How can we understand the local impact of climate change on our communities, now and in the future?
All year, we are keeping up our focus on the climate crisis here on the blog and podcast.
One of the biggest issues with understanding the impact of climate change is that the data on climate-related deaths and economic losses are incomplete. While FEMA’s National Risk Index attempts to help us understand the situation better, we have identified several opportunities to improve on their work by using more granular data and alternative methods drawing on data science.
This month’s podcast features a roundtable on the new Local Climate Impact pilot project from the RTI Rarity team. Special guests Aditya Vadalkar, MS, Bahamin Ayla Akhtari, MS, and Sourabh Deshmukh, MS discuss our project, focused on three domains: climate-related risk, community resilience, and community vulnerability.
We also present interactive maps of our in-progress dashboard featuring California and Florida. The dashboard allows us to explore those states’ current risk, resilience, and vulnerability along with simulation modeling of the potential local impact of climate change in the future.
Improving Measurement
- The NRI uses the Social Vulnerability Index (SVI) to measure community vulnerability. Drawing on 16 variables, the SVI is a composite measure based on factor analysis.
- Our team used the RTI Rarity dataset. We identified more than 100 measures of social and community context to better capture vulnerability at the Census tract level.
- The NRI uses the Baseline Resilience Indicators for Communities (BRIC) to measure local resilience. BRIC includes 49 county-level variables in their model. The variables are grouped into 6 domains, and the BRIC index is on a scale of 0 to 6.
- Our updated resilience sub-score includes 65 variables, is on a 0-100 scale, and is measured at the Census tract level.
Audio-only version available at Podomatic, Spotify, Apple, Amazon/Audible, and anywhere you get your podcasts!
Transcript below.
Lisa Lines: Hello and welcome to the Healthy Intersections podcast for August 2024. My name is Lisa Lines and I’m the host and producer of this podcast. I’m also a senior health services researcher at RTI International and I lead a team called RTI Rarity, where we are using data science to try to answer some of the tough questions that trouble our society in terms of health outcomes and the impact of some of the big health events that we’ve been experiencing and pandemics and syndemics and all of that.
Today I’m excited to have the opportunity to introduce to you three of our data science analysts on the RTI Rarity team, and we’re going to talk about an initiative we’ve been working on called the Local Climate Impact Score.
This is just a pilot project that we’ve been working on this summer, and the goal was to try to use our data science chops to try to model out what the local impact of the climate change is now and in the future.
So we’re going to be talking about that, and before we get to that, I want to take a moment to introduce everyone on the team. Actually, I’m going to have you guys introduce yourselves. Is that okay? All right, so we’ll just take it in alphabetical order by first name. So we’ll start with Aditya.
Aditya Vadalkar: Hello guys. I’m Aditya Vadalkar. I have been working with RTI for the past year. I started working as a data science intern and currently, I’m working as a data scientist.
I have done my master’s in computer science from the University of Southern California and I work at RTI. My main role is creating dashboards, working on machine learning models, and data pipelines. And I’m looking forward to more interesting work like this.
Lisa Lines: Thank you. Baha?
Bahamin Ayla Akhtari: Hi, my name is Bahamin Ayala Akhtari, but everyone calls me Baha or Ayla. I am currently an intern at RTI Rarity and I have just completed my master’s at the School of Biomedical Informatics. I’m excited to be here with everyone.
Lisa Lines: Thanks, Baha. And Sourabh?
Sourabh Deshmukh: Hello everyone. I’m a data scientist at RTI and I’ve been working as a data scientist for about a year over years. And just like Aditya, I completed my master’s in computer science from the University of Southern California, which is in LA, in May of 2024.
And yeah, it’s been a great time at RTI working on some challenging and pressing issues. Looking forward to sharing my ideas on this podcast.
Lisa Lines: Yeah, thank you so much all three of you for being here. I’m really excited to have this conversation about the work that we’ve been doing on the local climate impact model.
