In this episode we take a journey through the world of imagery, guided by Esri UK imagery expert Zoe Taylor. We explore some exciting developments within the imagery space, focusing on emerging technologies and how the GIS community is using the power of imagery.
Have any ideas or feedback? Get in touch.
Related Content
Discover ArcGIS Imagery Workflows documentation: authoritative resources to help you manage, analyse, and use your imagery and raster’s.
Get access to a library of tech sessions and demos from the 2021 Box Set, including some great videos covering imagery. You can binge watch, or dip in and come back for more later.
This course is for anyone who wants to learn - or learn more - about the many uses of imagery, the many types of imagery, and the rapidly growing integration of imagery with GIS. Sign up and join the waiting list.
Imagery paints a thousand features
Duration: 0:22:15
START AUDIO
Voiceover: The Spatial Jam, an Esri UK podcast.
Zoe: Using imagery can really help us understand transportation issues I think we are seeing across the world.
Sam: How are people actually using imagery data?
Zoe: We can really now combine the power of remote sensing and imagery analysis with the power of GIS.
Sam: Hello everyone, and welcome to another episode of The Spatial Jam podcast. Thanks for joining us this week as we talk about all things imagery. It doesn’t matter if you are an avid user of imagery data, or you’ve used it once upon a time for that long-lost university assignment. This episode will cater to all backgrounds to help us understand the role that imagery data plays in GIS.
My name is Sam Bark and today I'm joined by my co-host, Beth Adams, to unravel all things imagery, which to me and my very limited perspective is often a base map but actually, I understand that there is so much more to talk about.
Beth: Yes, and to help us out today we are joined by a special guest on the podcast, Zoe Taylor. Zoe works in the Professional Services team here at Esri UK. She has a diverse skill set that spans many, many topics and one of which, luckily for us, is imagery so thanks for joining us, Zoe.
Zoe: Thanks for having me.
Sam: It is actually really nice to have you on, Zoe, because it is a bit of a full-circle moment for me. I don’t know if you remember, but my first professional services onsite visit was shadowing you, looking at imagery data with a customer. It is quite nice to- just feels like a very full-circle moment to come back to talk about imagery with you on today’s podcast.
Zoe: I do remember and that was a scarily long time ago now.
Sam: It was, yes. We won’t mention how long ago. I want to kick off the episode by asking you, Zoe, why is imagery exciting? Obviously there are so many things about it, but what is it to you that is exciting?
Zoe: I think for me it is the last five years or so, imagery has become far more integrated into the platform. We can really now combine the power of remote sensing and imagery analysis with the power of GIS, and get way more out of it than we ever could before.
Sam: Yes, and there is so much changing with imagery. Beth, what have you observed from the recent years of imagery and what do you think is now changing?
Beth: I wouldn’t say it is necessarily recent years, but I was thinking about this the other day, that when I was a child you would have people come round trying to sell you aerial photos of your house, where they had flown over. People would buy them because you'd never seen the top of your house before.
These days, you can obviously go look on Google Earth and find it. You can go look on any imagery base map. But also, if you have a few thousand pounds to spare, who doesn’t these days? You can go and task a satellite to take pictures anywhere in the world, that is just mind blowing for me. I think it is just, yes, ridiculous how far we've come in regard to that.
Sam: I've never actually experienced that, but people used to knock on your door and sell you a picture of your house from a plane imagery, like a view?
Beth: Yes. Obviously I'm much older than you, Sam, so it was a long time ago. But yes, people used to just knock on the door, completely unannounced, and sell you photos of something you owned, so it is a strange one.
Sam: That is pretty cool, and then along came Google Earth and changed everything.
Beth: Exactly.
Sam: Pretty much. Zoe, is there anything from your perspective that you’ve noted, that is massively changing within imagery?
Zoe: Oh, it is huge. The volume of imagery has increased massively. Like Beth was saying, you can now task satellites. If, for example, a humanitarian disaster happens there are lots of satellite companies out there that can just go and task a satellite and check on the area of damage, and then we can assess that from there almost instantaneously.
We've got programs like Sentinel and Landsat, all of that data is available on the Living Atlas, so you can get that at the touch of a button. It is just massive. The volume is probably the one thing that has changed the most. Then, when you think about the ability to then combine that with GIS and our vector data, the possibilities are endless.
Sam: You say, obviously, there is more content out there and there is just more of everything. Is it easier to use and what kind of software do we have available to use all of this imagery data?
Zoe: Yes, it is loads, loads easier than it ever has been before. It is completely fully integrated into the ArcGIS system. We got Image Server if you want to use your imagery in an enterprise. We've got Image Analyst for ArcGIS Pro which is an extension that lets you do way more in-depth analysis, raster analysis, that kind of thing.
