00:00 (S. Tabibian): Welcome to The Climate Map podcast! The Climate Map, an initiative founded by Covalence Global, outlines the complexities of climate change on a streamlined, action-oriented mind map. This podcast is an archive for our research, highlighting conversations with entrepreneurs, scientists, policymakers, and designers. You can learn more about The Climate Map at theclimatemap.org. Today we are excited to welcome Dr. Moet Ashfaq, a scientist for the Computing and Computational Science Directorate at ORNL. In the episode, we discuss the role of scientific modeling in Pakistan flooding. 

00:33 (S. Tabibian): Tell us a little bit about yourself. Where are you from, and what is your journey in computational climate science?

00:40 (Dr. Ashfaq): I was born in Kashmir, that’s an area between India and Pakistan. My early education was in Kashmir, and then I started graduate studies in Pakistan in two cities, Lahore and Islamabad. My background originally was in math and physics. I have a physics master’s and a Master of Philosophy in computational physics. That’s a very different background than what I’m currently doing. So, when I finished my Master of Philosophy in physics, I was looking for a PhD position somewhere in the US in material science. But at the same time, I was looking for some lecture opportunities to work, and at that time a new center in Pakistan was established in Islamabad. It’s called Global Change Impact Studies Center. I applied there. (They were looking for people with computational science backgrounds.) I landed the job there, and the center was focused on climate change. That was the first exposure to climate science for me, and after I worked there a little bit, I started liking it, and I realized maybe that’s the career I want to pursue. So after doing work there for a couple of years, I looked for a PhD position again in the US, but this time for atmospheric science. I came to the US, and I did a PhD. That’s how I ended up being a climate scientist, but I started my career as a physicist. 

02:11 (S. Tabibian): Awesome, can you please provide a high-level explanation of what global climate models are and why they’re important for our world?

02:20 (Dr. Ashfaq): Global climate models are physical tools. They’re the most sophisticated tools to understand our system in the past, present, and future. We use them for weather forecasting, for sub-seasonal season prediction, and also these are the same models that we use to protect how climate may change if the greenhouse gas concentration keeps on increasing in the atmosphere. These models are very complex. They require supercomputers, and not everyone can run them. You know the IPCC; IPCC reports are based on about 50 of these models, which are developed across the globe. I will say these are the backbone of all we know today about the earth system. These are the models on what we heavily rely on on what may happen tomorrow, in a couple of weeks, in monsoon season, and in the longer timescales. Now, what we did in Southern Africa was instead of using climate models, there’s another approach that we call empirical modeling, which is more mathematical. It does not represent our system in the physical way, but what it does is based on our understanding of how, say, precipitation is being rated, and what mechanisms affect precipitation generation. If we understand those mechanisms, then we can try to understand when there’s more than normal precipitation and less than normal presentation, what are the drivers of that precipitation variability? Once we understand those drivers, we can use that kind of information to build empirical models. These are very, very simple mathematical models, which are built upon this knowledge. Sometimes these models are able to provide better understanding than the physical climate models. And the reason for that is the physical climate models have their own issues. Sometimes, they don’t work well in certain regions. Southern Africa is one of those regions. We found that the simple, empirical model that we did based on our understanding of the climate characteristics of that region outperformed physical models. We compared it with the two physical models that provide this forecasting. So we did seasonal prediction this winter as a test case (which is winter in the northern hemisphere, but it’s the summer in southern Africa), and we found that our prediction, using the empirical model, could outperform the two leading seasonal forecast systems. (One is in Europe and one is in the US.) 

04:59 (S. Tabibian): Thank you for making that distinction between empirical models and climate models. What’s the difference between statistical downscaling and dynamic downscaling, and how do they affect global climate models? 

