Geo for Good 2019: NASA's Servir use of Earth Engine
DAVID SAAH: So welcome, everybody I know this is the last session of the event I really appreciate all you guys coming here My name is David Saah And as you can see from this huge list of folks, I’m part of a large effort here And what we’re hoping to do today is to go through a couple different things We want to give you guys an overview of SERVIR for those of you who aren’t familiar with it It’s a pretty amazing program We want to talk to you guys a little bit about the way that we operate SERVIR is a little bit different than the traditional USAID or NASA project or program We’re going to give you some insights on how that actually works And last year, what we did was that we went through very briefly all the different tools and services that we were delivering through the hubs, which was great, because we got a lot of really excited folks about a lot of the code that we had and the tools that we were building But it seemed like we didn’t have enough room for conversations And so the last thing we want to do for today is really enable us to have a dialogue, break out into smaller groups if interested to take– we want you guys to take and learn and contribute to some of the efforts that we actually have Does that sound like a plan? Yay And in the title– we got stickers [LAUGHTER] All joking aside, we have a bunch of stuff here for you guys to take Everyone likes the meatball, including me I look like a meatball But take it Use it Distribute it, OK? So what is SERVIR? So SERVIR is a joint program between USAID and NASA It’s a real collaboration between those major organizations And the real idea, the purpose, the goal of it is to take space-based technologies– the tagline is take space-based technologies down to village-level decision making to make an impact Now, what does that actually mean? Does that mean that we’re going to actually take a satellite and give it to a villager, hoping that they’re going to be able to improve their farming? No The idea is to be able to take Earth observations, the analytics associated with Earth observations, and make that village or that community more informed about the choices that are being made for them or the choices that they’re actually making in order to have a more sustainable future All right, that’s the goal That’s the vision And there’s a variety of different service areas that we work in And you’re going to hear us talk a little bit about service, because we see this whole program as service to a community, service to the environment And we have a service delivery framework in mind And the places that we work is agriculture and food security All right, think about all the things around that And I know we’ve had a lot of talks this week about that There’s a whole land cover component We had a lot of talks about that Waters and disasters– lots of talks about that– weather and climate So all the tools that we’ve been hearing about all week, all the things that you can do with Earth Engine and the Google technologies fit really nicely into all the major services that we’re thinking about Now, where do we work? Where is the SERVIR program? How is the SERVIR program actually organized? Right now, we have five hubs We have one in the Mekong region We have one in the Himalaya Hindu Kush We have one in East Africa, one in West Africa Africa’s humongous We need two And then we just launched one right now And when I say, “we,” this is the global “we,” not me we– launched one in the Amazon And there’s also another effort being launched right now in Central America as well So if you look at this, it’s a large part of the planet that really needs these technologies to help them improve decision making It’s not a NASA drop program It’s not a USAID drop program The people that are actually implementing these projects are usually consortiums in some examples or institutions that have been in the region for a long time So it’s really enabling local folks to do things And what we’re really trying to do is we’re trying to bridge this gap And you guys, I mean, we’ve had this conversation I don’t know how many times over coffee But you’ve got the science folks And you’ve got the end user needs And that last mile or that keystone brick or that gap in this bridge is oftentimes missed And what we’re really trying to focus on is actually connecting this part, connecting this part so we can have a continuum of information flow, knowledge, and hopefully impact And we do this by leveraging the science, all kinds of cool satellites And every day, this spectrum, this constellation is changing in a bunch of dynamic ways to take advantage of all kinds of data that looks like this Imagine being able to see the whole planet live on an ongoing basis with technologies like Planet and some of the new stuff that’s coming out