But probably the most exciting thing is the recent addition of ArcGIS Image and ArcGIS Online. Basically that means that anyone can upload imagery into ArcGIS Online. It is like a software as a service so it will host the imagery for you.
You don’t need to worry about complicated image management having big systems and lots of storage, and then you can do the analysis right there and there, and in the web which, for me, just brings imagery to everyone.
Sam: Is that the tool? I think I've seen that. It has some machine learning capabilities analysis with it. Is that the one that I'm thinking, right?
Zoe: Yes, so all of the imagery tools include some level of machine learning. There are a lot of deep learning packages out there, some of which you can access through ArcGIS Online and the rest through Image Server or even just Pro.
Sam: Yes, that is cool. Yes, I think that is definitely a trend in where things are going. But I think, yes, to pick up on that point, I've seen so many different things on ArcGIS Online now and apps that you can view different imagery data.
I know the Sentinel-2 data that comes in, imagery data, I think there was a volcano that happened, I don’t even know when, couple of years ago. You can see, because of the temporal resolution or the frequency of when the image is taken, you can see the before, during and after impact of that volcano eruption and the impact on the surrounding area.
Then, within ArcGIS Online run analysis on a spectrum of different parts of the imagery. I think that is so much further than when I was studying at uni [sic], that has come leaps and bounds from what you could do back then.
Interesting to pick up on some of the apps that have come out. Beth, I was going to ask you if you’ve seen any different apps that have come out to facilitate this use of imagery?
Beth: Yes, there are some really good ones out there actually. You can go and explore the Sentinel data and also the Landsat data through online apps. You can compare two different images from different times. You can also, just really easily, do analysis.
Things like burn analysis, you just click a few buttons and it does it for you. You don’t need to know how to set it up, what it is doing in the background. It just does all the analysis and then you can export that as a feature service out, so that you can see it in ArcGIS Online and use it elsewhere.
I'd say that is a really easy way for any of our listeners to go out there and just try it out. Look for Sentinel, or Landsat Explorer, or the Landsat Viewer as well. They are all really easy-to-use apps.
Sam: All of these, Landsat, Sentinel, they are all accessible for all Esri users, is that correct?
Beth: Yes, so even if you are not an Esri user, you can go and look at them to start with. You'll only be able to see the latest imagery. If you sign in with an ArcGIS Online account, you will then be able to put in different dates. If you wanted to see that before and after for the volcanoes in the Canary Islands, for example, right now, you could go and do that. Yes, there is much more functionality for Esri users but anyone can have a look at it.
Sam: Beyond the two that we've just mentioned, Zoe, do you know about any further content, imagery content, that people can get their hands on and where would they find that?
Zoe: Yes, there is absolutely loads. We've got partnerships with Maxar and Planet. They provide satellite imagery. Planet, for example, allow you to do that tasking of satellites if you need to. Then you’ve got other types of data which come under the imagery banner, stuff like elevation is available in the Living Atlas or even already derived products.
For example, Natural England have done exactly what Beth was talking about. They’ve identified burnt areas and cut areas in heather. They’ve actually published a layer to the Living Atlas for moorland change. You don’t necessarily have to look for the imagery, but also the derived products from the imagery can also be really useful in your analysis.
Sam: That is really interesting. I feel like you’ve been reading my notes because my next question to you, Zoe, was actually, how are people actually using imagery data? Because I feel like, between us, we've got 101 different examples of somewhere where we've seen imagery being used in a cool and fascinating way. How have you observed the usage of it within the GIS?
Zoe: Yes, so that moorland change one is a good example. I suppose environmental uses is a common one, so even things like identifying areas of possible river pollution. You can look at slope combined with bare earth and you can see areas where, if it rains, it is likely you are going to get a lot of run off and river pollution, which is particularly important for environmental organisations that are trying to prevent that.
One of the best ones I've seen recently is Esri have just released a deep-learning package to look at SAR imagery, so that is the Sentinel-1 satellite data. They are actually using that to automatically detect ships, which is really, really interesting for things like port management, rescue missions, looking at cargo transmissions around the world, so that is probably one of the neatest ones I've seen recently.
Sam: Yes. Beth, you were talking about the port in California this morning, weren’t you? They’ve got a 40-ship queue or something waiting to dock in, was it LA?
Beth: Yes, there are 2 ports in California that take most of the shipping in for America. Normally, you would get 1 ship waiting to go in and they’ve currently got 40. I think some of these machine learning examples on using imagery can really help us understand just the sheer volume of transportation issues I think we are seeing across the world. It is not just the UK that has got shortages of important things like toys for Christmas, but it is all around the world as well.
Sam: You’ve kind of touched on that, Zoe, about the machine learning element of imagery. I think the recent product released by Esri called the ‘Land Cover,’ well, it is the 2020 Land Cover data set. It has just been announced this summer and was released into the Living Atlas.
