De-mystifying Martian Clouds

 Two of the lab's PhD students have just published analysis on how Martian clouds interact with sunlight in companion papers over in the Planetary Science Journal! The results represent work on thousands of images of the sky taken by the Curiosity Rover over more than ten (Earth) years and describe the thickness of the clouds we saw and give information on the crystals that make up those clouds. How can you use pictures of clouds to figure out what they are like on the scale of less than a thousandth of a cm? Read on to learn more!

by Alex Innanen & Conor Hayes

Here at PVL, we’re a big fan of Martian clouds. For over eleven (Earth) years, we’ve tasked our favourite PVL-er, the Mars Science Laboratory (MSL) Curiosity rover, with staring up at the sky for a handful of minutes every few sols to capture the clouds drifting over its home in Gale Crater. Recently, we had two papers accepted that discuss some of these cloud observations. Our papers are very similar – both look at how light interacts with water-ice clouds during the same time of year (the Aphelion Cloud Belt, or ACB, season) using the same cameras (Curiosity’s Navigation Cameras, or Navcams). Between our papers we have 33 figures and over 15 000 words. Reading through that much material can be a daunting proposition! So, presented below for the cloud-curious is a brief and hopefully engaging summary.

Over the years, clouds have made frequent appearances on this blog, but as a quick refresher, yes, there are clouds on Mars! Some are made of dust, some are made of carbon dioxide, and some are made of water-ice. The water-ice ones are the ones we’re interested in, particularly those that form as part of the ACB. Every year, when Mars approaches its furthest point from the sun, it sees an increase in water-ice cloud formation around the equator. Gale Crater, where Curiosity lives, is just five degrees south of the equator, so it sees the southern edge of this belt of clouds. This is really great for those of us on the environmental science team – we get an opportunity every year to study these clouds and look for patterns in their behaviour from year-to-year. 

ACB clouds above Gale crater tend to be fairly tenuous – think of those wispy cirrus clouds you might see on Earth (shown above). Like cirrus clouds, they’re made of tiny crystals of water ice. (On Earth we don’t tend to specify what kind of ice, but on Mars the atmosphere can be cold enough for both water and carbon dioxide to freeze, so it’s helpful to differentiate between the two.) These crystals form when water vapour in the atmosphere condenses on some kind of nucleus – usually dust particles.

On Earth, atmospheric water vapour tends to freeze into a certain set of shapes depending on the specific conditions present when the ice crystals are forming.  These shapes have been catalogued thanks to the fact that we can actually fly into and directly sample our clouds taking close-up pictures of the ice crystals. While it may be possible that the ice crystals in Martian clouds have similar shapes to those in terrestrial clouds, we unfortunately cannot (yet) directly image them like we can here on Earth. Instead, we have to rely on some physics tricks. One of these tricks is looking at how light interacts with the clouds. These light interactions can produce a myriad of cool optical effects, but more importantly can give us information about the size and shape of these particles. This information is captured by what’s called a ‘phase function’. The phase function is a mathematical description of how much light is scattered by some particle at different angles from 0 to 180°, visually represented by a curve. The exact shape of that curve depends on the shape and size of the ice crystals in the clouds, so we can attempt to determine the nature of martian ice crystals by comparing our phase function measurements to those that have been made for ice crystals on Earth.

To see how light is scattered by the clouds across different angles, former PVL member Brittney Cooper came up with the phase function sky survey: Curiosity takes a series of 9 small cloud movies looking in different directions all around the rover, like you can see in the gif below.

From this we get information about how sunlight is scattered by the clouds at different locations around the rover and put together an average phase function for the clouds we observe throughout the cloudy season at Gale Crater. We’ve been doing this observation through four Mars years so far, which means that we can compare different Mars years to see if there’s any change in the phase function. The ACB is a very stable feature – it doesn’t change much from year to year. Likewise, the average phase function at Gale doesn’t change much either. Nor does it show much difference between morning and afternoon observations. So, what does that tell us? Mostly it’s just that – the phase function doesn’t change much from year to year, or from morning to afternoon. But this could also suggest that the water-ice crystals that make up the clouds aren’t changing.

Which brings us back to the question of what those crystals look like. This problem has been tackled a number of ways in the past, often by comparing an observed phase function with one for a known ice crystal shape. Brittney took this approach, as did Alex in their Master’s work. However, we found that none of the modeled ice crystal shapes fit our curves very well. This could be for a couple reasons – the particles Brittney looked at were much bigger than the water-ice particles that we tend to see on Mars. It’s also likely that the water-ice crystals are not forming the same shapes they do on earth.

In the phase function paper, instead of forging along directly comparing our curves to known water-ice crystal shapes, we took a slightly different approach. It turns out you can make a simple approximation of a phase function mathematically using what’s called a Henyey-Greenstein (or HG) function. There are two values that go into making an HG function – the creatively named ‘b’ and ‘c’.  Helpfully for our purposes, the b and c values also give information about the particle shape. If we look at the b and c values we see in the Gale Crater phase functions, they’re close to b and c values for rough, irregularly shaped particles – not those relatively simpler geometries we see in Earth clouds. It’s not as exciting as actually having a picture of what a martian water-ice crystal looks like, but it is still a solid starting point.

