How to 3D Scan Dublin City: Professor at New York University, Debra Laefer - members only

Engineering innovation has created incredible technology and found new ways to use current technology to overcome engineering challenges.

LiDAR is an example of one of these revolutionary tools and today we hear from an art historian turned civil engineer who is using it to scan our cities. We hear about their creation of the world's densest urban aerial laser scanning dataset, which was conducted using a large slice of the centre of Dublin City, and the challenges they’ve overcome in transforming how we understand, plan, and protect our cities.

Our guest is a pioneering force in urban data science and has authored over 160 peer reviewed publications, been awarded four patents and worked as a professor in UCD Dublin. She is Professor at New York University's Centre for Urban science and Progress Dr Debra Laefer.

THINGS WE SPOKE ABOUT

Using LIDAR technology to create highly detailed 3D scans of cities
Developing methods to efficiently store, process, and analyse LIDAR data
How 3D scans are revolutionising urban flood modelling and emergency response
Applying the LIDAR data and 3D models to real-world engineering challenges
Exploring the use of 3D printing technology in conjunction with LIDAR data

GUEST DETAILS
With degrees from the University of Illinois Urbana-Champaign (MS, Ph.D.), NYU (MEng), and Columbia University (BS, BA), Prof. Debra Laefer has a wide-ranging background spanning from geotechnical and structural engineering to art history and historic preservation.

In her decade and a half as a faculty member in both the US and Europe, Prof. Laefer has served as the principal investigator for grants from a wide range of sponsors including the National Science Foundation, the US Federal Highway Administration, the National Endowment for the Arts, the National Endowment for the Humanities, Science Foundation Ireland, and the European Research Council (including a €1.5 million single investigator award from the flagship ERC program for which she is the only civil engineer to have been funded in Ireland in the program’s 11 year history).

Prof. Laefer has authored over 160 peer-reviewed publications, been awarded 4 patents, and has supervised 15 doctoral and 20 Masters theses. Among many honors from IEEE, ISPRS, and other professional societies, the most notable is perhaps the 2016 commissioning and hanging of her portrait by the Royal Irish Academy as one of eight researchers selected for the Women on Walls project to celebrate Irish women in science and engineering.

https://www.linkedin.com/in/debra-f-laefer-09510a11/

MORE INFORMATION
Looking for ways to explore or advance a career in the field of engineering? Visit Engineers Ireland to learn more about the many programs and resources on offer. https://www.engineersireland.ie/

Engineers Journal AMPLIFIED is produced by DustPod.io for Engineers Ireland.

QUOTES
"We didn't devise a sensor, we didn't even improve the sensor, but we took a fundamentally engineering approach to it. We took this more systematic approach of; let's reverse engineer the process, figure out what we want to get and figure out how to use the sensor to obtain that data." - Dr. Debra Laefer

"This technique has been used in fields as far from civil engineering as breast cancer research. So that not only has it been transferred to other LIDAR applications, but people have used it for other remote sensing and medical imaging datasets." - Dr. Debra Laefer

"It's good to make mistakes, and it's good to have senior engineers check them." - Dr. Debra Laefer

"As crazy as your idea may seem, a lot of the time the best ideas are initially too far ahead of the curve, so don't give up on them." - Dr. Debra Laefer

KEYWORDS
#buildings #data #dublin #engineers #civilengineering #lidar #computationalmodel

TRANSCRIPTION
For your convenience, we include an automated AI transcription

Dusty Rhodes 00:00
Right now on AMPLIFIED, we're about to find out how to make a 3D scan of Dublin.

Debra Laefer 00:05
If people think they have a good idea, they shouldn't give up, that if you push on it hard enough long enough, it will happen. As crazy as your idea may seem, and a lot of times the best ideas are initially too far ahead of the curve. But don't give up on it.

Dusty Rhodes 00:24
Hi there. My name is Dusty Rhodes and welcome to AMPLIFIED the Engineers Journal podcast. We're very familiar with the BIM and LiDAR in civil engineering but how far can you go with those technologies? Could you apply them to a full city and still get millimetre level accuracy. Our guest today is behind the world's densest urban aerial laser scanning dataset, which was conducted using a large slice of the centre of Dublin City. She is a pioneering force in urban data science, addressing the challenge of handling massive amounts of information collected by drones, satellites and laser measurements, and then using smart ways to store search and turn that data into useful visuals. She is a qualified civil engineer, has authored over 160 peer reviewed publications, been awarded four patents, worked as a professor in UCD Dublin, and is currently serving as a professor at New York University's Centre for Urban science and Progress. I am thrilled to welcome Dr. Debra Laefer to the podcast. How are you, Debra?

