The Clouds of distant worlds: a chat with Dr Sushuang Ma

Welcome to Growing Worlds, the podcast where we explore the science of exoplanet through conversation with early career researchers on the front line of discovery. I'm Lorenzo.

And I'm Jenny. In each episode, we meet a guest who takes us behind the science of planets, stars, and the search for life in the universe.

Today we are turning our gaze upward to one of the tickiest components of exoplanet science, clouds. When you hear clouds on other planets, you might think of fluffly white, vapor-like on Earth, but it can be more exotic than that. On exoplanet's clouds can be silicates, sulfur compounds or entirely unfamiliar particles, and they play a profound role in what we observe.

To help us understand how scientists model and observe those clouds and why they matter for our understanding of distant worlds, we're joined by Dr. Sushang Ma from King's College London. Sushang, thanks so much

Hello.

for being here.

Thank you for inviting me.

Sushang is a researcher at King's College London whose work centres on modelling and retrieving of clouds in exoplanetary atmospheres. she is the lead author of the study Yun Ma, which represents a new modelling framework for clouds' microphysis and additive transit in exoplanet atmospheres. In her research, she explores how to infer cloud composition and size distribution from the observed transmission spectrum of exoplanets. And how we must go beyond the simple opaque cloud model if we are to interpret the next generation of observations.

So Sushang, we always like to begin with your personal journey. So why don't you tell us what kind of drew you into studying exoplanet atmospheres and how did you get into cloud? Was there a particular moment or was it a slow evolution?

It was both and it was a long journey. a lot of people,

that's good.

yeah, like a lot of people, I was always fascinated by the extra solar world in general, but I was never like thinking I can actually do it. So I chose an engineering degree to do, but I was good at it, but I didn't enjoy it. whenever,

Okay, yeah.

yeah, yeah, it was a little bit. I don't know, it was not easy, but meanwhile it's not challenging in way that I want to be. So whenever I get time in the break, I just go to library, right, I find a corner and there are magazines and books of general physics or astrophysics and I was

Yeah.

going to watch, like read them. Of course I never told any of my lecturers, like I'm going to read

no!

something.

No, of course not. They always thought I'm going to pursue engineering or at least a computer science route. Yeah. And one day I just read about this James Webb telescope launch, but I remembered I read a similar article a while ago when I was a teenager. I read it before. I was like, how could like it's launched twice. Then

Yeah.

it was

Yeah, that is not your fault.

Yeah.

I think it's probably on JWST

Yeah

that it takes a while to actually be launched.

Yeah.

I thought I might have dreamed the first time I read the content. I might then read again. Okay, it was too late.

Yeah. Yeah.

And that way though, it's even more delayed. It was actually launched, not launched until I was almost to start my PhD.

Mmm.

So I first read it when I was a small kid.

Yeah, yeah, I think if I recall correctly, the original planned launch date for JWST was 2007 with a $500 million budget and it was eventually 2021 with a $10 billion budget. So 20 times over budget, 14 years late, better late than never because that was fantastic.

I'm pretty

and

sure that when they designed the telescope thinking about the next generation of scientists, they weren't thinking about growing up together with the next generation of scientists from middle school to actual the doctorate.

That's true, yeah.

Yeah, it was, but it was a moment because I

Mmm.

read the article carefully because I was thinking maybe I had a weird dream and created a future or something. And I read, read it, wow, they can actually do these by exoplanets. I feel that the magazine was more general. It's not towards exoplanets, especially like, but different topics in astrophysics. Okay, I hope one day can use a telescope, right?

And then I got really bored in meantime. Well, I hope no audience is studying engineering or like a problem.

Hey, but look, that's okay because just because it's not your cup of tea doesn't mean it's not anyone else's cup of tea.

