , the UM does not currently feature any explicit parameterisation for rapid vertical uplifting other than by aeolian-driven processes .

Both schemes omit some dust microphysics due to the difficulty in accurately characterising them within the dust scheme, namely surface crusting and surface re-entrainment , which likely contributes to the disparity between GCM outputs (without forcing) and observations. The differences in spatial distribution could also be caused by the differences in microphysics parameterisation between models, and therefore refining this would undoubtedly improve the representation of dust on Mars. The magnitude of this potential improvement, however, will remain uncertain until an inter-GCM comparison takes place where the initial and boundary conditions are identical. Such work would be able to identify differences which are solely due to these differences in parameterisations. Studies have been conducted on a mesoscale level in conjunction with the Mars 2020 lander but took place prior to recent major improvements in parameterisations in current Mars GCMs . Global comparisons have been extensively applied to Earth GCMs through CMIP6 projects

Full list of projects available at https://www.wcrp-climate.org/modelling-wgcm-mip-catalogue/modelling-wgcm-cmip6-endorsed-mips (last access: 16 January 2023).