Debra Laefer 01:24
Great, thanks so much for having me.

Dusty Rhodes 01:26
It's a delight to have you with us. Before we get into the world's densest LiDAR data set, which you generated here in Ireland, you had a very interesting route into engineering, you had a kind of an art history degree and then got into civil engineering. What, tell me the story behind that.

Debra Laefer 01:43
So I fell in love with painting and old buildings and decided I wanted to become an art historian. So I applied to the best program in the United States and got admitted to Columbia University. And as part of this, I gotten involved with creating a student art gallery. So this was supposed to be a place for students and faculty and alumni could show their own artworks. And then we were informed that we had to temporarily move out of space, because they were going to do construction through it to put in some new telecom lines. So this was back in the mid late to late 80s. And I was concerned because it was a historic building. So I started asking around, started doing some investigation, and found out that not only was the district building, but it had been damaged. In fact, it had been damaged the last time they did construction near it, so called the Landmarks Preservation people to confirm that yes, the building was protected. We reached out to a people at the historic preservation program at Columbia University. And they put me in touch with one of their students who is a civil engineer. And I was so impressed with this, this young woman, her name is Marie Ennis, she's actually still a practising engineer here in New York City. And that she could combine this large toolbox of thoughts and knowledge and conveyed in a way that was meaningful to people in practice. And I was very, I think, influenced by that. And over the next few months, I started thinking, well, if I really love old buildings, maybe this is what I want to do, maybe I need to be come a civil engineer. So there I was finishing my last year in my Bachelors of Arts degree in art history. And starting my first year in my Bachelors of Engineering and civil engineering. And ultimately, I persevered with this, I been worked in the construction industry, at a time where it was it was pretty rough, a lot of organised crime, a lot of violence on the sides. It was a pretty exciting time in New York construction. But I really was happy doing that. And I thought, Oh, well, you know, I feel like I still don't know enough. So we're going to start a master's program at night part time. And as part of that, I met some amazing people in the geotechnical engineering realm. So that's a division of civil engineering. And a lot of them surprisingly, had PhDs. And they were in the midst of really important a lot of amazing technologies from Europe, into the United States, things that were very well suited to protect existing structures when you did excavation or drilling or blasting or de watering, or tunnelling near them. So I finished my master's degree I applied and got a Fulbright to Italy, I spent a year at the Polytechnic of Milan, really studying brick masonry and its vulnerabilities and then I came back to the US, and I took my PhD and Geotechnics. But I had also the opportunity to do some travelling as part of that research. And we went to Korea, and I got a chance to come over to the UK, and to spend some time, particularly some people from McDonald, looking at the Jubilee line. So at the time, this was the most expensive tunnelling project that had ever happened, it was about, I think, 2 billion pounds. And about 25% of this was being spent either on predicting which buildings were going to move, monitoring them, or, you know, kind of free tunnelling intervention where they were pumping grout under the ground, in particular, under Big Bend. And despite this huge investment, a lot of buildings did get damaged. So when we spoke to the engineers, who simply what kind of great computational Shanell models are you using to predict which buildings are going to get damaged, and they said, Oh, we don't use the computational models at home, what's wrong with them? And they said, No, because there's nothing wrong with the model. But we do not have documentation of all of the above ground buildings. Many of these buildings date back hundreds and hundreds of years, and to go out and to survey each building, and then convert to generate drawings. And then to convert that into a computational model would be impossible for the hundreds and hundreds of buildings that are along this tunnel route. So instead, we're using a fairly simplistic set of numerical equations that date back mostly to the 50s, but then kind of improved in the early 70s. So here we are, we're pushing the turn of the millennium. And we're using stuff that's at least 30, if not 40 years old.

Dusty Rhodes 06:53
So that I understand it, you find yourself in London, and you have all of this tunnelling going on, they can't correctly tunnel because they can't do the computations for all of the buildings because it's just too big an area.

Debra Laefer 07:07
It's not that they can't tunnel if they can tunnel but it at greater risk to the structures than need be.