So, let’s get started. Let’s just dive right in. First, let’s set it up and give our audience a little bit of background information about… First of all, we start with looking at existing models that are out there and trying to figure out how we could improve on what’s already out there.
We started looking at all these different things that are out there that others have done, and we landed on the National Risk Index, which is something from FEMA, which is an index that tries to get at community vulnerability, community resilience, and climate risk. So, we started with the National Risk Index and trying to figure out how to actually improve upon it. Let’s talk about some of the drawbacks.
The National Risk Index, they use the baseline resilience indicators for communities, BRIC scores. Baha, can you tell us a little bit about the BRIC and what it consists of, and then maybe some of the reasons why we wanted to do something a little bit different?
Bahamin Ayla Akhtari: The BRIC index, it was developed by the University of South Carolina, and they basically categorized resilience variables into six categories. And that’s environmental, community, infrastructure, institutional, economic, and I think I’m missing one more. Oh, social. Social as well.
Yes, so what we did was we looked at their resilience variables and we had also looked at our resilience variables. Basically, we had gathered a bucket of resilience variables and we had looked at theirs to see if we are lacking in any. And basically, the ones that we had not included that the BRIC index did have, we also implemented in our analysis.
The baseline resilience index for communities was basically created as a way to monitor existing resilience variables to natural hazards. And what is different is that they did it at the county level, whereas we’re doing it at a census tract level.
Also, they had a score of zero to six, which is a bit simplified, if we want to be frank. It’s a bit simplified for resilience factors for communities, and we wanted to get a bit more granular to be able to show how communities are resilient.
Lisa Lines: So how many variables do we actually include in our resilience category right now, do you know?
Bahamin Ayla Akhtari: Currently, we have about 55, but we’re adding another 10.
Lisa Lines: Okay. Yeah, that’s great. And then let’s talk about vulnerability a little bit. The way that the National Risk Index measures vulnerability is with the social vulnerability index. I think that, Saurabh, you might have something to say about the SVI and why we are doing something completely different.
Sourabh Deshmukh: Right. As you just said, the NRI uses the social vulnerability, which accounts for the vulnerability of a community. Where we set ourselves apart in calculating the LCI is, I think the SVI has just 16 variables, if I’m not wrong. They only use 16 socioeconomic factors. And on the other hand, what we have is we have a pool of about 109 variables across different buckets like economics, health. Yeah, that’s something which sets us apart.
And apart from that, we do not take into account race and ethnicity, which we believe is not something that we can work on changing in a community. Because ultimately, race is not a vulnerability, racism is. So that’s one more thing.
And apart from that, some of the variables that we have that are not in the SVI bucket of theirs would be conditions that represent infrastructure, for example, like number of bridges in poor condition at a census tract level, is something that we have. And yeah, that’s I think what sets the vulnerability bucket apart from the SVI.
Lisa Lines: Thanks. We have one other category to talk about. Aditya, tell us about climate risk. What have we got now for climate risk?
Aditya Vadalkar: Talking about climate risk, we thought, what are the different factors that go and involve a community or impact the community the most? That would be the natural disasters that we face or the toxins or pollutants that are released from the everyday things that happen around the community.
We went ahead and got data from OpenFEMA regarding the expected annual loss in dollars regarding a particular natural disaster. We got that data for 18 different hazard types, ranging from avalanche, coastal flooding to earthquakes and heat waves and so on. All of these are talking about expected annual loss, that is, amount in dollars, which signifies which quantifies the loss for a relevant consequence type.
And when we talk about loss, we mean the amount of loss in dollars that a community might be impacted by. When we also have variables like the gross median income or the financial situation of a community, we can sort of understand how much a particular community will be affected when we see the annual loss, as well as how financially settled that committee is.
And apart from that, we also collected data from EJScreen regarding toxins, pollutants, data like particulate matter or toxic releases into the air, to get a better understanding and a better information for the model, to get a clear understanding of what all environmental impacts it is facing.
