My name is Shayn Hawthorne, and I've been here at AWS for about three and a half years, and have thirty years of experience in the space industry before I came to AWS. I started at AWS by building the AWS Ground Station, and I am now the Space Technology Leader for one of AWS’s newest organizations, the Aerospace and Satellite Solutions Division, and I'm working in that organization with a lot of great customers, like Maxar. So today I'm joined by Dr. Walter Scott, the Executive Vice President and CTO of Maxar Technologies. And we're going to be jumping in and talking to you guys about how AWS has been supporting and providing services and cloud capability to Maxar to help them advance the future of space in the cloud. We're going to do that by first off providing an introduction into the Aerospace and Satellite Solutions division, where I'll talk to you about what we've learned and how come we started this new organization to help our AWS space customers.
And then we're going to segue over to a great presentation by Dr. Scott, where he's going to talk about how Maxar truly is applying the cloud in space. He's going to do that by talking to you about some technologies that Maxar has been using in order to help their satellites image the world and do better things for the world, taking care of different types of natural disasters, taking care of helping people figure out how to respond to COVID. And then we're going to expand out into how Maxar is actually supporting interplanetary exploration, how they're helping the world to start moving towards colonizing and working around the moon. And then finally, talking about how Maxar does a number of incredible things, working with the weather.
So, what are we doing with AWS Aerospace and Satellite Solutions? AWS as everybody knows, provides secure, flexible, scalable and cost efficient cloud solutions to help our commercial and our public sector customers, build satellites, conduct space operations, conduct launch operations with rockets, and then reimagine how space exploration can be done in the future. And we're doing that and building for a key reason. That is mainly that AWS has the most experience in cloud, we've been building enterprise systems on the cloud since 2006, and our space customers need that level of experience to help them expand into Earth orbit, to the moon, and then later. We also provide the most functionality of any of the cloud services to support space with more than 175 services in order to provide support to space workloads on Earth, in an orbit around the Earth, and then someday even out to the moon, we're able to give that functionality and help people build the space enterprises that they need, using cloud services, microservices, and products.
We have industry leading artificial intelligence and machine learning, which allows you to work both on Earth's surface in the cloud and up in space to apply machine learning and artificial intelligence to some incredible workloads that can help you identify buildings, they can help you figure out behaviors and infer decisions from the data collected on the satellite, so that you could tip and cue satellites on orbit, or do the same types of operations down on the ground. And when you're getting that data down to the ground, you want to get that data down to the ground quickly into a large network around the world. That's where AWS Ground Station as a service comes in. It allows you to get your space data faster, you can downlink that data to AWS, get it through our Ground Station as a service and into the cloud quickly, so that you can start using these services and products to help process your data, help store it and distribute it to your customers around the world. And then if you need support, if you want to build things at a greater scale, we have the largest community of partners available so that you can access thousands of systems integrators and independent software vendors, our ISVs, as we call them, to help you build new things in space and continue growing the world's efforts to grow into space, to the moon, and beyond. And, why do we do that? We do that because even in space, customers want a service experience.
And, that's what AWS delivers. It delivers you first of all security. We have a resilient global network of data centers composed into our AWS regions and availability zones, to provide you unrivaled redundancy, processing, and velocity for your data around the world and in processing it. Then, we provide you the elasticity that you need to be able to scale your services by the minute.
That means that based on different use cases and workflows that you might be working on, you're actually able to scale out when you need to, get all the extra capacity that you need, the storage, the ability to move your data, and then scale back in when you don't need that service. And that allows you to be efficient and save your capex for different uses other than building Ground Station networks, or building your own ground operations. And then, you're taking advantage of AWS’s worldwide network to provide unrivaled low latency access to your data. If you're using Ground Station as a service, and then moving the data that you receive at the AWS Ground Station service around our network, you can actually get data at any of our Ground Stations around the world to your virtual private cloud in hundreds of milliseconds. That allows you to get your data where you want it and when you need it in near real time. And that gives you immediate fulfillment, so that your customers are able to get the data and the decisions and the inferences that you're providing them.
Or you're able to get your data to them at the velocity that they need. So, let me give you an example of where that scaling out and scaling in, and that access to AWS services can really enable the space service experience. Let me show you this picture of the southern part of the United States. As you all well know, we've been watching the southern United States for a number of weather events, hurricanes, and other issues that we need to be careful of because we want to protect people, and we want to warn them when there's issues that they have to be aware of.
