Sensors and Systems
Breaking News
Maxar Secures NOAA Approval to Provide Non-Earth Imaging Services to Government and Commercial Customers
Rating12345WESTMINSTER, Colo.- Maxar Technologies (NYSE:MAXR) (TSX:MAXR), provider of comprehensive...
AEye Announces Groundbreaking Immersive Lidar Experience for Attendees at CES 2023
Rating12345DUBLIN, Calif.- AEye, Inc. (NASDAQ: LIDR), a global leader in...
WIMI Hologram Academy: Multi-Dimensional Holographic Vision Opens A New Chapter In Cyberspace Mapping
Rating12345HONG KONG – WIMI Hologram Academy, working in partnership...
  • Rating12345

Moore_Rebecca_thumbGoogle.org unveiled the Earth Engine concept at the COP15 climate talks in Copenhagen, with a focus on monitoring tropical forests to assess their ecosystem services. V1 editor Matt Ball spoke with Rebecca Moore, the Engineering manager of Earth Engine and Google Earth Outreach, about the design and functionality of Earth Engine, and the overall aims of this mission.

Moore_Rebecca

Google.org unveiled the Earth Engine concept at the COP15 climate talks in Copenhagen, with a focus on monitoring tropical forests to assess their ecosystem services. V1 editor Matt Ball spoke with Rebecca Moore, the Engineering manager of Earth Engine and Google Earth Outreach, about the design and functionality of Earth Engine, and the overall aims of this mission.

V1: I think it would be interesting to start with a little history, because I know that you’ve been working on Google Earth Outreach for far longer than this recent effort on the Earth Engine. Could you fill us in on how this all got started?

Moore:
Google Earth Outreach started for me as a Google 20% project in September of 2005, right when I joined Google. Every engineer at Google can spend one day a week, or 20% of your time working on something outside of your formal job responsibilities that you’re passionate about, as long as it can be tied to Google’s mission to organize the world’s information and make it universally available.

I was the technical lead for the Google Earth Layers on the Google Earth engineering team at the time, and also an ardent environmentalist. I was at that time using Google Earth successfully to raise awareness in my Santa Cruz Mountain community about a proposed plan to log a thousand acres of redwoods in a local canyon.

Through creation of an animated 3D “virtual flyover” of the logging plan  in Google Earth, I was able eventually to show that the plan was not legal and we stopped the logging. That success of using Google Earth for grassroots environmental advocacy caught the attention of environmental groups around the world who began to contact me asking for advice.

Many of these groups, such as the Sierra Club, Greenpeace, etc., had thought there might a way to use Google Earth for environmental protection, but they hadn’t yet been able to figure out how to do it.   Keep in mind that Google Earth had only just been released, a few months earlier in June of 2005…and most people’s experience was limited to flying to their homes.  At roughly that same time in September of 2005, Google Earth was used on the ground when Hurricane Katrina hit to save the lives of more than 4,000 people who were trapped on their roofs by the flooding

It became very clear within these two applications, environmental and humanitarian, that Google Earth had the potential to be a significant and meaningful tool for public benefit.  So that’s when I launched Google Earth Outreach with the help of other  Googlers willing to contribute their 20% time to work with non profits to help them understand how to use our mapping tools (Google Earth and Google Maps) for their projects.

V1: What was the response that you received from the non-profit community?

Moore: We worked with more than 100 different non-profit organizations over the next 18 months, from tiny grass roots NGO’s like Appalachian Voices who wanted to highlight mountain top removal coal mining all the way up to the United Nations Environmental Program (UNEP) who wanted to animate their coffee table book called “Atlas of Our Changing Environment.”

The book highlights more than 100 locations around the world that have undergone extreme environmental degradation over the last 30 years. They had Landsat images from 30 years ago and today, showing receding glaciers, desertification, tropical deforestation, and these sorts of extreme changes. They wanted to bring that online to users on the Internet, and we have more than half a billion users of Google Earth around the world, in more than 40 languages today. It provided an unprecedented way for UNEP to make their atlas more broadly available.

In working with these groups we developed tools and resources that we published online. In June 2007, we took the lessons learned from consulting with these organization and created a set of tutorials, online video, case studies and a showcase gallery of public benefit uses of Google Earth and Google Maps for the Google Earth Outreach site for the United States. We also give grants of Google Earth Pro, which normally costs $400, and we give that away for free.

We subsequently launched Google Earth Outreach in Europe and Brazil, and then last October, we launched Google Earth Outreach in Africa. In each of these cases, my team and I went to those countries, gave workshops to more than a thousand NGOs in these different countries, we translated our Google Earth Outreach tutorials into the local language, and we worked with local NGO’s in those countries to create showcase examples in advance of the launch to show the use of mapping for public benefit.