Lisa Lines: Usually with RTI Rarity, we are usually using an outcome of some sort and then predicting it with machine learning. Why aren’t we doing that here? Why are we doing it in a different way than what we normally do?
Sourabh Deshmukh: Yeah, I think over here, potentially someone who’s going to look at what we’re trying to do is EPA, who are largely not interested in machine learning approaches [inaudible 00:11:00]. And over that, just keeping it simple is what we are trying to go for.
What a linear combination would do for us is make it extremely interpretable and highly explainable, instead of a complicated machine learning based approach to generate a risk score. I think that’s why we are going for a simple linear equation, which represents all of these three buckets and their interactions to generate local climate impact.
Lisa Lines: And also, we know that, in fact, while we have some expected annual loss information, the real death toll from climate change is currently invisible.
The official statistics actually say only 200 people died from heat in 2023, but we actually know that it was probably more like over 2000. It’s just not coming into the data in that magnitude. So we’re dealing with an absence of an outcome, in a sense.
The idea is here to try to do a better job of understanding the potential impact of climate change events. And then we have this whole interesting idea of simulation. I think maybe it’s now time to bring up the dashboard. Aditya, are you ready to share?
Aditya Vadalkar: Yup.
Lisa Lines: Now, while he’s loading that, I’ll tell you we did the pilot project in two states, California and Florida. So, hopefully, this is coming through for you all and you can see the local climate impact score is currently being shown on the screen.
And here, we’re looking at the Bay Area and you can see around San Francisco, mostly pretty low risk, although there are pockets of higher potential impact. And if we actually break it down, actually, if you look at the risk independent of the…
So we’re going to just show you risk now. And you can see, here is showing the higher-risk areas in the South Bay around there, and in Monterey Bay as well, also inland. That’s possibly a lot of the heat index and flood events, things like that.
And so, we have a way to actually do some simulation to show the effects of a 10 percentage point increase in risk, or actually we can do vulnerability increases if we think that communities may become more vulnerable in the future, there’s an opportunity to look at that.
And then same with resilience. What happens if we invest in climate adaptive strategies, like in protecting flood buffer zones, wetlands, and things like that? Or investing in training more citizens in the citizen disaster core or things like that? And it’s interesting too, to think about the way that risk may, in fact, go completely off the charts as we think about sea level rise.
They’re projecting one foot of sea level rise in the next 10, 20 years. And so, if that happens, we can actually look at which communities will be affected by sea level rise, as well. We can zoom in and out. We can click on a place and see its score.
And I want to just note, on the left-hand side of the screen, the risk score range. Let me show that a little bit. Yeah. So this is actually just showing you that the distributions of these different sub scores, they’re all on a zero-one scale, but they don’t all max out at one. So, as you can see, as we’re looking at combining these three things, we’re actually taking the risk plus the vulnerability and then subtracting the resilience. So it’s a very straightforward combination of these three sub scores, which is great.
And then let’s go back and look at the Florida map. We can actually select Florida and see that load up. And same thing there with sea level rise. It’s a real concern. Actually, Aditya, do you have the sea level rise in a different window?
Aditya Vadalkar: Yes.
Lisa Lines: Great. Yeah, okay. Right. Yes. So this is 2020, 2040, 2050. And as you can see, we’re using flood gauge, tide gauge indicators out in the ocean to model out what sea level rise is actually going to look like. And we’ll be adding that to the map in a later iteration, which is very exciting.
Okay. So let’s talk some more about the score itself. If a community’s at higher risk, that could be a number of different things. That could be they’re high on vulnerability or low on resilience, or both. Even a medium-risk community on the climate side of things, the impact can be multiplied because of high vulnerability and low resilience. And so, that’s where we want to try to encourage communities to invest in mitigating some of those vulnerable situations and the vulnerability of communities to the impact to climate change.