This is doable And this one– if you look at this one– is it actually showing up? Or do I have to push it again? Aw, man Darn it Go! So if you did see it– we’ll make the slides available But when you run it on your local machine, what you would see is an accumulation of precipitation in real time So some of the new sensors are actually capturing that information as well So imagine being able to take all that data and all that science and putting it together to fit the end user need Now, the way that we actually do this is going to be discussed by Tim You want to get up here and talk about this? How do we actually– how do we do it? TIM MAYER: So SERVIR is very interesting in the way that we approach the services themselves So the model that SERVIR uses is really to generate these services through a service planning framework So there’s three main steps that go into that So it really centers on the collaboration with stakeholders, so doing needs assessments and identifying and prioritizing what those development challenges are, and really focuses on designing a tool or a service that is detailed So that way, as David had mentioned with the service and the end users, often people like to come in and just make a tool But that’s not our approach at all So we want to start from the very beginning and do those needs assessments and lay out all the components that go into that service design And then the final component of that is the actual implementation, so making sure that we have sustained efforts in the region And when we’re releasing or implementing some of these tools, we’re there And the hub members are also there to make sure that it’s being addressed and taken into adaptation and is being utilized by the end user So it’s a cyclical process But you can see as it moves across, there’s that first needs assessment and then service design and delivery Those are the main staples of this framework So we really like this idea And we’ve implement it now across all of the hubs So here’s just an example from the Mekong’s website So when we say a service, it doesn’t necessarily have to be just a geospatial tool, for instance Sometimes we like those But it could be something like a technical consultation or training or capacity building That’s another example of some of the needs assessment that– once we do a needs assessment, we go in place and do some of those as well So it could be data sets, tools, capacity building trainings, or technical advice So those are some of the components of the things that we do at SERVIR So with that being said, you have kind of an idea of the model now We’re going to go into some examples from all the folks who are here from SERVIR And we can move to the cafe discussion as well So here’s the first example from the Mekong DAVID SAAH: Why don’t we talk about our format? TIM MAYER: Yeah DAVID SAAH: So I’ll be right there So the way that we’re going to do this is we’re going to invite a representative from each of the hubs to come and give us a couple-minute example of some of the service designs that they’re doing And then afterwards– let’s hold the questions, because we’re going to break out into little subgroups where you can actually dig in and grab the code and understand what they’re actually doing Does that sound like it’s OK with everybody? All right So the first example– let’s bring in the Mekong group Biplov, yay Give Biplov– [CLAPS] Yay, Biplov [APPLAUSE] BIPLOV BHANDARI: Thanks My name is Biplov I work as a GS and remote sensing developer for the SERVIR Mekong project So one of the services that we have been delivering to the five countries in the Mekong is the Regional Land Cover Monitoring System So interesting idea with this system is that it’s based on the primitive layers So primitive layers are the biophysical layers And we have these– I think more than 15 classes And then the beauty of the system is that each of the countries can then use those primitive layers to design their own land cover system So as you can see, the– sorry DAVID SAAH: Not that one, this one– the laser one BIPLOV BHANDARI: OK, so once you have the primitive layers, then we invite the stakeholders from different countries to design their own land cover system and use those primitive layers to generate the final [INAUDIBLE] classes So this is one of the main services that we have been working on Please go and explore the system and play around in this website given below And the methodology is explained there Thank you [APPLAUSE] DAVID SAAH: You want to do CEO– the Collect Earth?