Yeah, yeah, don't be wrong. Engineering is very meaningful. And yeah, if you like it, please keep going. And I started to apply for anything in astrophysics, planetary science for a master's degree. But it wasn't easy because it was a jump of the major. but somehow I got enrolled into that time was the University College of London. And I got into this course for planetary science. I was very excited. was like, okay, it was also COVID. was, yeah, my family wasn't supporting

Yeah, true.

it. Definitely not. And I was like, no, I need to go because this is the only chance I can change my major. Then it was difficult because a lot of astrophysics courses, I've never taken them. I only have the knowledge of the basic physics and mathematics. Yeah, so that year was particularly hard, but

Yeah, like a really steep learning curve.

Yeah. And also it's a self learning progress. And then I met Joanna, who's my supervisor. She's the PI of Ariel mission. Lawrence has something to say. Yeah.

Mm.

I just want to clarify for the listeners that when Sushang so friendly say Giovanna she means Professor Giovanna Tinetti, is the PI of Aerospace Mission.

Hehehehe

Yeah, somehow she just think I'm a very keen person. So she gave me this opportunity. thank you.

think you are! You seem very keen on this podcast, so...

Yeah, of course, like I like sharing the story and inspiring the people who are also interested. Because it wasn't straightforward to me to be in this field. Like a lot of things come by luck, but also come by effort.

Yes.

I forgot to mention, even when I was in my bachelor, I go to the relative courses. So I finished my major. And meanwhile, I sit in other classes and I beg the professor, I take the exams?

at the same time. So I was doing two majors, but I only get one degree because that time it wasn't like, it was a journey.

my goodness. Yeah.

And Joanna later gave me this project to do, right? And then she think, you did well. How about considering PhD? I like, yes.

Nice. I love that. Yeah. Cause

Mm-hmm.

it was a bit similar for me. Like when I was doing like the masters, it was some of my lectures said to me like, have you thought about doing a PhD? And I was like, no, not, not because like I didn't think I was capable, but just because like no one in my family had done a PhD and I didn't know anyone who had like done a PhD. So it wasn't like something on my radar. So I really like it when the lectures like encourage you to take that next step and everything.

Yeah, I think it's fair to say that you find the right mentor can really change your life.

Yeah, 100%.

Yeah, absolutely. I feel it's half like the research you do before choosing, half it's really by the opportunity and luck. Let's see, yeah.

Yeah, it is, it's like half what you know, half who you know. Because you've got to have the right people around you, but you've also got to have like that capability.

Yeah, yeah, absolutely.

So then you get onto the PhD and the PhD is all about clouds.

No, it's actually not. So I finished Clouds. Well, the Clouds was the starting point because it will ensure I can graduate from PhD. Also, I have started doing it while in Masters. I just refined it, especially in the Retrieval Framework and tested it, published a paper, then put there for some people in the group to use. And then we got the James Webdata public.

Like it was after the commissioning of the first batch of data and initial publications. And I saw this data and I was like, okay, maybe I can, I can do something now with my cloud model because they didn't contain that in, into their initial analysis, like a comprehensive way of cloud microphysics. Then I realized it's not only the cloud.

So cloud is just a part that needed to be optimized, but it's really about the chemistry, the temperature of the atmosphere and the cloud together. So I started to refine this methodology. I don't know how chance to talk in detail later, but it's about how to find a sweet point between you do the complex physical models using the theories, but meanwhile you have the observation.

There are data limitations. There are certain information in the data. How do we go from two direction, meet the sweet point of the best fit of the model to the data? And in that research, put the Watt's for Therianbiasi case. And I found like there might be a new explanation of the atmosphere using a new chemistry there.

That was exciting. It wasn't planned at the beginning of the MUTD. To be honest, at that time I was doing the large scale simulation. Lorenzo... I was doing it

You

Lorenzo. Can I say it?

Of course you get, of course.

So we were doing the large scale simulation, basically a population simulation and retrieval to prepare the aerial mission. And then one day I disappeared. Lorenzo kept asking

Ha ha!

me what I doing.

and I'm still waiting for her to reappear in my radar.

I'm so sorry. That was because I found these new things in this planet and I have to focus like it was just out

Mmm.

of plan.