Dusty Rhodes 07:14
So, you said to yourself, aha, here's a problem. I'm gonna come up with a solution for this.

Debra Laefer 07:20
Not quite, I just went home and thought, Wow, I'm surprised this is a problem. Yeah. So I went, I bet back, I finished my degree. And about two, three years later, I had just finished, I had moved to North Carolina to become a young faculty member. And 911 happened. And having spent many years in New York studying and working, having Mitch family there. My parents were born there. My grandparents were born there, many people from our families still live there. It was a very disturbing and moving kind of time. And I was very interested in what they were doing, how they were trying to do the rescuing, because I'd been involved with some kind of post disaster earthquake work while I was also at the university. So it was kind of a little tied into the emergency management community at that point. And I started to learn about a fellow named Dave Bloomquist, who is a faculty member down at the University of Florida in Gainesville. And his work with NOAA, and the work that they did to basically put up a small plane, and to do heat detection and LIDAR over the World Trade Centre disaster zone, so that it could help them both figure out where there might be fires happening underground still, and how to start to remove debris at that point, they already realised that there are no survivors. So but it started looking at these 3d models, or 3d representations using this LIDAR data. And I thought, wow, that's really interesting. And about four months later, I was up in New York, and I had an opportunity to work to get to know an engineer who was really helping coordinate a lot of that removal, and had the opportunity to actually go down into the site. So this is like January 2 2002. And the site is still on fire. Even with a gas mask, it was very hard to go through. But I'm looking around and I'm seeing you know, these buildings on the damage and thinking about the work that Bloomquist did, and I said, No, no, maybe we could use LIDAR to document all these structures. So I called him up and he was very generous and he helped share some of experience and help get my group started and we started doing some work in this area. And we started doing some work for the owner Emergency Management Agency. Looking at prediction have trees falling across roadways, where we would go and MIT from the LIDAR they already had, we could measure the height of the tree and the distance to the road and make estimates to what extent if the tree fell over, it would either partially, completely or not at all block the road. So that was kind of our first foray into that. And once I started, I was completely locked.

Dusty Rhodes 10:25
Okay, so now, it sounds like you are looking for data over huge areas of land and very highly populated land, with a lot of buildings in it. That's a huge amount of information that you need LIDAR piqued your attention. For engineers who are not working in this space. Can you explain how that technology works? How do you 3d scan an area?

Debra Laefer 10:48
Yeah, so it's a technology that can be used from multiple platforms from even right now through your iPhone, or from some type of stationary unit, the unit can be mounted on a car, it could be a small lens mounted on drones, on helicopters on airplanes, the technology is fundamentally the same, you're sending out a laser signal, kind of a beam of light, you know, what time it left your piece of equipment.

Dusty Rhodes 11:27
And you know, where kind of in the world your equipment is, is it on the ground? Is it above is it at the bottom of the broad of a craft.

Debra Laefer 11:29
So, that beam part of it will come back, it will hit something and it will come back. And you will know and the equipment will record the time that it comes back. So based on the change in time, we have a certain distance that can be calculated because we know what the speed of light is. And we use that to determine what they call the range.

Dusty Rhodes 11:52
And you were doing this millions and millions. And it's I imagine millions and millions of times a second.

Debra Laefer 11:58
Yeah, I mean, ultimately, obviously some of its limited by your equipment, but it's actually more limited by the how much data the equipment can take back. And how long your battery is good for.

Dusty Rhodes 12:15
Let's put a picture on it. Okay, you somehow found yourself in Dublin and you decided Grafton Street. Okay, we're going to 3d scan that we're gonna measure that down to what kind of measurement Did you get it down to what scale?

Debra Laefer 12:28
The first scan we did was about four centimetres. Wow. Okay. And the second I think was down to about two and a half centimetres. Wow. So tell me maybe a actually even less than that. So made that a centimetre.

Dusty Rhodes 12:47
Tell me about this story about how you use helicopters, drones, whatever centre of Dublin Grafton Street, the whole block and you measured it and 3d scanner to within a centimetre.