I guess I’ll do Collect Earth Online So can I add to your slide just a little bit? BIPLOV BHANDARI: Yes, please DAVID SAAH: So if you look at the bottom, there’s two URLs The Regional Land Cover Monitoring System is probably an hour-long conversation If you want to go to the URL, you could see how we actually do land cover monitoring at scale It has all the code It has the way that we pull stuff together It even has the code to build this interface If you wanted to take this and take off our names and put your names on there, feel free to do so The other challenge that we had when building a land cover monitoring system, just in general, is collecting ground reference data How many of you guys have problems with collecting reference data? One, I see two Come on, people Tell the truth [LAUGHS] Collecting reference data is hard, right? And so what we ended up doing here is we ended up striking a collaboration with FAO, the UN’s Food and Agriculture Organization, to piggy back on top of their Collect Earth system, where we’re were able to take that– I don’t know if you guys have ever played with it– is to take that platform and turn it into an online platform where we could crowdsource a series of experts to collect information that we would need to map out different things that we put together for land cover mapping And what’s nice about it is that that tool was built specifically for the Land Cover Monitoring System for us But there was a knock-on effect where they actually ended up– “they” being FAO– ended up taking that tool And now they’re using it for their Global Forest Resource Assessment And they’re collaborating with us So it’s another example where we co-design and co-develop these products with folks that are able to pick it up and use it on their own The next set of examples that we have come from West Africa Come on up Yay [APPLAUSE] SALIOU NDOYE: OK, good morning, everybody I am Saliou, User Engagement Lead And I’m with– ABENA ASARE-ANSAH: OK, I’m Abena from Ghana SALIOU NDOYE: OK, I will just give you a quick overview of what we’re doing at West Africa And Abena will go in more detail in one of the services we are working on In West Africa, we are working under these four focal areas We have developed different services Under the weather and climate, we are working on developing a service for dust management It was a requirement from the UNCCD, who we are working with Under the water resources, we have a service dealing with ephemeral water bodies It’s a need for herders in the Ferlo region And we’re working with the Centre de Suivi Ecologique to develop a service In the next one, we have another service also dealing with ground water monitoring We are working on with AGRHYMET, trying to map out groundwater resources in two regions in Niger, in Tillaberi and– in Tillaberi and– I don’t know the other region But it’s in two regions in Niger The other service we are working on– we, in collaboration with FAO and CLCPRO– is locust monitoring We’re trying to develop an early warning tool, trying to prevent locust outbreak in the region And this is an important service covering the West African region We’re working with FAO And under the land use, land cover, we have different services we’re working on And here we are collaborating with the UNCCD And we presented a report we drew up for the UNCCD for the West Africa region We presented at the last COP in New Delhi And now we have been asked by UNCCD to develop a series of maps of land vulnerability to droughts This is one of the services we will be working on And we are working also on illegal mining And Abena will present in detail that service Also with the same institution, we are working also on charcoal production I will let Abena fill on this ABENA ASARE-ANSAH: Thank you very much So talking about illegal mining in Ghana, what mostly happens is that these people are mining along river bodies or along rivers And then they are also mining into protected forest areas So realize that people are cutting down trees along these protected areas and in these forest reserves And what we have here projected is a radar image
from Sentinel-1 that was captured from 2015 to 2019 And so we set a threshold to be able to illustrate these galamsey, or illegal mining sites And so the dark blue areas show the extent of illegal mining for one of the years And realize that this illegal mining causes a lot of threat to the community members It ends up polluting the water bodies as a result of the mercury deposits So some people end up losing their lives And then we are also losing a lot of trees So and another thing, too, is that these areas are quite inaccessible So the best thing to be able to monitor this was to just assess these areas using satellite images And another thing, too, we have done is that we’ve created an app, a mobile app, to be able to use as a validation or reporting platform so that members or people who are able to access these areas will just update us on our server so that we will know the specific areas that are really illegal mining sites or not Thank you [APPLAUSE] DAVID SAAH: Just to be clear– or just to be clear– this area– is it water? Or is it illegal mining? ABENA ASARE-ANSAH: That’s illegal mining DAVID SAAH: That’s illegal mining ABENA