I'm glad that you are mentioning this because this is real life in science. You're just working

Yeah.

with somebody on something very exciting and then let somebody disappear because something more exciting is just coming up and then eventually they will be

Yeah.

back, I hope, right?

For a fact, I think I disappeared for not three years, but almost.

my goodness, are you going back soon?

Yeah, very soon. I will wait for three years to reach three years and then go back. Yeah.

Yeah, on the anniversary. you

I'm glad that this is our record now. You can back down now because it's recorded. I can use this against you.

yeah exactly it's written

No way!

yeah he's written on the airwaves so

but focusing on your other work, the one more exciting that brought you away from me basically.

Mm-hmm

You say that you need to find the sweet point between complex models, including the clouds, to understand the chemistry and to get actually understanding of the data that we got from space. to help us understanding this complex thing that you are doing and you are bringing us today, why don't you start walking us through why clouds are so important in exoplanetary studies?

and how they are changing our understanding of what we see.

First of all, from the current understanding, of course we don't know everything about exoplanets because these are distant worlds, we have limited observational data, it's just the beginning. Many exoplanets, we see they have clouds in it. Even before James Webb we only have very limited data points. People were thinking they were proposing some of the continuum.

or when you see a flex spectrum, these are caused by clouds. That seems to be a most reasonable explanation. And we also see the clouds in the solar system planets. So it should be there. But why we don't have these cloud models before and I needed to work on it, it was because it was just a couple of years ago before James Webb, we only have limited data. For example, Hubble.

WFC3, 1.1 micrometer to 1.7 micrometer. This band, we only see certain absorbing features from water or mizan or a couple more, but not very direct signatures. And that part, usually the cloud particles, the features are quite featureless. it's

Okay.

like, yeah, for example, if it's 10 to 30 micrometer, it would be flat, almost flat.

completely. Maybe you'll see a little bit of the scaffolding slope, but not significantly. So there was no need. There's a word that Lorenzo taught me, the information content of the spectrum. I don't know if you created it or not.

I haven't created that but thank you for giving me the credits.

It was quite useful. If your data doesn't contain this much information, then you should probably not to put this much complexity in your model and

Yeah.

trying to fit it because that might be biased by your model. It's not

Yeah.

going to be constrained by your data. But because the cloud, if it's thick, it's going to block the signatures from below the cloud deck. If it's semi-transparent, it will block a little bit so make it more flat seems to be more flat In this case, it makes it harder to determine for the major traces, for example, commonly we see water coming out of the site. How abundant are they in the atmosphere? It's difficult because the cloud has attenuated signals. And for

Yeah. Yeah.

those minor traces, because their signature wasn't significant at the beginning, even if there's no cloud, but if you have cloud, you cannot see them.

And to have a comprehensive understanding of the atmosphere, including all the aspects, let's say the chemistry, clouds, temperature, or where it becomes opaque, we do need to understand this component, at least its impact on the spectrum.

When we say spectrum, it's the transit depth spectrum. The transit depth we measure when the planet transit in between the observer

Okay.

and the parent star. The atmospheric part of the exoplanet will block, but not entirely, the starlight.

Yes, yes, so some will like make its way through that atmosphere.

Yeah, and it's waveless dependent.

Okay, so depending on how that transit dip changes, depending on the wavelength you're looking at, you can then get information about the atmosphere and the clouds.

Yeah, exactly. But usually the cloud part of the signature is not that obvious.

Mm.

But when we have wide coverage of wavelengths, some...

which we have now with like JWST instead of just Hubble. I see.

Exactly. And also the forthcoming Aero mission where Professor John Tintez appeared.

Yeah.

Yeah, it makes it possible for students to optimize, of course, the impact of the cloud on the spectrum. And it helps us to better determine the abundance of the gas phase chemistry.

Okay, because the makeup of the clouds gives you an insight into that.

I would say it's the necessary part we need to into fitting the spectrum.

So what sort of clouds can exoplanets have? Because I guess our clouds are water-based, right, mostly, like

Yeah.

pretty much. are we always looking for water-based clouds or are we looking for clouds of different materials? I guess it would depend on like temperature and things like that.