Debra Laefer 12:57
So what we really wanted to do was to provide these representations of these buildings to the engineering community. So you have to set your mind back to 2004. Celtic Tiger, Ireland's booming, and a lot of discussion about putting in Dublin's first metro to go from the bottom of Grafton Street, or more specifically, in the northwest corner of St. Stephen's Green, yeah, up to the airport. So Dublin, and Ireland, in general, at that point had had almost no tunnelling. And obviously, here we are in a country that has limited experience with this technology, you have a very complicated geology with a lot of small, granular material mixed in with kind of big boulders and stuff. So it's a tough thing to tunnel through without a lot of disturbance of the ground. And here you have this amazing architectural resource in terms of the centre of Dublin, that at that moment was actually under consideration as a World Heritage Site. So you have this kind of conflict happening about preserving and the future and at risk. So I was fortunate put together a proposal to science foundation Ireland, with a colleague at UC Dublin, Hamish Carr and a colleague up at our collaborator up at Trinity, we were able to come up with kind of a plan of not only how to acquire this data, but how to process it and make it usable.

Dusty Rhodes 14:39
Okay, tell me about acquiring it.

Debra Laefer 14:40
So, when most people even today, put this kind of unit under a plane or a drone or helicopter, it faces down so the unit swings, and depending on the equipment, it might swing thing, just left to right. Or it may have kind of an arc to it. But it's kind of, you know, it's not just capturing exactly what's below it, but kind of a swath, but it's pretty much focused on what's directly underneath. And as the LIDAR unit swings to the side, the quality and quantity of data that you get, when it intersects a building facade is pretty limited. So most of the good data that you're getting is roads, and roofs. But if the thing that you're interested in knowing about and protecting is the building's facade and its structure, knowing about its roof and knowing about the street next to it's not going to help very much. So we kind of took a big step back, and Hamish and I really like, Well, how do we capture these building facades? And we said, well, let's let's think about the equipment, how does the equipment work? And how do they traditionally fly? Even though old kind of medieval city, like Dublin has a kind of pattern to it. And much of it's a grid. So typically, what they do is they say, Okay, we're gonna fly from x to y, and from A to B, this is our kind of area, and they will fly along the grid line, they'll go down, turn around, back down, back. And then when they're finished with that, they'll come around 90 degrees and do it the other way. It's great for the pilots, they really get lost, it's not so great for the data acquisition for the con, we watched you. So we show through geometry that if you flew diagonal, to the street grid, that you could pick up significant like basically double the information, just because the angle, just because the angle, the other thing we realised is that the amount of overlap that they fly was only enough to basically sew together, you know, one group of data from the next. So when you're flying down one street, they would go over, they wouldn't necessarily do the next street, that they would position themselves so that there was only about a 10% overlap. So if you're from the geomatics community, and you're interested in mapping, and you're interested in floodplain, or using this for floodplain risk analysis, this is great. But if you're interested in looking at these facades, it's not so great. And it basically really limits what you can pick up, because it's in that swing at the edge of the scan, that we're picking up the facades, right, because we're looking down, we're looking at the street, and now we're swinging to the left. And only at the end of that swing, do we start picking up the data. So again, we went back to basic geometry, and established that we needed about a 60% overlap, to achieve a complete scan so that we didn't have these is good blank spots, because you have with this line of sight technology, if you can't see it, you can't capture it, like the camera. So you have a situation if you're in front of one, if you're standing in front of a building, you obviously can't see what's on the back. But also, sometimes buildings preclude you seeing a building behind them. So if you're up in the air, and you've got a tallest building, and maybe there's a small one across the street, maybe you can't see that. So again, we had to kind of compensate for a lot of these things. When we originally did this, people thought we were insane. They're like, why do you want to do this, you know, like, trust us trust us. And it was very hard to even find a contractor to do it. And when we got the data, they were astonished. They're like, Wow, we had no idea we could get this kind of data.

Dusty Rhodes 19:08
Let me just say that there is a YouTube video of the data that you got. And when you watch the video, your jaw will drop and go, Oh, my God. And I have put a link directly to that video in the show notes in the description area of this podcast that we're listening to right now. So you can just click on it. And you can see it. Apologies, Deborah go on.

Debra Laefer 19:28
No, thank you. I think that's one of the best demonstrations of it. Because we're taking we didn't devise a sensor. We didn't even improve the sensor. But we took a fundamentally engineering approach to it. So I think that the way the technology had been used this idea was like more data is better and you just get we can you smash it together and you kind of muddle through the best you can. And we took this more systematic approach of let's reverse engineer the process. fear what we want to get and figure out how to use the sensor to obtain that data.