ASARE-ANSAH: Yes DAVID SAAH: Look how much there is There’s a lot of it, right? Thank you If you look at the series of tools that you can look at, that have been built on top of Earth Engine technologies, there’s a whole series of them So during the breakout session, please feel free to ask about any of these, OK? Thank you guys Eastern and Southern Africa, you ready? Come on, Steve [LAUGHS] Here you go STEVE FIRSAKE: OK, so I’m Steve, a geospatial engineer We work within RCMRD under SERVIR We’ve been to two countries in Africa There’s a whole bunch of– DAVID SAAH: [INAUDIBLE] STEVE FIRSAKE: There’s a whole bunch of tools that we’ve created And I would say ever since 2017, we’ve been using Google Earth Engine to create most of the products for the services But the biggest service that we’ve done on Google Earth Engine is this one So when you hear about space to village, this tool actually really brings it down to that We, in collaboration with the government of Rwanda, actually came up with this tool, the Rwanda Land Use Decision Support System, to actually manage all the affairs of what the country is doing to actually push their development forward to the sustainable development goals, which we hope to achieve globally So basically, what they did is they have a bunch of layers They were really concentrated about stuff with landslide, land cover change, forest degradation And they wanted a vulnerability map created out of that so that they can make decisions about where they relocate people to Which pieces of land do they actually focus on changing them back to get reforested, and which areas do they actually start practicing crop farming on it, on them? So previously, they used to use statistical data, basically tabular data, no spatial information from that But then using this tool, which we actually gave to them officially in May this year, I’m also hoping to look– and I’ll be writing papers and articles and tweeting about all the changes that Rwanda does using this tool So thank you DAVID SAAH: Thank you [APPLAUSE] AUDIENCE: Can you just tell us what the colors mean here? [INAUDIBLE] DAVID SAAH: What is red? What is yellow? What is blue? Is blue mining? STEVE FIRSAKE: No, no This is the– [LAUGHTER] This is just one of the layers Later on, we’ll show you We have brochures and the links for this Later on, it you come in later This is just woody cover changes It’s one of the layers that they are actually concentrated on I know you can’t see the different layers on here But this is deforestation, degradation, [INAUDIBLE] DAVID SAAH: Steve, Steve, Steve, here, use that STEVE FIRSAKE: Yeah, but then you can open up this link Yeah, you can open up this link, or see me afterwards Let me know [INAUDIBLE] time into the other guys who are going to speak But me and Alan– Alan, wave Alan has brochures of all the services of the things that we’ve done And feel free to see me after to look at this and ways in which you can implement in any country that you are interested in OK, thanks DAVID SAAH: Thank you, Steve The Himalaya Hindu Kush team Here’s the– or you can use this If you’re going to walk, use this one, OK? SUDIP PRADHAN: Good morning, everyone I’m Sudip Pradhan And me and my colleague, Kabir Uddin, we are from Hindu Kush Himalaya
Actually, if you are looking at the green patch there, so it’s the Hindu Kush Himalaya So it’s a mountainous region We are based in– our hub, Hindu Kush Himalaya– is based in ICIMOD And we are working for eight different member countries, from Afghanistan, India, Pakistan, Myanmar, Bangladesh, Buhtan, China, and Nepal And we use Google Earth Engine for various different apps, various service areas Land cover is one of them, also flood mapping And then my colleague will be talking more about our application of Google Earth Engine and wheat area mapping in Afghanistan And we also do lots of capacity building program in our member countries, actually teaching folks how to use Google Earth Engine KABIR UDDIN: Thank you, Sudip And my name is Kabir Uddin I’m working as a [INAUDIBLE] remote sensing specialist Wait, I’ll– [LAUGHTER] DAVID SAAH: Sudip, come here, and hang out with me KABIR UDDIN: As Sudip was mentioning this large area and Tuesday morning, you have seen now a presentation on how they are climbing It’s mountain region It’s not easy to climb or go in those areas easily And if you see, also, IPCC report, they used to say mountain region is a special data gap So under the SERVIR, we are working Those are the land cover mapping, as Sudip was mentioning a bit earlier But now this work demanded by [INAUDIBLE],, we need some information [INAUDIBLE] But [INAUDIBLE] when we are doing, it was very much challenging for us But when Google Earth Engine came, and then this became very much easier for us And also, when we see David Saah’s presentation, he was mentioning about the service area We are working for our research Or we are using for use One of the important things for us to do when we are working– we are closely working with government So we are not developing OK, please take it and use this We are not doing We are always together OK, in the workshop we are together When we are defining the need of this, we are together We are maybe coding together But we have maybe someone doing