So in a typical day, we have a low weather or no weather environment, the cloud is able to allow our customers to really probably just move along at a very low rate, they may be doing some level of dynamic analysis, some level of prediction, but overall, it's a quiet day, there's not a lot of work going on. So there's a low amount of compute being used, and a low amount of storage. But, let's say that a weather event does start to move in.
Now things start to spool up. First responders and emergency reaction people, state governments, national government, FEMA, everybody needs to start monitoring this weather and start getting rather ready to weather the event. And now your processing starts to spool up, you're starting to collect data, you're starting to distribute data through content delivery networks, you're starting to see people querying and getting information and then putting information back into the cloud, to enable them to coordinate responses and start to do work.
This is where the cloud comes in to be so valuable. As you spool up, and you need more and more services and more and more capacity, more and more storage, you have the ability to do that. And you're able to use that information to benefit people, to alert people and help warn people. And then when you go back to that previous quiescent state where you don't have much going on, you can spool down all of those servers, all the EC2 instances, you can start to decrease your S3 storage that you're using to monitor that event. And pretty soon you get back to normal, and you're not paying for any of the services that you needed during that big active event, because they're simply not needed at that time. That's the type of power that we're able to provide to our AWS customers.
And so, I wanted to give you that example. And then once again, welcome Dr. Walter Scott, to present his discussion about how Maxar Technologies is applying the cloud in space. Over to you, Walter.
DR. WALTER SCOTT: Thanks, Shayn. Maxar is a company that's behind many of the things that we depend on, whether it's satellite imagery that we see on our mobile phones, or creates the maps that we use to go from place to place. The 3D models that are used for placement of cellular infrastructure, or satellites that deliver satellite radio, satellite TV, satellite internet, or even things that go to other planets. I want to talk about one of the first examples of how Maxar and AWS are working together.
Hurricane Laura was a devastating wind storm that went through the southern United States this past summer. And what Maxar does, in response to any disaster of this type, is start with our massive library of up to date satellite imagery. About 100 petabytes that’s stored in AWS.
We pull that imagery from Glacier and S3, and use that as part of something that we call The Digital Globe, which is a living digital twin of the surface of the earth. It consists of the imagery, and a variety of other information layers, many of which we extract from the imagery. And the way that we extract those data layers from the imagery is using something that we call DeepCore that runs inside the AWS Cloud. DeepCore starts by pulling imagery from that 100 petabyte library, the most current imagery that's needed to support the next step, which is applying AI.
Now, when we're talking about the volumes of data that we store in AWS, there's no possible way to use humans to do all the work. And that's the reason why we use a set of trained AI models to extract various information layers. Those models are for things like roads, or buildings, or solar panels, or any of a number of other objects that you might want to recognize in the imagery.
We label, meaning we create the things that the model recognizes. We label the imagery, store that on RDS, and then we use the cloud to perform two functions. One of them is to train the model in the first place.
And the other is to score the resulting labels for the model as a way of improving the quality of the AI model and then updating the model inside S3. We also use Bitbucket and Jenkins as a way of continuously deploying the code into the DeepCore architecture. So after the wind storm passed through, the day after we collected post-event satellite imagery of the same areas, and then overlaid that with the information layers that we had extracted from the pre-event, the most up to date, pre-event imagery, providing things like roads, and building footprints that were then ingested into the applications that the first responders were using to find safe routes, and to find a way to damaged structures to evacuate or otherwise provide necessary services. Well, it's a wind storm. And one of the questions is, where's that windstorm going to hit? Well, Maxar has been working in the weather domain as well. The NOAA supercomputer currently generates forecasts in about 100 minutes, and it does that 4 times a day, at the top of the hour.
Well, we decided to take that and put that in AWS into the model. And we were able to deliver that 58% faster, saving basically an hour in the amount of time that it takes to generate the forecast. Well, when a storm track is moving along the ground, the storm track changes.
And so having that forecast an hour ahead of time, means an hour extra time to prepare -- to board up the windows, to plan an evacuation, or to hunt down the pet that decided to hide at the worst possible time. Now, the way that we do that in AWS, you see here on this chart, and the benefit of working in AWS is not only do we get performance, but we get resiliency, it operates in a number of availability zones, so that we're not reliant on a single location. And it's also able to scale up and scale down. So we only run the model when we need to, instead of having a data center that's operating 24/7 and not needed. The net result of running this, I mentioned 58% less compute time, basically an hour faster than the NOAA model, for much less cost than the NOAA supercomputer.
Again, because of that scalability. We generated the same forecast quality as the high performance models from NOAA. And we believe it's applicable to other weather models and we're in the process of updating it to support other weather models. More on that in a moment. So by being able to deliver the forecast 58% faster that opens up the possibility of running the forecast more times per day. And we'll talk about how that's applicable in a moment.