05:12 (Dr. Ashfaq): Now, first we need to understand downscaling. There are climate models that are used for future climate projections. As I mentioned earlier they are computationally very expensive. They operate with supercomputers to run, and due to that, most models have quite coarse resolution, around 100-150 kilometers. At that resolution, we can’t tell how a regional climate may change or how it will change in the next coming years or decades. But if you want to understand climate impacts at much smaller or local regions (a city, community, or state level), that 100-150 kilometer resolution is not sufficient. So for that, we do downscaling. We refine the climate change signal in these global climate models. There are two approaches, one is called statistical downscaling, and another is called dynamical downscaling. Statistical downscaling is just like empirical modeling; we develop mathematical relationships between high resolution simulations and global model simulations, and based on that, we downscale and refine those simulations at high resolution. There are some drawbacks of statistical downscaling, one that it depends on the relationships that it learns in the historical period, and we know that those relationships may not remain stationary. In the future, those may change, but that’s not the case in statistical downscaling. It considers them as stationary relationships. And other than that, statistical downscaling can correct the errors in the distribution of temperature and precipitation, but it cannot correct the processes in the models that generate errors in the precipitation distribution. On the other side, dynamical downscaling is just like running a climate model again, but at a limited or smaller area. (That’s why we call them limited area models or regional climate models.) These models use boundary forcings from the global model simulations, and because they are running over smaller regions, they are relatively less expensive, and so we can run them at higher resolutions. But again, the challenge is these are physical models. They are not perfect; they have errors in them. And even if we have them at high resolution, they may also have computational costs. So not everyone will be able to run regional climate models, compared to statistical downscaling, which anyone can do. Both methods have their benefits and limitations, and it depends on the kind of science question you’re addressing. You can make your best scientific judgment on which way is the best to go if you want to measure climate change impacts. 

07:57 (S. Tabibian): So when we’re looking at Pakistan, do you incorporate downscaling into that region? And if so, do you choose to use statistical or dynamic?

08:07 (Dr. Ashfaq): So in our research, we have done downscaling over South Asia multiple times. (Not in Pakistan, but in South Asia as a whole.) And we have done dynamic downscaling; we have not done statistical downscaling. One of the major reasons for not doing statistical downscaling for that region is that it heavily depends on good observations. And we don’t have good observations over South Asia. This is one of the challenges that we have in South Asia; that there’s not good density and network of ground stations and observations, especially in Pakistan. The last time that we did dynamic downscaling, it became part of the IPCC assessment reports. One of the chapters in IPCC assessment is regional climate change, and the data that we produced was part of that effort. 

09:07 (S. Tabibian): Wow, that’s very impressive. Focusing more on Pakistan, I was reading about how you discussed that the combination of multiple forcings is really what produces both extreme monsoons and also the frequency of excessive rains in Pakistan’s western margins. Can you elaborate on that and why Pakistan is facing such extreme weather lately?

09:29 (Dr. Ashfaq): Well yes, it’s true that at the turn of the twenty first century, Pakistan has experienced more extremes. At the start of the twenty-first century, we had a very exceptional drought that lasted several years, and then in 2010 we had strong flooding, and in 2022 we had exceptional flooding. We definitely had both kinds of extremes, wet and dry. In our research, we try to understand what is driving this increased variability of precipitation. Now, note that floods and droughts are common with this region; it’s not something that is new. If you go back in time, the region has always experienced floods and droughts, with some light breaks. So, we understood that there’s definitely a rule of natural variability because these things happen in the past, so we wanted to understand how much a role climate change has and how much a role natural variability has in it. Based on our investigation, we found that there are multiple drivers of precipitation variability over Pakistan, and what has interestingly happened over the last 25 years is that these multiple drivers are co-occurring. You know that if there’s one driver that can cause an extreme, and if there’s another driver that’s also present, then you can have a compounding impact. What we found is that over the last multiple decades these drivers are co-occurring and because of their co-occurrence, they have a compounding impact that is causing either extreme flooding or extreme droughts in the region. 

10:53 (S. Tabibian): Definitely, so why can’t we necessarily relate the cause of Pakistan’s monsoons to climate change?

11:01 (Dr. Ashfaq): The thing is we know there’s a common perception that the recent extremes in Pakistan is because of climate change, which is true in one sense and may not be true in another sense. What we have tried to convey is that we should not jump directly to attributing everything to climate change. There is an increase in extremes; we want to understand the drivers. And what we did is we removed the trends from precipitation temperature because we know those trends are different from climate change, then tried to investigate precipitation variability. We found that more than 70% of precipitation variability over Pakistan will last two decades, which can be explained just based on natural forcings. Now what are natural forcings? These are anomalies in mean surface temperatures, in different oceanic basins, these may be the atmospheric modes, like jetstream patterns or wave activity in the atmosphere, which naturally occur. We could explain, just by using these natural forcings, up to more than 70% of precipitation variability, just based on those factors. But as I mentioned earlier, most of these forcings (oceanic, atmospheric) are co-occurring. Now this co-occurrence is what is causing the compounding impact in this extreme flooding and drought events. And while the causes of these extremes are natural, the co-occurrence may be because of climate change. So what I’m trying to convey is that the direct effect is coming from natural forcings, but climate change is perhaps indirectly contributing to these extremes, just by causing them to co-occur. There’s another factor that we know that because of the warmer temperatures, there’s more rapid transpiration, and more rapid transpiration means that you have the same weather system twenty years earlier, it creates x amount of precipitation, that weather system today may be able to produce more because there’s more moisture in the atmosphere if there are warmer temperatures. That’s another indirect effect of climate change. The point I’m trying to make is based on our understanding and research of that region, we’ve found that natural variability is the direct cause of flooding and droughts in Pakistan and its increased variability, but the way these things are occurring and have occurred in recent pasts – that may be because of climate change. So climate change definitely has an indirect role, but may or may not have a direct role in causing these extremes. 