Dusty Rhodes 20:06
So now you have the data, what the problem is, is that you have an enormous amount of data. And that's the next problem. The next challenge, what do you do with it? I mean, how do you sort? Those many ones and zeros? Yeah.

Debra Laefer 20:22
So the basic storage of it, the I would call the static store, it's just, you know, putting it somewhere is not so much a problem. My brother used to work for Google, and he would joke, you know, what's a petabyte between friends? So it's not the storage, it's, as you said, it's the sorting. It's the what they call the queering. It's the retrieving of the data. And so we really, with my help, my long term colleague and collaborator Michela Berta loto, at UCD, really sat down and looked at a lot of the work that she had done in database and database structures, and talked about, well, what were our needs? How is the community currently doing at least some of this work? And what was that opportunity. And so in about 2006, we really started in earnest, taking on that problem. And I would say that that work really culminated about nine years later, when we graduated, jointly drew graduated a PhD student who demonstrated that you could very effectively use the data structure as the fundamental building block for post processing algorithms. So that you already have stored the data in a way that is highly usable. The paper that on jeuveau are joins graduate student who is still in Ireland is the lead author on is in the top point, zero 1% of all papers cited. For the years published, this technique has been used in fields as far from civil engineering as breast cancer research, so that not only has it been transferred to other LIDAR applications, but people have used it for other remote sensing and medical imaging datasets. So which is really amazing.

Dusty Rhodes 22:24
So you put some banners on the data, then how do you integrate it with other technologies? So a lot of people talk about GIS and bi M and stuff like that? How do you get that data then interacting with them? So engineers can actually use it?

Debra Laefer 22:37
Yeah, I'd say actually, most of the ways that us engineers use it reaction, computational models. So certainly, there are ways to tie it to GIS systems, the time we were working, there wasn't even a full 3d solution, which meant that it was what they call two and a half d solution, which means that every Z point, every elevation point, there could only be one unique one for every xy point. So if you had a building that was truly straight, there was no way at that x, y point, you know, at that corner of your street, or the corner of your building, to represent both the bottom and the top. So you had this kind of slightly wedding cake effect, where the points were actually slightly offset. Obviously, the technology has moved on from now, these GIS systems can both produce and host 3d models. But to just to give you a sense of kind of where we were, you know, with us BIM kind of really wasn't even really a thing by then. And the challenge is, is that unlike a photograph, when you have a photograph, and you look at it, every pixel, every little space is filled, right? There's no blank spots. With the LIDAR data. It's not that way. Maybe the beam went through a window, and it didn't come back or you know, maybe it went through a tree and you it came back in like six different pieces. So you get this data set that's very non homogeneous. These often refer to a sparse, it's an ordered. So there's not no, there's no natural order, when you get it back from the vendor. mean, it's been geo referenced. But there's, it doesn't like say, Oh, this point belongs to a building. And this other point belongs to a building.

Dusty Rhodes 24:30
It's a bit scattered.

Debra Laefer 24:31
It's a bit scattered. It's a bit chaotic. So So we continue to pioneer really groundbreaking work in how to fundamentally store that data. Because very early on, it became clear that the sheer size of the data, it was a major impediment to people using it. And it's still content used to be so some of my most recent work that we've not published yet, really looks at how do you take a billion points and process them actually Just on a regular PC, you know, can you do that? So, for us, the bigger question was, how do you get these points into a computational model? We don't care, we're labelling them necessarily, may eventually want to label them to a certain extent, because you want to know if your material model for each piece is correct. But the bigger thing is, how do you generate what they call a watertight model. And this is where the 3d printing starts to sneak in. So to do a computational model, you have to do create what they call a watertight mesh, which means that every point has to be connected to other points through a set of lines, but these lines must connect at nodes, they can't overlap, they can't be a little short of them, right. And when you use the traditional transfer transformation processes that were available for ticket in the late 2000s, you ended up having to do a huge amount of manual correction. And if you're looking at a terabyte of data, that's not gonna work, right. Yeah. So we really had to kind of think about how do you overcome those problems. And through this wonderful kind of collaboration between computer science and through civil engineering that we had going on at UCD, between myself and Hamish Carr and our students, Tommy Hanks, and Lynch ronghong, we had this kind of Eureka moment that the way the computational models were set up, many of them used would look like almost a little bricks, that they were these eight noted elements, and that these eight nodes had no elements, which didn't have to be squares, that could be rectangles looked an awful like, the elements that we were using to do the storage axis that this is acting like a key is that, yeah,