One of our previous podcasts, we actually talked about Florida quite a bit, went through a whole discussion of the northern part of Florida along the I-10 corridor. That whole area typically has worse health outcomes than southern Florida, and the vulnerability to climate change is much higher in certain of those communities as well.
Okay. There’s our Florida map. And yeah, they’re high LCI, meaning high climate impact along the coast, all through northern Florida there, and in the center. Not so much on the [inaudible 00:17:18] beach communities. Often those are wealthier communities, but as you can see, there are pockets of high LCI all along the coast as well.
And we can imagine, as we look at some of these overlaps, you can see what’s driving that LCI score is high risk and maybe some lower resilience areas as well.
Other thoughts and comments on what we are seeing here?
Aditya Vadalkar: A way of looking at this map is we have the risk vulnerability and resilience in different shades of red, blue, and green. One interesting thing is, when these colors overlap and blend, you can see how each community has a blend of the resilience, risk and vulnerability.
For example, the bluer ones are on a side of more vulnerability as compared to complete green ones that have very high resilience. Or at the same time, you can see how each of the scores interacts with each other and combines to form the LCI. So when you finally view the LCI, you see it as a final score.
Lisa Lines: And this is hot off the presses, right? I mean, we have been playing with the color scheme quite a bit, trying to understand how to show visually this rather complex, three-dimensional index. And audience, you’ll have to tell us have we succeeded in our goal of making it interpretable and easy to understand. That’s the goal here, and to be completely transparent about what goes into the model.
We’ve got 55-plus variables in the [resilience] domain and 18 different measures of climate risk. It’s a, we hope, a good improvement on the national risk index to be able to see not only the impact now, but also the potential impact in the future, as we can do some simulation modeling.
Let’s talk some more about what we’re seeing here on the map. What goes into that resilience bucket besides the BRIC? We have some other things going in there, positively scaled things, right? Baha?
Bahamin Ayla Akhtari: Yes. One of the important factors in the resilience bucket are health outcomes, healthcare, and infrastructure, such as walkability index.
When we’re looking at resilience, it’s important to also notice that each community will have different resilience variables and measures, so whenever you take a look at a community’s resilience and vulnerability, you can see how a community would need to allocate resources in order to improve the resilience and mitigate vulnerability.
I do want to take a look at Florida right now. If we take a look at the eastern coast, we can see that, although it does have higher resilience, it will also have a pretty high LCI. Although a community is pretty resilient, you also need to look at factors like vulnerability in order to lower the LCI score.
Lisa Lines: Yeah. Other thoughts, comments?
Sourabh Deshmukh: One thing which I’d like to say is, separating these three buckets out into three sub scores essentially gives us an idea that if we work on either of these three buckets, we can make the LCI look better.
So, for example, if we talk about the resilience bucket, if we work on the factors that go into that, we can literally counteract the effects of the factors which we have counted into the vulnerability bucket. So that’s something which is very interesting over here.
Lisa Lines: Definitely, and hopefully food for thought for communities at higher risk. There are all kinds of strategies for lowering a community’s vulnerability and increasing resilience.
Aditya Vadalkar: I would like to add one thing.
Lisa Lines: Yeah.
Aditya Vadalkar: One thing that I just showed right now on the dashboard was having a higher resilience score. If the resilience score was increased by 20%, you could see that the whole map of Florida was showing a color of light brown, which was low climate impact. And then I edited it back to the current scores that we have, which is the baseline. A lot of the areas, you can see, are in high local climate impact.
So it just goes to support what Sourabh said, that if we are able to work on these modifiable variables that we [inaudible 00:22:46], then we can see how the climate impact will reduce on each community.
Lisa Lines: For sure. Our call to action would be for communities to organize, hopefully, and tell their legislators and their policymakers, “We’re at high risk for big impact from climate change and we need to increase our resilience, we need to reduce our vulnerability.” And that’s how we can prepare.