What Jenny's asking is

Absolutely.

that if she's looking at the sky while it's lying on a green grass field on an exoplanet, what kind of shapes can she see on the sky basically?

Yeah, that's the real question. Like am I, no, am I seeing a horse? I'm No, but seriously, it's like what sort of like clouds content, are they all water vapor clouds or are we looking at different types of clouds?

No different types of class, but all in the bear shapes. I'm joking.

Yeah, all impatient! You

You're correct. It depends on the temperature. But there's another component is what's the baseline, those gases that are going to form the clouds. So imagine you have this amount of water on the Earth. And certain part of it is going to form clouds because it saturates. So it's two major factors. But also, The gas phase abundance depends on the temperature and other factors like the elemental ratios and type of chemistry in the atmosphere. My temperature is really a key parameter in determining composition of the clouds. Like in the colder region, like the Earth's or the Jupiter temperature, we have the ammonia and the water clouds.

Okay.

Going further, you might have those chloride or sulfurite clouds. And when you reach like a thousand to a thousand five hundred Kelvin, now it's in hot Jupiter region, it's a critical hot. We don't have this planet

Yeah, too hot. Yeah, yeah.

in our solar system. These are closing planets to their parent stars. Silica clouds don't mean it. This is our current understanding, but maybe, maybe.

Yeah,

Yeah, maybe a month later.

like silicates like like rocks are made of silicates. So you've got like floating Like yeah

Yeah, yeah, either sky, can imagine?

God that's wild. I guess this is where, because you see sometimes they're like, on this planet, like, rains, diamonds. And I guess like that's the kind of thing that they're on about because it's like carbon and it's hot or something and so like it makes, it makes diamonds. I guess that's like, so you can get all sorts of crazy clouds.

Yeah, possibly. if they

Yeah.

have the drastic temperature changes, it may form something really crystallizes. Yeah,

yet.

the dimension might be possible. And different colors. Then going

Baby! Yeah.

further, guess what? You might have iron clouds.

my goodness. Yeah, if it's hot enough. Gah,

Yeah.

that's wild.

And how are all these different kinds of cloud affecting the actual transmission spectrum that we were mentioning before?

their scattering and absorption features are different. So first, the composition will shape a little bit of these overall scattering and absorption features. And we see in different wavelengths in transit, we mentioned before. And then the specific particle sizes will also change a little bit.

where you are going to get more light, where you're going to absorb more light.

So are these changes, they're really sort of subtle or are there certain ones that kind of pop out?

certain ones pop out.

okay.

Yeah, but it's.

And that's why they're so important to factor into the models and things like that.

Mm-hmm. Yeah. And it's really case dependent. And

Okay.

yeah,

Okay, so it's like the the signals of like different gases in the atmosphere will always be the strongest but then the clouds have a very unique signature that's smaller but it's like you can't mistake it that's a cloud signature.

It's not a, but a continuum, a list of small signatures. Some pop out a little bit more than others.

Yeah. Okay, yeah.

That's why we need large wave length coverage to determine which one is it.

So how do you kind of factor all of this into a model to compare to the data to kind of like analyze what's going on? Like how do you, yeah, how do you pull all of this together? All of the different effects of all the particles and I just, how, yeah, how? I just can't even conceive how you even go about starting to like make models to then interpret the data.

I need to refer to a technique first. It's called the biogen technique, retrieval. If you have heard about it, the retrieval studies in the exoplanetary transistors spectroscopy, we have the data, we have the model, how to fit them together. So

Mmm.

in iterations, we explore a parameter space. So basically this space. I might have this parameter for cloud, this parameter for water, this parameter for methane, and we try different settings. When I say

Okay.

different sets, it's like 10 to four to 10 to six sets, this much.