Well, the world is a really big place. And so, one of the core questions that Maxar has to address is where do we take the pictures? Well, there's some things that help us but the most frequent change occurs where most people live. And you can see in this map, that there's a high correlation between population density and the density of change on the surface of the earth. The world’s population is concentrated. About 8% of the land is considered populated. More than half the population lives on less than 1% of the land in densely populated urban areas.
But 95% of the population live south of 50 degrees latitude. And what Maxar finds is about 5% of the land drives about 95% of our revenue. So, figuring out where in that really big planet to take pictures is a pretty important job for us. And, we have some sophisticated algorithms we use to do that optimization. But it's also important to figure out when to take the pictures.
For example, if it's cloudy, there's not a lot of use in looking at the ground because the cloud obscures everything. Well, after Hurricane Laura hit here are six image strips that were taken by our Maxar satellites in the day after the hurricane hit. And you can see that the only areas that are cloud free are the ones that are outlined in purple. So, having an up to date weather forecast at the last possible minute, would have allowed us to be more efficient in the way that we used our satellites, and also use less storage, storing data that is otherwise not usable. And we're in the process of taking our weather forecasting model and using it for precisely that purpose. What you're seeing is, all of those aspects that I talked about, are living in the AWS environment, as Maxar is all in on AWS, and whether that's the ingest of data both from our own satellites or other sources, the weather forecasting, the Digital Globe, DeepCore to extract a variety of information layers, supporting the cloud, all of this running over and over continuously inside the AWS environment.
Well, let me bridge to another example. Many of you have been following the wildfires that have been devastating communities in the western United States. Well, one of the problems with a wildfire is where there's fire, there's smoke. But fortunately Maxar has the only commercial satellite that's able to see through the smoke, Worldview 3, to be able to identify where the fire lines are, where burned structures are, and to provide that to first responders and communities, which we do via the Open Data program stored on AWS. Well, let's talk about another aspect of satellites, which is, when do they collect their data? Now, for the longest time, Earth observation satellites have generally collected in either the late morning or in the early afternoon. I'm not going to get into the reasons for that.
But it boils down to synchronizing the orbits with the rotation of the Earth and the Sun so that if you're taking an image at, say, 10:30 in the morning in New York, and the satellite zips around the planet, 90 minutes later, it's over Chicago, it's going to be more or less about the same time in Chicago, plus or minus. Well, not everything happens at 10:30 in the morning or 1:30 in the afternoon, wouldn't it be nice to be able to get looks throughout the day? Well, of course, if you have looks throughout the day, you're going to need those weather forecasts to be done much more rapidly to be able to support those multiple looks. Well, we're actually doing that. We're building a set of satellites called Worldview Legion that will be joining the Maxar constellation next year, the first couple of launches will be happening next year for an initial block of six legions. Well, in addition to forecasting faster, with the ability to look at the ground so many more times per day, you want to be able to get the data, and the way Earth observation satellites get data to the ground is when they're in range of a ground station, they use radio to downlink the data to the ground station. And these shaded areas you can see are the areas of the ground stations that Maxar beams its satellites to.
Well, not everything's in range of the ground stations, so we have to store data on the satellite. And, that can take time. So adding more ground stations allows us to get the data to the ground faster.
And that's one of the ways we've been working with AWS Ground Station as a way of getting more often the data to the ground faster. How fast? Well, actually, we've been able to demonstrate from collection of imagery to being in an S3 bucket in the cloud in less than a minute. That's pretty fast.
Well, let's go to another planet. I mentioned Maxar supports a number of planetary missions, one of them is we built the robot arm for NASA JPL’s Mars Insight Lander, which you can see here in this artist’s view. This is the robot arm in operation in the lab on earth, and the robot arm is used to pick up some of the instruments that were taken off the body of the lander and then placed onto the Martian soil to perform a variety of measurements.
Well, one of those instruments which you can see in this next picture, the one that you see on the right, it's a mole digger, and it was designed to dig under the soil on the Martian surface to go make measurements below the surface. But if you were following the story around the Insight Lander, there was a problem, the mole essentially couldn't get a grip, it was spinning around, it wasn't able to grab on and dig the way that it was designed to do. So the very smart engineers at NASA JPL came up with an idea, which was to use the Maxar robot arm to do something it wasn't designed to do. They put a scoop on the robot arm and stuck it on top of the mole, pushed mole into the Martian soil, so that the mole digger was able to grab onto the soil and dig to do its job. First case of interplanetary whack-a-mole as far as I'm aware.