13:33 (S. Tabibian): Thank you for highlighting that nuance, how climate change may contribute to that co-occurrence, but the direct relationship comes from those natural forcings. In your articles, I think you also emphasize that we need to improve sub-seasonal variation representation in order to better predict extreme South Asian monsoons. What type of scientific breakthrough is necessary for this to happen?

13:59 (Dr. Ashfaq): Again, coming back to these extremes in Pakistan, what we found is there are five different drivers. Three of them are oceanic, which means they are related to the sea surface temperature anomalies, but two of them are atmospheric variability. The variability of jet stream anomalies, and other wave patterns in the atmosphere. Now, whenever we have a sea surface temperature anomaly, that tends to persist for a longer time. The most prominent example is ENSO. ENSO is a sea surface anomaly in equatorial pacific. When it goes in one phase, it stays there for months, right? Because it stays there for months, or whatever impact it has, it persists for a longer time so we can easily predict that. But if a region is affected by atmospheric variability, it has different challenges because atmospheric variability is more variable than time. A sea-surface temperature anomaly may persist for months, but the atmospheric anomaly may not persist for months, so it’s difficult to predict. So my point was, if we want to be able to predict these kinds of extremes that happen at sub-seasonal scale, we should have the ability to predict these atmospheric modes of variability, which currently we don’t have. Unless we have that ability, we will not be able to improve prediction of extremes at sub-seasonal scales, like the ones we have in Pakistan.

15:21 (S. Tabibian): Moving on from the predicting aspect and thinking about building up the capacity to have local data in the South Asian region, what are the issues right now with Pakistan only accessing publicly accessible global data for their weather predictions, and how do you suggest that the region builds up their capacity?

15:43 (Dr. Ashfaq): There are multiple challenges that Pakistan is facing that are also contributing to the reasons why the region suffers a lot whenever there’s a climate extreme. One, we don’t have any educational infrastructure. I can tell you that when I interviewed for that job long back before my PhD, I had zero exposure to climate science. Our curriculum and early education does not have any climate science as we have in the US. And even today, not a single grad school across Pakistan provides a comprehensive degree boarding program. That’s, I think, the unique situation that Pakistan has across South Asia. Other countries have much better ways, at least in educational infrastructure. That’s one challenge – so when you don’t have the educational infrastructure in place, you’re not able to develop indigenous leadership. Secondly, we have one of the poorest observational networks in the region. Nepal is 15 times smaller than Pakistan and has three times more observational networks. India has thousands of stations, while Pakistan only has 30-40 stations, and even that data is not available publicly. So the issue is that when climate modelers develop models, they validate their models using station and ground truth observations, right? And if they don’t have access to this kind of data, how can they improve models over this rate? Most of the models are very unreliable, partly because we don’t fully understand climate statistics of the region, and again the reason is that we don’t have good data. So, without having good data and without having educational infrastructure, it’s very difficult to build leadership and build the capacity to do forecasting. Forecasting, again, requires a set of models that you need to use, and if you don’t have that kind of educational infrastructure, you will not have any ability to develop those models and run those models operationally. There are other requirements too, like you need supercomputing and other sources. That’s why Pakistan 100% relies on the seasonal forecasts around the global prediction centers, which actually is not wrong because many countries which are less resourceful use global forecasting from these global forecasters. But the other issue with Pakistan is there’s now clear mandate given to any agency currently about who’s going to do this kind of forecasting, and they have a meteorological department, which should be doing these kind of forecasts and seasonal predictions, but these days you can see there’s a national disaster management agency that is dealing with  it. This is kind of unusual because NDMA (that’s what they call it) is like FEMA in the US. FEMA doesn’t do forecasting here; FEMA only appears once there’s a national disaster related to climate, and they try to bring solutions to suffering communities. In my opinion, this lack of infrastructure creates a quagmire where no one knows what to do. There are multiple challenges that they have. So they need to improve educational infrastructure; they need to have a clear mandate given to one agency, and equip that agency so that they can do seasonal forecasting indigenously. And then also they need to have an observational network much better than what they have currently, and they should also make it public if they want models to do better in that region. 