coding more But someone always from their local knowledge, they are always this So as one example, we are going to show in Afghanistan In Afghanistan, 80% people depending on this table for this wheat and land [INAUDIBLE] also If they are not distributing their wheat production, which only there is very challenging for them So under the SERVIR, we develop one tool, very much close collaboration with the Ministry of Agriculture and Livestock And they are well-trained, actually– this system in the code base they can run to monitoring their wheat growing area And besides, here is some few things has been applied, like understanding the crop [INAUDIBLE] like crop calendar, chronology and all aspect being considered And it’s well accepted by Afghanistan ministry So with that, actually, thank you so much [APPLAUSE] SUDIP PRADHAN: We also put together an app for normal users who can select a district and get their statistics By the way it’s App Engine app DAVID SAAH: Thank you, guys Amazonia, come on up So as they’re walking up, you can kind of see the spectrum and breadth of the different sorts of things we’re working on And we’re just giving you some of the highlights When you actually– if you were interested in some of them, you could talk to the individual groups And they’ll give you a more detailed list, OK KARIS TENNESON: Hi So the Amazonia hub is a little bit newer We’re six months old So we’re just in the stages of doing our user engagement But we have a couple first year services that we’re developing So the first one is building on Terra-i Do you want to say a bit about Terra-i? And then I’ll talk about the phase 2 OK PAULA PAZ GARCIA: OK, Terra-i is a monitoring system to analyze land use cover change and deforestation We want to analyze and use Google Earth Engine and TensorFlow to classify different crops in the Amazon And currently, we have tools available
It’s Conservation Amazonica, ACCA It’s available and analyzes the deforestation It’s near real time, monitoring deforestation They have a lot of examples And they use high resolution imagery and Sentinel to analyze, for example, fires and deforestation in Brasilia They create a simple app to analyze the focus of fires and deforestation And this is the link available And I work with a Terra-i project It’s a monitoring system to detect deforestation We have the available data set on the tropics And SEPAL, yeah KARIS TENNESON: And then the last service that we’ll showcase is– we’ve been working with the Ministry of Environment in Ecuador in partnership with FAO, so UN FAO, to move some of those common scripts that we’ve all been learning about to track forest loss or forest degradation from the Earth Engine scripting environment into a GUI that’s specific for monitoring your forest lands for greenhouse gas reporting So the app is called SEPAL And it’s sepal.io And we’ve moved– and the big “we,” here is FAO and SERVIR– have moved some of these applications into this platform So for those of you that went to the Area 2 estimator this morning, there’s a GUI built around those in SEPAL You can do deforestation using the app in SEPAL And then we started building some degradation applications and moving it in here and linking it up with Collect Earth online as well, which is what the Mekong hub already presented So this is an example where you can do some of the sample design to place your points, to do area estimates of your forest loss layers, build that within SEPAL, and then, oh– well, the next slide that’s not there would be moving it into Collect Earth online So we’re starting to integrate some of these pieces and make it a little bit easier for countries looking at reporting their greenhouse gas emissions to just use one tool and have all of these connections already premade for you, so DAVID SAAH: Perfect Thank you, guys [APPLAUSE] So before we move into the cafe style, I just want to say a couple of additional things about SERVIR So one thing is a common question that we hear is how do we get involved? You guys are working in an area that we’re working in And we’d like to collaborate We are very open to that If there is anything that you think is complementary or that could help lead to an impact on the ground, feel free to reach out to us And we will lock you into the appropriate hub and the appropriate contact person Tim, raise your hand Emil, raise your hand Those are our NASA folks Most people know Emil via Twitter [LAUGHS] Tweet at him, please, and also friend him– follow him What’s the proper word? EMIL CHERRINGTON: Follow DAVID SAAH: Follow Follow him [LAUGHS] Thank you And we’ll connect you in The other common question that we hear quite a bit about is, hey, is that– I’ve said this a few times I want to say it a few more times Can we actually gain access to your code or your applications? Yes, yes, yes, yes, yes Please take it And there’s a variety of different ways that we actually distribute it Some feedback would be nice If you do take something, if you improve on it, please let us know so we can improve it for the regular system Or if there’s a better distribution method of how do we share code in a way that’s better than GitHub GitHub seems to work really well for programmer types of folks But what we’re also noticing is that people are