It is essentially a global data set that shows or categorises every section of land into 1 of 8 categories, I think it is. They use machine learning to essentially identify the different areas and well, yes, just work out the, on average, over the whole year based on, I think, 50 or 60 different image data sets.
They basically calculated what the average response was and then used machine learning to say, “This is a built-up area. This is trees. This is forest,” which I thought was too good to be true. Then I explored the data and yes, it literally is the entire world and it is so, so interesting.
Then, a couple of weeks later, they dropped a predictive Land Cover data set for 2050. You can now go and see, based again on machine learning, I think they mapped, it was like early 2010s and then to 2020.
They used that 10-year period to then say, “Okay, well, in 40 years’ time, on this trajectory, what are we going it look like? What areas are growing? What areas are decreasing?” Now they have a 2050 Land Cover data set which kind of blew my mind when I was looking at it, but it is so fascinating to see the power of machine learning.
Zoe: Oh, definitely. Those kinds of things, they’ve always been the core of imagery analysis, things like image classification, object detection and deep learning. All the different packages that we can build around those are just going to increase that capability even more.
We've got things, that you can now automatically detect trees and you detect roofs and roof types. Again, kind of similar to the ship one. All that kind of stuff can be useful in lots of different situations.
Sam: Actually, just to dive in there, so if anyone is really interested in the machine learning element of GIS and how it is used with the Esri system, we actually, fortunately, recorded a podcast episode, I think episode three, on machine learning, where we had a really good guest on, Richard, to talk about machine learning. If you want to dive into that, and haven’t listened to it already, definitely explore some of our other episodes to find out a bit more there.
But actually, just to move things on, again, I'm really interested to see how other people are using it. Some of the examples that I've seem more recently you mentioned already, planes flying over houses to take pictures, satellites and various different types of satellites.
But one of the more recent ways of capturing imagery data is through drones. Zoe, have you seen much about that, and what kind of imagery is available through drone capture?
Zoe: Yes, again, another massively expanding field and actually a lot of options in the system to be able to use drone imagery. We've got things like Site Scan that gives you that end-to-end SaaS offering, so you can do all the way through from flight planning. It uploads the imagery to the cloud and then you can use Site Scan to create your 2D and 3D products, so things like 3D meshes and point clouds.
But if you are not interested in necessarily having the end-to-end offering, you can just use drone to map to process the imagery on your desktop, or if you’ve already got Pro and you just want to do a bit of ortho mapping, that is another option.
But I suppose the most exciting thing is a new product called ‘SURE for ArcGIS,’ which creates these amazing true ortho images at scale, really, really accurate 3D meshes. You can completely transform the imagery and the uses for it.
Sam: Sorry, just to ask, I don’t actually know what ortho mapping is. Could you quickly explain that?
Zoe: That is probably a key thing to have mentioned at the beginning, sorry. Yes, so ortho mapping, it is the process of stretching an imagery to match the spatial accuracy of the map. When imagery is taken it might be taken at an angle, and it is that process of using a combination of the location, elevation data and information from the sensor itself to put that image into coordinate space, so into our map space that we are used to as GIS lovers.
Sam: Makes a lot of sense.
Zoe: But there is this amazing new tool that means that you can look at both the top down imagery that we are all used to in our map space, at the same time as looking at it in what we call ‘image space,’ so that is if you have an image that has been taken from an oblique angle like, for example, a plane flying down a coastline, you can view the imagery at the angle the plane took it at in a program called ‘ArcGIS Excalibur’.
The really amazing thing about that is that you can capture data in either/or of the two views. You could draw around the roof of a building, for example, if you want to capture building polygons, and then that will appear on the image space map.
One really good use case of this is in disaster response situations. You could get a plane to fly an area that has been damaged by, say, an earthquake, and then you can view the spatial distribution of damage, but view the actual damage on the image space image.
Sam: A lot of images but I think that makes a lot of sense. Because if there were buildings that were crumbled down, would you be able to do a before and after comparison of the space? Is that essentially what you can do?
Zoe: Yes, so the tool itself has a swipe tool. You can swipe back and forth and determine whether the building was damaged by the event you are interested in. Then, you can also collect vector data on top of that, so then you could do some kind of time analysis based on that vector data of say, damaged or not damaged, and then create different apps all in the ArcGIS system. You create a dashboard based on that data. You are not just interested in the original imagery but you manage to create other useful products from it.
Sam: Yes, that is really cool. I feel like we could keep talking about the different examples of how and where people are using it, or customers are using it. I just find it so fascinating to see the applications of imagery in literally all corners of GIS.