19:26 (S. Tabibian): Absolutely, that was actually one of my questions, “Do you think that Pakistan should create a more specialized agency just for forecasting, like the US?” And I was also wondering, in terms of building that network, do you think that Pakistan should allocate its own funds toward that, or do you think that it should look towards aid from other countries that are wealthy and high-emitting?

19:48 (Dr. Ashfaq): Well, again the thing is, say you build a hospital. You bring a lot of modern machinery, but you don’t have technicians, and you don’t have good doctors to use it. Interpreting the results is no use. After some time, it will become useless; it will stop working. That’s the same situation. I think Pakistan needs to invest its own resources more into educational infrastructure. They will not get that through foreign aid, and they also devise a clear mandate about who is going to do what in the climate domain. Any aid agency is not going to tell them what to do. This is what they have to figure out on their own. They have the institutions; they have a meteorological department. That department has thousands of employees. They have a good network. They have people who can interpret climate information from other sources and ingest local knowledge and expertise and provide better forecasts. I think it’s good for them to take these kinds of steps on their own. Also, this habit of reliance on foreign aid is not good. If you develop this habit, you will not ever develop your own capacity. You have examples in India. India also suffers from floods, droughts, and other natural disasters. I don’t remember in the recent past that they’ve ever asked for international aid to deal with their challenges. Pakistan, on the other hand, whenever they have some sort of issue related to the climate, they just go to foreign aid agencies. I understand they don’t have much resources to deal with that, but I think they need to make more efforts to become more self-reliant on this. 

21:42 (S. Tabibian): Absolutely, overreliance on foreign aid is really eroding their internal systems and abilities to address their own problems themselves. I really liked your article on the at-risk populations. There’s clearly a deeper-rooted corruption existing within the country. What are the current issues, both political and societal, with climate justice for at-risk populations, specifically in relation to flooding infrastructure?

22:09 (Dr. Ashfaq): One thing honestly I can tell you is in Pakistan, I don’t think they even know which population is at risk because they don’t have any developed knowledge about that. One of the major challenges that they have is political instability. What political instability does is it causes governments to make policies which are not sustainable. Every three or four years a government changes, a new government comes, brings on policies, so one knows what may happen five years down the road. So you don’t have any long-term sustainable policies, which are needed to develop and establish climate-resilient communities. Secondly, as you mentioned, Pakistan is a country with a very poor justice system. If you don’t have a strong justice system, there’s no accountability. And if you have no accountability, that leads to corruption. And corruption, if you look at past records, is deeply rooted, especially when foreign aid comes to the country after a disaster. Most of the aid does not reach the communities; it goes somewhere else. And because of that, you get aid in the name of the poor at-risk communities to develop the resilience to climate change, but that does not trickle down. And because of that, it’s a vicious cycle where you get to an extreme, then people suffer, you get aid, aid will go somewhere else, people still suffer. So people who are below poverty stay below poverty, and they never have a chance to actually get out of poverty. There’s no easy fix for a country that has multi-dimensional problems, so I don’t think that they have a clear or easy solution for this. But I think the climate crisis is one of the major crises that they have no clue how to deal with because of these multidimensional challenges.

24:16 (S. Tabibian): Definitely, thank you for that perspective. To close off, looking forward, what are some of the goals in your career to continue with modeling both empirical and global climate models. Will that perhaps relate to the South Asian or Pakistan Region?

24:33 (Dr. Ashfaq): There are many goals. One of the goals is to develop systems that we can use to do better predictions. If you can develop empirical and physical modeling systems that can inform about subseasonal extremes well ahead of time, that may enable governments in the region to prepare and do better disaster planning. That’s one of the goals, and we are continuously doing research to identify different mechanisms and the factors that contribute to extremes and trying to devise ways to use this knowledge to convert that into a solution system for better predictions, that’s one. Secondly, one of my goals is to develop better climate services, which means that I would like to make the observational record public so that it’s accessible to scientific communities to validate our models and create better models that are more reliable. Also, we’re trying to interact with the younger generation, trying to see whenever there are opportunities to train them and build future leadership in climate science. Other than that, we do high-resolution climate modeling over South Asia, that’s one of the regions. and we do have plans for doing our next iteration of high-resolution modeling. Pakistan is part of it, and that data is going to hopefully be used in an IPCC report in cities, which will be available in 2027-2028. South Asia one of the more focused regions for that. 

26:29 (S. Tabibian): Awesome, well those all sound like very impactful goals. Thank you so much for your time!

26:35 (Dr. Ashfaq): Thank you very much. 

26:36 (S. Tabibian): And that is it for The Climate Map today. Please visit theclimatemap.org to learn more about how you can get involved with us.