Debra Laefer 27:02
in this, so we divided the data in something called an octree. So we're you chop it up into basically eight quadrants. And if there's no data in the quadrant, you forget about that part. And then you keep kind of digging down either until you only have a certain amount of points in a box. Maybe that's the amount of points that can be stored in the computer's cache and dealt with comfortably. Or maybe you do it more generically. And say you're just going to do you know, five divisions of these things. We realise that the octree and the computation model in the finite element, they looked a lot like so then we came up with two really pioneering algorithms to do that transformation. So as we were thinking about watertight, so what is starting to happen, the patents for the original 3d printers are expiring, and we're starting to see this boom of three. So actually, this is about 2012 2013. Starting to see this boom of home 3d printers, or the you know, low low in 3d printers. There's 3d printers this 3d printed that people are even talking about maybe can you 3d print a house, you know, all these things that people are now really excited the same way? Everybody's talking about AI now. So if you cast yourself back to 2013, everybody started with 3d printing through their predict 3d printed clothing and hats, and, and, and, and everything. How do we start saying, Well, if we're creating this watertight model, couldn't we use that same watertight model approach for 3d printing, because that's what you need. The input files for 3d printing have to be these watertight models. So there was an opportunity to apply for a competitive commercialisation type grant through the EU. At that point, I had received the European Research Council Award, which was the single largest single PI award that you could obtain at that time. And they had a program they wanted to really try to commercialise work. So you could then say, Okay, this piece of work came from his project. And we'd now like to try to commercialise it. So went through that competitive program, and we got a good amount of money. And we said, Okay, we have this wonderful used 3d printer, commercial grade metal 3d printer that we were able to acquire. And it quickly became apparent that this is a very expensive thing to run. It requires a huge amount of knowledge that you have to keep in the group. And we said, how are we going to sustain this? So we said, well, there's no 3d printing centres in Ireland. What if we just opened one, and that's that so that was our next big adventure. And we started to acquire other funding and other equipment and we really, you know, kind of graduated a whole class of people who then went on many of them to really lead the introduction of what are called Advanced Manufacturing in Ireland, including a guy named Brian Marin, who came to us off the dole through a, you know, train to work program. And it was so successful, that Brian became the main initial technician for the first National Advanced Manufacturing Centre. So I think, a real success.

Dusty Rhodes 30:31
Let me put this into some kind of a context, then on the engineering and design side of things because of our use of BIM. And we're used to digital twins, and you're able to play around the buildings and change things and see how it looks. How can you do that? Like, can you use this technology that you are working on to do that on a city wide level?

Debra Laefer 30:50
Yeah. So obviously, acquiring the data takes a while processing the data. So it's not something that you just go out and do every day. But we do see that communities, municipalities, even states are doing this now on a pretty regular basis, if not once a year, once every two years, in fact, the United States, we are having the completion of our first national scam, which is pretty exceptional. The difficulty of processing and storing that data is in part related to the quantity of the data. If you want really good sub centimetre data, it's gonna be a lot. So that always has to be part of it. But we have certainly taken that data when we've, you know, generated middle little 3d models of parts of Dublin.

Dusty Rhodes 31:41
And where would an engineer be able to use that if he is looking as as a city planner? What kind of things would he be able to do with it?

Debra Laefer 31:48
So I think one of our very early visions when particularly we got our first data set back in 2008, we said status, so good, but it's not quite good enough. That the I think the aspiration was to have a data set that was so good, you could pick out the curb height. Wow. And I think that is what we really achieved in 2015, that the data was so good that we could determine bather, basically whether the edge of the sidewalk was handicapped accessible. So I think that's a very easy, accessible use case. We've now most recently moved, we just completed a project called Urban Ark, with UCD. And with work off under her up at Queen's University, Belfast, looking at urban flooding. And one of the key components to that was the detection of subsurface spaces, basements, parking garages, things like that. And although Lidar is a line of sight technology, we can get a pretty good understanding of some things, that there are the spaces and some extent the size of them. Based on if you're you know, the angle, you're collecting the data, you might see kind of the stairwell that many of our Georgian town-houses have, or even be able to capture some of the data through the windows, or the entrances to parking garages. And by incorporating that into a larger flooding model, we can determine more effectively where the risks really are, where's the water going? Are we over predicting, or these people are particularly at risk. And we've generated flood models that show that the subsurface spaces really have an impact of where the water's going, and how fast it's building up. So if you're trying to evacuate parts of the city, or deploy emergency services, you want to know where to do that you don't want to send people to the wrong places.