Wow, so tens of thousands, even

Yeah.

hundreds of thousands of options. So it's like, one will have say 10 parts per million of methane, and then the next one's got like 15 parts per million of methane, and then

Hahaha.

the next one's like 20, and then you've got to start changing the carbon dioxide, and then you've got to like change the

yeah!

water, so you've got like all these different degrees of freedom, and all of them have to be varied, and wow.

Then we let the computer decide which sets, like we won't decide one of these is the best, we decide the region that's more likely to be the ground truth.

Hey, yeah.

So same for clouds, when we think about it, if we think from a formation perspective, you have the nucleation, you have the growth, the condensation, calculation, but does the spectrum contain this much information? The answer is no. So we start to think from the observational perspective, so different direction, and see, because we are going to use this biotron techniques.

or let's put directly retrieval here. What can we retrieve? So the most direct thing that determine the cloud impact on the spectrum is going to be most likely retrieved. What are they? So the composition of the cloud species and the cloud particle site distribution. So species, yeah, we can still infer from the temperature and the composition of the atmosphere. But the particle size distribution is really complex. What

Mmm.

can we do? So there are studies of simplifying a certain distribution into one radii for a certain layer. So there's simplification. Then we can retrieve, like, which is the radii that can give us this relatively close to the actual signatures of clouds into the transit depth.

So is it like the, you do sort of lab experiments that will then tell you, this range of size of particles is possible, but nothing else is. So it kind of limits your options a little bit and makes it easier. Is it like that?

or kind of possible compared to the data. Let's see.

You mean the lab results of the ready-to-be-transfer properties or like in this condition what sizes

the sort of...

can it form?

yeah, the sizes of the particles.

this part is from the settling models. So what's it?

Okay, so it's not lab experiments.

Yeah, it's more from the gas kinetics and observation of the Earth's clouds.

Okay, yeah, so like, our planet is the laboratory.

Yeah, of course, because it's easier for us to do the in situ observations. I believe there are more data from the solar system planets like Venus. Yeah,

Yeah, okay. Yeah.

but it's more complex because they have a very thick cloud and like it's different color layers combined together. I haven't seen the data, but I hope in the future

Yeah.

I can have a look.

So like by looking at Earth's atmosphere and the size of the particles are like in

Mm-hmm.

our atmosphere, depending on like the pressure and the temperature, then that goes into the kind of exoplanet models, is that right?

We do consider the draft balance. So it's a balance between the upward turbulent mixing and down sedimentation. And with that draft balance, give the properties of the atmosphere. So you think about a particle that's like traveling into this viscosity flow and either the growth or it just dropped down or

Yeah.

it just suspend

Okay.

like, yeah.

Yeah.

Yes, and this approach is originally from Echman-Mardy. They have Cropas model. For me, it's more like from a numerical perspective to make it more accurate. And then how to combine it, connect to the radio transfer, and combine it into this 10 to 4 to 10 to 6 times iteration.

Yeah, okay, yeah. And so then once you've got your sort of the area of parameters that fit, then that gives you your idea about what the temperature and the pressures like in these clouds, what they're made of, like what the size of the particles are, we have plus or minus a bit, right?

Yeah, I need to mention everything is down together with the chemistry, like the gas phase abundances, because you cannot ignore those signals. And also

Yeah.

the baseline gas abundance really indicate a little bit how much these clouds are going to form.

Yeah, yeah.

Okay.

So you say though that it's quite difficult to get all these details and there is a balance between what you can model and what you can actually detect, right? You mentioned before that fitting the data mostly depends on the data and moving from Hubble Space Telescope to JWST we have larger data set, larger wavelength range, and we can do more. So now that we are moving to the area that as you mentioned about JWST and aerial and possibly beyond, what do you think will be the best improvement that we're going to have? An improvement in instrumentation or in modeling? What can be the difference in the future? What are we going to learn more compared to what we know now?

You mean from the modeling side or from the data side?

I am an informed data side because I'm a technical guy, but I'm happy to listen also what you see on the model side.

Okay, I'll be a friend. from this, said, in simulations, tried to decrease or let's say increase the SNR because a simulation can do that. When you increase the SNR, you can really constrain more things like even the eddy diffusion, like to have more...