V1: Do you have a favorite showcase example?

Moore: We worked with the Amazon Surui tribe in Brazil, who are really a model of a tribe that  has been fighting to protect and preserve their culture despite decimation from disease and other problems introduced after first contact with the modern world in 1969. They were for a time successful protecting their territory using bow and arrow, but eventually reached the limits of that.

Chief Almir, their young elected chief, was the first member of their tribe to attend university. While he was there he tried Google Earth in an Internet Cafe, and like the rest of us the first thing that he did was fly to his home. It’s quite striking what he saw, and what anyone can see, you see a verdant green island of healthy rainforest that is surrounded by complete devastation of clear cut rainforest. You can even see the illegal logging incursions into their land in the satellite imagery.

He told me that when he saw that he had this “aha moment” that if the world could see this that it would help mobilize support for better protection of their land. He said the time had come to put down the bow and arrow and pick up the laptop. He came to Google in 2007, and asked for our help to train his people on how to use our mapping tools to create the first rich indigenous cultural map in Google Earth.

If you look at the Amazon today in Google Earth, it’s almost blank. If you go to almost any populated urban area in the world, you’ll see that we have lots of rich content, with roads, place names, hotels, and YouTube videos and photographs contributed by our users. But if you go to the Amazon it’s pretty empty, and he said that noone would know that any people live there.

We trained his people on how to put their cultural information on Google Earth as a way to preserve their history and culture and to showcase the projects that they’re doing to create a sustainable future for the tribe. For example, the chief has organized a project to replant 7,000 hectares of land that had been illegally logged. Through this project he hopes to gain access to the carbon offset marketplace. It’s called the Surui Carbon Project, and it looks like it will be a model for indigenous tribes throughout the Amazon, and potentially throughout the world.

We did the original training in 2008, and then went back again in September of 2009 to train them on how to use Google Android smart phones with opensource software called Open Data Kit that allows them to do field collection of data. They’re using this to build a system for reporting and alerting when illegal logging is happening. You can take a video that is time and date stamped, GPS located, and can immediately go online in Google Earth and beam that around the world. They hope to use this to create a global community watch program and to support stronger law enforcement. 

To learn more, see “Trading Bows and Arrows for Laptops” on YouTube.

V1: Where did the idea of Earth Engine come from?

Moore: While we were in Brazil in June 2008 launching the Google Earth Outreach program, I was approached by Brazilian scientists and other non-profit representatives at the workshops that we gave in the capital of Brasilia. They said that Google Earth Outreach was great, but what they really need Google to do is help them monitor deforestation in the Amazon. As I look back, this is where Earth Engine was born.

Tropical deforestation is happening at a rather alarming rate, and I believe more than a million acres a year are lost in the Amazon. From a climate perspective, deforestation is said to account for between 14 and 20 percent of greenhouse gas emissions. More than all the transportation in the world put together (cars, planes, trucks). That is because when you cut the trees, they are no longer able to absorb carbon dioxide and sequester that carbon. And then following that cutting the area is burned to convert to farmland or ranch land and that releases enormous amounts of so-called black carbon that causes a huge amount of greenhouse gas emissions.

At the Copenhagen Climate Change Conference, and prior to that, many nations are coming together to work toward a proposal for reducing tropical deforestation. It’s considered the low-hanging fruit of mitigating climate change, because all we have to do is stop cutting the forest. It’s not complicated.

The problem however is how does one effect a change like this. The reason the forest is being cut is that it provides an income in those areas. To offset that there needs to be some sort of financial mechanism to compensate those who would otherwise cut the forest. There are a number of systems to create a voluntary marketplace, and there is discussion of the provision called REDD (reduction of emissions from degradation and deforestation) whereby wealthy nations would pay the tropical forest nations not to cut their forests down.

For such a system to work, it will be necessary for these tropical nations to reliably monitor their forests, measure changes in those forests, and report that in a way that is open, transparent and verifiable. If you look at the technical capacity of the rainforest nations today, it’s fairly low. A representative from one of those countries told me that their computers are 12 years old, and that they don’t have the ability in-country to do this.

Back to the request that came through our Google Outreach launch in Brasilia. Dr. Carlos Souza, who’s a leading remote sensing scientist with Imazon, said they have a fully operational system that provides monthly reports on new hot spots of deforestation that is detected using MODIS data. He said that what they don’t have is the technical capacity to do the level of monitoring and the resolution of monitoring that they would like as frequently as they would like, and with the timeliness that they would like.  Also other tropical rainforest nations would like to run the IMAZON software, but don’t have a way to do this today.