Dusty Rhodes 33:45
So listen, you've done Dublin City, I believe you're going for something slightly bigger for your next project.

Debra Laefer 33:51
We don't so much bigger, but maybe more technically advanced. So we've recently completed a one square kilometre area in south-west Brooklyn. And what's really special about that dataset is not only do you have this great LIDAR data, pretty much the equivalent but we didn't delve in a little bit denser. But we've coupled it with something called hyperspectral data, and hyperspectral data, ours is in the bottom of the shortwave range downward. So if you have materials that are known, and you can get the what they call the spectral signal from them, we can match that spectral signal with things in the built environment. So a computational model has two important components. One is the geometry. And that project through our lab work and those of others has largely been solved. But the assigning of those materials and those material properties has not and hyperspectral gives us that opportunity to start doing that.

Dusty Rhodes 35:02
So some exciting stuff happening, a lot going on. I love talking to you, because you're talking about things you did 20 years ago, that are almost like cutting edge. Now you have a type of brain that just thinks 2030 4050 years in advance. So I have to ask you, what do you consider now the main challenges that engineers today need to start thinking about?

Debra Laefer 35:25
I would say one of the main challenges is coupling these major weather systems or storm systems with that kind of urban level weather system, we have huge investments of, you know, trying to predict where hurricanes are going, and how much rain and how much storm surge. But what's really happening from the street level, say, up to the first 100 or two feet, there are not a lot of models. And yet all of that's controlling all these urban heat problems that we're having. And we just don't have those couple of models. So it's that multi-scale physics, that we're kind of missing right now, where people like me very much on the crowd at the bottom, could work with somebody like one of my collaborators, Olivia police here in our Chaos Group, which is our weather group in our maths department. Between Olivia and myself, there's a big space. And there's not too many people in that space. So I think that that's really where we need to start going.

Dusty Rhodes 36:32
I also wanted to ask you, because I was watching an interview with you, you were talking about the advice you give to your students, which I mean, it just rang true with me. It's when you're a student, make mistakes, because you learn more from mistakes. And when things go right. Now, that's great to say to students, we're all engineers listening to this podcast. Can you apply that to engineering in a real life professional situation? Or should you just have made audio mistakes in college?

Debra Laefer 37:02
One of the highlights of my educational career was getting to meet a geotechnical engineer named Ralph back. And Ralph peck at the time was the most important living geotechnical engineer in the world. And he had been a student of Karl Terzaghi, who is the founder of geotechnical engineering. And he would come at the age of, you know, 79. And he would give these talks about when he was a young engineer, working under Karl Terzaghi, and all the mistakes that he made. So it's good to make mistakes. And it's good to have senior engineers check them.

Dusty Rhodes 37:42
Keep trying new things, regardless. Yeah, kind of wrap up then by just ask, is there anything else that you'd like to add to our chat today that I haven't thought of? Or haven't brought up?

Debra Laefer 37:53
Yeah, I mean, I think that if people think they have a good idea, they shouldn't give up. That if you push on it hard enough, long enough, it will happen. As crazy as your idea may seem that a lot of times the best ideas are initially too far ahead of the curve. But don't give up on them.

Dusty Rhodes 38:15
If you'd like to find out more about Debra and some of the topics that we spoke about today, you'll find notes and link details in the description area of this podcast. But for now, Professor Debra Laefer from NYU Centre for Urban Science and Progress, thank you so much for an absolutely fascinating chat.

Debra Laefer 38:31
Lovely to be here. Thanks for the invitation.

Dusty Rhodes 38:35
And if you enjoyed our podcast today do share with a friend in the business just tell them to search for Engineers Ireland in their podcast player. The podcast is produced by dustpod.io for Engineers Ireland. For pre-released episodes, more information on engineering across Ireland or career development opportunities, there are libraries of information on our website at engineersireland.ie. But for now, until next time, from myself, Dusty Rhodes, thank you for listening.