also asking to transfer some of these tools to their regions, not necessarily the code to build it themselves, but the application to use themselves That’s a new common request that we’re hearing So if you hear about that as well, and you have some grand ideas about that, please let us know Are there any other general questions before we break out into groups? Yes? EMIL CHERRINGTON: So Dave, since you’ve been kind of [INAUDIBLE]– DAVID SAAH: Come up Can you come up here and talk through here? EMIL CHERRINGTON: It’s a question for you, though DAVID SAAH: Oh, because they’re recording I’m just trying to follow the rules EMIL CHERRINGTON: So just a question for you, Dave– so since you were really the– you connecting with Nick and with the Google Earth Engine team, Tyler, Noel, all the great folks, et cetera, and also
Jamie Shout out to Jamie Favors here from NASA headquarters He’s the partnerships lead at the ESD level Could you tell us just a little bit, because some people might be interested in knowing kind of why you thought, hey, Earth Engine is something that the SERVIR network could take advantage of? Could you maybe speak a little bit to that? DAVID SAAH: Sure, yeah, definitely So you guys heard us talking a little bit about this being a needs driven program– service, code development, all these fancy words– and it’s all true And the first step in all of this process is finding local collaborators who could actually use this stuff and make a difference And when we do that, the idea is to go through and try to identify all those barriers to entry that is currently preventing them from actually using Earth observations to make an impact That’s our goal And what we found was that, hey, there’s a limitation in terms of access to data There’s a limitation in terms of understanding Earth observations and remote sensing, what you could do with it There’s a limitation in terms of compute There’s a limitation in terms of collaboration How do you share this stuff? And when we went out there and we started looking at all the different technologies and solutions and approaches, at that time, Earth Engine was just getting going And it was a natural fit It eliminated a lot of the barriers and gaps that we had in order to be able to do that code development process with folks on the ground It allowed us to get to the science quickly, the questions quickly, without all that overhead upfront That was not a planted question, was it? EMIL CHERRINGTON: [INAUDIBLE] DAVID SAAH: Wait, wait, microphone please EMIL CHERRINGTON: Maybe we could just– also, just for people’s general knowledge, since again SERVIR is a program within the capacity building program at NASA, could you maybe talk a little bit about the partnership between NASA and Google? DAVID SAAH: Yeah, I mean, so– Jamie, you have to come up here Well, he went part of the way up JAMIE FAVORS: The cameras– at what point do I get into the camera? DAVID SAAH: Come in Come in You’re good Stop JAMIE FAVORS: We’re good OK So in general, I mean, I think we’re just trying to– NASA is trying to get better at– we do really well partnering with academic institutions, people doing research outright that are used to downloading NetCDF and unpacking data sets By working with the commercial groups, it’s trying to understand what are those other opportunities So we’re doing work on the Earth side, doing stories in voyage– you probably saw some of those last night– a lot of work on the Earth Engine side now, which is the data sets But there’s programs like SERVIR that don’t have to go through this process of a large scale partnership that can do very specific program related things, so DAVID SAAH: Thank you JAMIE FAVORS: Thank you DAVID SAAH: How many stickers do you want? EMIL CHERRINGTON: [INAUDIBLE] not planted question Could we maybe put Tyler on the spot from the Earth Engine team? And again, thanks Everybody– big, big round of applause for Tyler and the great work that their team is doing [APPLAUSE] Could you maybe tell us a little bit about your perspective on, again, working with diverse user groups like the SERVIR hubs? TYLER ERICKSON: Yeah, I guess I could I mean, we’re very excited to have a group like SERVIR making such good use of the tools that we’ve developed, because our reason for building them in the first place is trying to enable this type of impact in the world And I find that it has a sort of a multiplicative effect If you find a group that works together that can support each other from a programming standpoint, they can move much faster than just individuals that are maybe doing this research in academia, founded on their own But they’re working alone So I think that is what I’ve been really impressed with, with the SERVIR group, is just seeing that there is so much collaboration going on within the group that they’re really willing to share outside as well And I think that’s why they’ve been able to make such rapid progress So we’re really happy, actually, with what we see come out of SERVIR [LAUGHTER] DAVID SAAH: What other surprises do you got for us, Emil? No, that’s good? All right, are there any other general questions? I really want to break you guys out so you can actually start digging in and getting stuff That’s what I’m really excited about [APPLAUSE]