But also what I'm fairly interested in as well is to understand what is coming up. Imagery is obviously quite an exciting space at the moment within GIS. Zoe, could you tell us a little bit more about what is coming up within imagery? What else have we got to look forward to?
Zoe: For me, I think oriented imagery is going to be a big one in the future, so that is looking at imagery data collected from street level, so collecting images on your mobile phone, if you know the height of your phone and the angle you were pointing it at, and your phone has GPS enabled. You can then project those images onto a map.
Sam: That is cool.
Zoe: There are other organisations that do drive throughs, and you can combine that data and build 3D meshes from that.
Sam: Just to dive in there. Beth, in the content space, have you seen any of our partners using this technology?
Beth: Yes, so we've got one, CycloMedia, that we've been working really closely with. Some of the things that you can do with their data, you can measure things like accessibility of kerbs. If you’ve got dropped kerbs, can a disabled person actually use that pavement, that kind of thing, or measuring for utilities. How far is it from the pavement to the manhole cover? Do they need to dig up the road? Do they need to close a road to do works? Things like that.
There are a lot of really interesting use cases for things like this that we just haven’t had in the past, because you’ve either had an image that you could potentially derive things from, maybe on a manual or visual basis, but now we actually have that data sitting behind it in the point cloud Zoe talked about, which you can use for full-blown analysis and measuring, and things like that. It is a really interesting space that is coming to the forefront now as well, so that is- yes.
Sam: Yes. Zoe, we talked about multidimensional data earlier as well, as an element of imagery that is coming up. Could you expand on that a little bit?
Zoe: Yes, so multidimensional is definitely another space that is going to grow, so looking at data that is not just an image, a pixel on a map, but looking at lots of different information all combined. Weather data is a good example. It happens at a certain time, in a certain location, but it also includes things like temperature, rainfall amount, which strictly, you could argue, would be hard to display on a map because that is a lot of different attributes all happening at once.
We have a new capability called ‘Space Time Cubes,’ which allow you slice the data in lots of different ways to gain more insight from it, so that is definitely something to look out for and have a play with, if you’ve never used that before.
Sam: Is that displayed in the Voxel layers? That was new to Pro 2.8, I think, which was the multidimensional Minecraft-y [sic] looking 3D thing that you could create. I think that is multidimensional Space Time Cube things. Is that the thing that you are referring to?
Zoe: Yes, all very New Age and all very exciting.
Sam: Yes, probably the worst description of it but it looks quite cool from a cartographer’s perspective. It piqued my interest, definitely.
Beth: But you’ve got to make sure that your colleagues don’t think you are playing Minecraft when you are in the office.
Sam: (Laughter) I can just blame it on the weather data or temperature data, that is fine, get away with it.
Beth: Yes, definitely work, definitely.
Sam: Definitely work. Okay, so we've touched on such a variety of different topics and elements of imagery, which I think, hopefully, is really useful for those listening. I think one of the key resources that I wanted to point people towards was just this year, we did both an annual conference and a box set, which we did virtually through our website.
If anyone wants to basically check out a bit more about imagery and the certain topics that we've discussed today, I think that is probably one of the best places to go and check it out. Because I know the conference we did, there was a really good presentation from Claire about ArcGIS Image. She showed a really nice example of how you can use it. Then, in the box set videos there is definitely a handful of different imagery-related topics that you can definitely learn a little bit more about.
But have either of you got any more resources that you would like to share with anyone listening?
Zoe: My favourite place is the Imagery Workflows website. It has just had a bit of an overhaul. It looks amazing and it is basically a one-stop shop for all imagery best practice. If you are struggling with something or you want to know the best way to manage different types of imagery, that is definitely the place I would head to.
Beth: I think from my side, as always, it is using the resources available to you from the Esri Learn Paths and things like that. I know there has been an imagery MOOC in the past as well, so a massive online course. They are not running one at the moment, but you can put it onto your wish list so that when they do run one you can notified about it.
Those courses are really good. They are over several weeks so you don’t have to do everything all in one go, but really useful lessons that you can learn and then utilise in your own day-to-day work, or even if you are just interested in it and you normally don’t have to touch imagery at all.
Sam: They are free so it is a win-win.
Beth: Yes, everyone loves something free.
Sam: Awesome then. Hopefully everyone listening has learnt as much about imagery as I have. I came into this episode with fairly minimal knowledge, so it has been really good to talk to you both, but particularly Zoe. You’ve got far more knowledge on this area than Beth or I, so it has been really good to have you.
Thank you to everyone who has been listening. We’d love to hear what you think, so please do get in touch at podcast@esriuk.com. Please don’t forget to subscribe and rate us on your chosen podcast channel, because it actually really does make a difference. We hope you join us again for the next episode and thanks for listening.
Voiceover: The views of the presenters may differ from those of Esri UK.
Sam: That is a wrap.
END AUDIO