Wait a sec, when you say increase, sorry if I'm not adapting, when you say increase in

Yeah.

SNR you mean that the signal is easier

yeah, I forgot to mention it.

to detect compared to what is actually our uncertainty on the measurement.

Of course, SNR is signal to noise ratio. So signal is more clear. are absolutely correct. So it gets much easier. Basically, the information content you taught me is going to increase a lot. So you can constrain much more precisely the atmospheric processes that's related to cloud formation. In that case, People can think about even more detailed analysis on the cloud particle size distribution. And remember, the clouds are not composed of single type like chemical compounds. It's actually a mix, but we were not able to do it at the moment. But I'm saying...

When I increase SNR, I increase it to much higher than the capability of James Webb and Ariel, unfortunately. So realistically,

Haha, yeah.

what can this instrument do now is to limit the effect of clouds in determining the gas phase chemistry processes. We might propose different scenarios at the same time because there are some uncertainties of different type of clouds and their partial set distributions. even

Mmm.

in the relative transfer, even you have the partial set distribution composition correct in simulating its impact on the spectrum, you have this laboratory data that's not very certain. Yeah.

Yeah. So what do you say is the biggest kind of challenge now for trying to infer all of these cloud properties? it the models that are limited or is it that signal to noise ratio in the data?

Can I say it's a signal to noise ratio of the data? And yeah, yeah.

You can, you can say that. That's absolutely fine. I'm here to mediate between the modeller and the data

Hahaha

analysis, yeah.

Yeah, and... Yeah, who's fault?

It's always our fault.

Yeah so you think, so do you think like the modelers have kind of got a pretty good handle on the atmospheric science but now we need that more sensitive data which we're starting to get with JWST but it's really about waiting for those future generation of telescopes like aerial.

You know what? Because our model are based on the Earth data. Or like if we increase a little bit the boundaries, it's to the solar system data. But exoplanets are different. Our solar system is quite unique. If we

Yeah

don't have data, we cannot refine our model. It's like the chicken and egg, But in this case,

It is, yeah.

I really think the way to push it is the data. And not only the exoplanet data. also the aerosol, like when I say this, reactive properties of these cloud particles from laboratories. But one thing is

Okay, yeah.

difficult is how can we make these, for example, silica clouds on the earth in a laboratory? It's a very high temperature.

Yeah, yeah, it is quite difficult. So yeah, so

Yeah.

I think it's a little bit of column A and a little bit of column B in the sense of we need more sensitive data, more higher resolution data from the telescopes, but the models need to also have some work with more feeding from the lab experiments and more detailed studies of atmospherics to kind of feed into them. So look at me keeping a happy medium.

Between the two of you Yeah,

Thank you Jenny, you're so diplomatic.

Thank you.

I try to be I try to be

So we will be in much better mood like when the three year hits, when we start the collaboration again. Yes.

The aerial stuff, yeah.

So I think before we wrap up, we always need to finish our last signature question. And our signature question is this one. So if you had to pick one celestial body, star, planet, cloud layer, something else entirely that reflect your work or how you think about it, what would it be and why?

for sure to be an excellent with clouds.

Yeah.

I think I'm someone who likes the big pictures. I always think from the big pictures, then go into deeper questions, like specifically one point I'm interested in and I think in a complex way. But I'll start with the pictures. So it will be something like bigger, like the hot computers. I'd

Yes.

say like the one I'm actually studying was for 39, it's a good example. So it has the cloud layers, right? So you do go deep and think what's beneath there. Yeah,

Yeah.

I'll pick this one.

Nice. I like that.

That's a really nice one. Thank you, Sushang. Thank you for lifting the veil of clouds, showing us what hides in the sky of other worlds that often hold the key to understand them.

And thank you to everyone listening. Join us next time on Growing Worlds as we explore another corner of the universe and the people bringing it into focus. Thanks for sticking around, Shishan. It's been brilliant.

Thank you.