The phase function is important not just because it gives us information about the shapes and sizes of the ice crystals in the clouds, but also because it is a critical input into various models. These include Martian global climate models (GCMs), which must include the effects of clouds on the amount of light that is transmitted through the atmosphere. It is also important for the topic of our second paper: the opacity of the ACB.

A cloud’s opacity basically describes how thick it is. An opacity (or “tau”) of zero means that there are no clouds and all light passes through.  As tau increases, more and more light is blocked, either through absorption or reflection into new directions (also called elastic scattering). In theory, tau can be arbitrarily high, but at a certain point so much light is blocked that it can barely be measured. During the Mars Year 34 global dust storm in 2018, Curiosity measured a tau as high as 8.5, meaning that about 99.98% of the Sun’s light was blocked by atmospheric dust (hence why the solar-powered Opportunity rover did not survive the storm).

We use a fairly simple model to determine the opacity of water-ice clouds. The math is not particularly exciting, but in essence it takes the amount of sunlight reaching Mars at the top of its atmosphere and determines how much “stuff” there has to be between the top of the atmosphere and the ground to explain the amount of light that Curiosity measures. The phase function is important here because it tells us how much of that light is being indirectly scattered towards the rover by the clouds.

The opacity of ACB clouds has been the topic of a number of PVL papers before this one, most recently by former member Jake Kloos in 2018. That paper covered the first two Mars Years of measurements. When we began writing this paper, we had just passed five Mars Years at Gale, so we were very much due for an update! We had initially hoped that this would be a fairly straightforward paper to put together. Because our model had already been well-established in our previous papers, we thought it would just be a matter of running the new data through the old model. Unfortunately, once we did so, the results pretty obviously made no sense. As previously noted, the ACB doesn’t change much from one year to the next, so we’d expect that the opacities would stay pretty much the same from year-to-year. Instead, the opacities output by our model for the new data were all over the place! They were neither consistent with each other nor with the old data, so we had to go hunting for a reason why.

It didn’t take much digging to find the cause. When we plotted the opacities as a function of each measurement’s distance from the Sun on the sky, there was a sharp increase as we got closer to the Sun. There’s no physical reason why clouds near the Sun should be thicker than those elsewhere, so our model was clearly breaking down in this area. The culprit, as it turned out, was the phase function. All of our previous opacity papers had assumed that the phase function was flat, taking on a single value of 1/15 at all angles. The results from the phase function sky survey have shown that this is very much not the case near the Sun, where the value of the phase function rapidly increases.

By assuming that the cloud opacities shouldn’t change very much over the ACB season, we were able to derive another phase function for ACB clouds, one that is reasonably similar to the one found using the phase function sky survey (which is good since we're all looking at the same clouds using the same cameras on the same rover!). After adding this new phase function into our opacity model, we were finally able to take a proper look at how ACB opacities have changed over five Mars Years.

In short, much like the phase function itself, they don’t really change at all, which makes sense given the consistency of the ACB between years. Notably, these new results invalidated one of the findings of Jake’s 2018 paper: that ACB clouds in the morning tend to be thicker than those in the afternoon. Although thicker clouds do appear more frequently in the morning in our new data, it doesn’t seem that this is the case generally. In fact, we found that observations in the morning tend to be taken closer to the Sun than those in the afternoon, which was artificially increasing their opacity values when using a flat phase function. Why didn't Jake include this in his paper? Without access to as much data as we have now, he simply didn't know that martian clouds behaved this way! (no one did) Therefore, while it can feel a little awkward calling out a former labmate’s paper as incomplete, science ultimately moves forward through incremental methodological improvements.

Just for fun, we also compared our opacity measurements with those taken by two cameras orbiting Mars: the MARs Colour Imager (MARCI) onboard the Mars Reconnaissance Orbiter (MRO), and the Emirates Exploration Imager (EXI) onboard the Emirates Mars Mission (EMM) Hope probe. Our methods did feel a little cyclical (assume the opacities don’t change much to derive a phase function, then use that phase function to conclude that the opacities don’t change much), so if we can match our ground-based measurements with those taken from orbit, we can have more confidence in our results.

Happily, the agreement between the MSL and MARCI/EXI measurements ended up being excellent, matching almost exactly with a few differences that can generally be explained by regional dust storm events that aren’t accounted for in the orbital data’s models. Thus, we can confidently say that our results reflect reality and probably aren’t a consequence of any assumptions that we made.

And don't forget to check out the papers themselves, available open-access at

Hayes et al. (2024)
Five Mars Years of Cloud Observations at Gale Crater: Opacities, Variability, and Ice Crystal Habits
&
Innanen et al. (2024)
Three Years of ACB Phase Function Observations from the Mars Science Laboratory: Interannual and Diurnal Variability and Constraints on Ice Crystal Habit