Let's come back closer to the ground or closer to Earth. I mentioned that Maxar builds communication satellites and you can see a couple of them here. Those satellites sit out 25,000 miles above the ground.
So to be able to make those satellites work, they need a lot of power to get that signal all the way down to the ground. Well, we've used that technology to support a NASA mission called the Lunar Gateway. That's a space station that orbits the moon that can move from place to place because it's used as a staging point for astronauts on their way to do various missions to the surface of the moon. It's powered by very large solar arrays, the same kinds that we use for our communication satellites, and it uses a propulsion technology called Electric Propulsion.
And it's one that Maxar has been a leader in for a number of years. Those are those really cool blue exhaust lines you see coming out of the back. Well, what's important about Electric Propulsion, is it gets a lot more gas mileage, it basically gets 10 more times as much gas mileage as you can get from chemical fuels. So you can go farther on a tank of gas.
Of course, what happens when you run out of gas? Well, we're actually working on that too. There's a mission we're doing for NASA, OSAM-1 and SPIDER. And that is to go up, grab onto a satellite, refuel the satellite in space, and release the satellite. The other thing we're doing on the same mission is demonstrating the ability for a satellite to use robotics to assemble itself. In this case, assembling a large reflector dish as a way of moving from having to launch satellites that could only fit inside the nose cone on a rocket, to being able to build much bigger structures in space.
Well, I mentioned communication satellites. And one of the satellites that we built is for the Australian National Broadcasting Network. Sky Muster II, and it was used to beam high speed internet to roughly 400,000 Australians scattered throughout the interior. Australia is a huge, huge continent.
One of the things that it was used for is distance learning because many of the school children in remote communities wouldn't be able to effectively commute to go to school. Well, obviously, distance learning is something that we've become a lot more familiar with in the time of COVID. Speaking of COVID, Maxar has been supporting COVID relief efforts by releasing information, satellite information and other data, for COVID relief responses in about 20 countries around the world, as part of the Open Data program where we release the data using AWS S3. I'll just pick a few examples.
In Sierra Leone, here is a city above Sierra Leone. We worked with the government of Sierra Leone and a number of other partners to support their COVID response efforts. For example, one of the partners is Grid 3.
The census data that Sierra Leone had before the COVID outbreak was old. And in managing an outbreak you need to know where the people are. So Grid 3 developed population estimates based on the most current Maxar satellite information, satellite imagery, and used that to feed into the COVID response efforts for the government. And it all fed a web portal that the government used to provide information to its citizens, as well as informing many of the mitigation strategies -- social distancing, evacuation, delivery of relief supplies. There are a few other examples as well.
When the Philippines were hit by Typhoon Vongfong, Maxar Open Data was used to support the evacuation and relief efforts in a way that maintained social distancing. In South Africa, Maxar data was used to map informal settlements. That data was used to then plan how to deal with an outbreak if one occurred and how to safely evacuate or deliver supplies to the people who were affected. In Malawi, our open data program was used to match the location of mobile payments to structures to mitigate fraud.
In Kenya, which was experiencing a major locust outbreak, the open data program was used, while farmers were told to stay home to monitor food security remotely, while maintaining social distancing. And in Bangladesh, open data was used to support mapping of dense urban areas to support government efforts at contact tracing. So these are just a few of the many examples of how Maxar and AWS work together for a better world. Thank you. SHAYN HAWTHORNE: Wow, thank you, Walter.
That was really an incredible discussion. And it really does show that Maxar is applying the cloud in space, I wanted to finish up with a couple of key takeaways for everybody to just show you and discuss how AWS really is offering a cloud that offers significant value to the space enterprises and their data. As you can see several examples in that neat discussion, we're offering scalable mission data processing, so that you can actually ramp up the capacity that you need to support your space mission, and then take it back down so that you're only paying for what you need.
And then also, you saw an incredible use of data analytics and storage, where that data is not just sitting there, it's not just waiting for somebody to come look it up again. The data is being used every second of the day to analyze, predict, estimate, and do great things to help the world. And then finally, that leads into machine learning and artificial intelligence. They're doing fantastic things using that AI and ML to help them find buildings, help them show where disasters are causing problems, helping to predict where people and populations are for COVID purposes, and help people.
And that all comes from satellite products and data dissemination being used on the cloud with services when you need it, where you need it, around the world. So, those are the key takeaways. That’s showing you how AWS and our new Aerospace and Satellite Solutions team is working hard with incredible customers like Maxar Technologies to do great things for the world.
Thank you for coming today. We really appreciate your time.
2021-08-19