We have to get ready. I mean this is definitely happening. It’s happening now. I mean we’re seeing all kinds of big impacts across the globe, and our country is not immune. We’ve got to take our heads out of the sand and put our energies together and organize to tackle this challenge.
It’s a huge problem for public health, beyond the fact that we cannot have a healthy economy without a healthy population. And as climate change continues to affect the population with heat stroke and hurricanes and floods and all kinds of natural disasters like that, they’re all increasing in severity. And so, from my perspective, this is critical that we engage with our communities and pull together to really tackle this issue.
Hopefully, we can publish this work and you’ll be able to, if you live in one of these two states, check out your community’s risk, vulnerability, and resilience and overall LCI, local climate impact, score.
I want to hear from all three of you about, any final things that you’ve taken away from this project, lessons learned?
Bahamin Ayla Akhtari: I just wanted to say, since you were speaking about community, community approach is crucial, especially when it comes to natural hazards, because not only does it encourage collective action, we are able to allocate resources and knowledge, support systems. Whereas an individualistic approach can leave people even more vulnerable and isolated.
I want to just talk about, quickly, where I’m from. I’m from Houston, Texas. And as a lot of people know, we’ve been through a lot of storms and hurricanes and we are a pretty vulnerable city. Our infrastructure is pretty old, so whenever a hurricane comes through, a lot of people are displaced or out of power for a while. But I will say, speaking on this past hurricane that we had, hurricane Beryl, our community was able to push through, through community resilience. And that’s an important factor that I think a lot of places should consider.
Lisa Lines: Absolutely. Any other lessons learned for you from this project or from your time here so far?
Bahamin Ayla Akhtari: So far, honestly, I think one of the biggest lessons I’ve learned from working on this project is how, even if you have resilience, if you don’t work on a community’s vulnerability, you will still have significant damage to a community.
I always thought that if you were pretty resilient, that you could push through. But no, it’s also very important to look at vulnerability as well. And climate risk is… It’s a ticking clock.
Lisa Lines: Yeah, for sure.
Sourabh Deshmukh: Definitely. One thing I’d like to say over here is, as Lisa said, the call to action. I feel what you saw right now is future generations are going to bear the fruits of that.
So I think if you start working on climate-related issues right now, that’s when our future generations are going to get a better form of earth. So I think it’s our responsibility to just get started and reach as many people as we can. And I think this initiative does exactly that.
Lisa Lines: Thank you.
Aditya Vadalkar: I would like to add on to all of these things that these guys said and… The diagram that we saw regarding the relative sea level and the rise in sea level across each decade, that also shows us how much each of the variables in our risk bucket will be affected because most of them are the expected annual loss from a particular natural disaster. And one of the main major things contributing to that or signifying that is sea level rise.
So, getting a collectivistic control over how we are impacting climate through these pollutants, toxins, or understanding why these natural disasters are occurring so often now is a great way to get started on trying to save this planet.
Lisa Lines: Yeah. The planet will probably be okay, it’s our human culture that I’m worried about. Yeah.
I guess with that, we can wrap this up for this month. That’s our episode. I really want to thank our special guests for joining me on this podcast today.
I’m so incredibly happy to be working with all three of you. I feel so lucky to know you all and to be able to work together on these important projects. And I just want to thank you so much for all your hard work. This was a huge, huge, massive pilot project. We got a lot done and I’m really happy with how it’s turned out.
So, audience, I hope that you will stay tuned because there’s much more to come. And thanks once again for listening to the Healthy Intersections podcast, sponsored by the American Public Health Association’s Medical Care section.
And we’re hosted on The Medical Care Blog. If you go to the blog, you can see the transcript of this episode, you can see some links to some of the work that we’ve been doing, and you can leave your comments right there on the blog post. So, looking forward to hearing from you, audience. And with that, wish you a happy rest of the summer. Bye-bye.
Lisa M. Lines