To monitor change in the Amazon requires many terabytes of satellite imagery data. The software analysis can take months to run on a conventional computer.  So to do this more effectively takes a lot of computational horse power and storage for the massive amounts of data, and these institutions don’t have this capacity. The request came to see if Google would be willing to create a platform that would host all of the world’s remote sensing data (satellite imagery and earth observation data) and provide access to a high-performance imagery processing engine online in the Google cloud to run these types of analysis.

V1: So the Earth Engine is not just a repository, but the means to analyze as well?

Moore: Yes. The idea will be to ultimately provide for public benefit an online repository that brings together all of the Earth’s observation data (satellite imagery, terrain datasets, vector data such as roads, borders, population centers, soil information, climate information) into one large georeferenced data store. And then to provide through an easy to use application programming framework access to our computational resources for analyzing that data.

We see this as an unprecedented platform for data-mining meaningful information out of this treasure trove of historical, current and future earth observation data. Ultimately it will be many petabytes of earth observation information.

V1: In terms of the kinds of remote sensing data that will be housed in this repository are their partnerships with the various agencies around the world?

Moore: We’re working with the Group on Earth Observations (GEO) an international consortium of all of the world’s space agencies (NASA, JAXA in Japan, ESA in Europe, INPE in Brazil, and many others). Their goal is to support public benefit applications of earth observation data, and to collaborate and coordinate their data collection and their data access policies to support better public utilization of earth observation data.

Our platform will be a way to showcase and make available really for the first time all of this data in one place in a way that the data is co-georegistered so that you will be able to leverage analysis on multiple datasets simultaneously.

The first use case or application area is forest monitoring, because of the current level of interest and need in a global platform to monitor changes in the earth’s forests over time.

Quite an important point is that Google will not by and large be creating the scientific algorithms that will do the monitoring. We will be providing this engine that provides easy access to the data, and to a high performance computational platform. Tens of thousands of computers in our data center will potentially be made available for analysis, and through an open API (programming environment). It will be up to the remote sensing scientists of the world to develop the algorithms.

Our hope is that this will make remote sensing much easier than it is today. That it will do for remote sensing in a way what Google Earth did for geo-literacy among the general population.

If a university student wants to try out a new algorithm, this platform should allow them to do that within seconds or minutes rather than laboriously gathering data, doing orthorectification and cloud correction and all these sorts of things. We’ll take care of the preprocessing of the data to get it to a standard form that is immediately ready for advanced analysis.

V1: Where does development of Earth Engine stand right now?

Moore: We announced Earth Engine in Copenhagen in December. We demonstrated a prototype that is not yet released to the public. The prototype we developed in close cooperation with Greg Asner of the Carnegie Institution for Science and Carlos Souza of Imazon. We wanted to work with a couple of leading scientists in forest monitoring to understand what their requirements are for such a platform.

It’s been a great honor to work with both of them and to learn deeply what kind of services we need to offer. What we built and demonstrated in Copenhagen included a subset of Landsat and MODIS data and an implementation on this platform of Carnegie Landsat Analysis System (CLASLite) by Greg Asner and Systema de Alerta de Desmatamento (SAD) by Carlos Souza.

Now we’re working to create a fully operational platform that we would launch this year. We’ve announced that we will make this available to the world for public benefit prior to COP 16, which is the next international climate change conference that will take place in Mexico City. We will make Earth Engine available to the rainforest nations in advance of COP 16, and our goal is to now develop the core platform that will host data and a processing framework an Earth Engine API so that scientists can begin to get access to the data and our platform. We expect to roll this out to trusted testers, and then more broadly as we get experience with it.

The way we’re designing Earth Engine is not limited to forest and landcover types of analysis. We’re building a general-purpose remote sensing and processing engine that could be used for other applications as well, such as water resource mapping or humanitarian crisis mapping, and there are many ocean and marine applications. We see those capabilities coming online over time.

V1: Will there be tools for collaboration, say for scientists working in the same region to share their results and their research objectives?

Moore: There are a couple of ways that I think we’ll see Earth Engine be utilized. One is it will be under the covers of applications developed by scientists, and those may be found on their own platforms or systems, which are independent from any collaborative platform. But the other thing that you’ll see is that we’ll offer a platform that will include a data download service, which I haven’t mentioned yet. We’ve been asked to aggregate all the data in one place, and create these much more useful higher-level data products such as already orthorectified, georegistered to a common DEM, atmospherically corrected data sets. We would make those data sets available for download so that remote sensing scientists could use them in their own desktop programs.

I can also imagine, although this is still very open for feedback from the remote sensing community. We’re still very much in the early stages of what Earth Engine will look like. We have been asked by a number of scientists to create the kind of collaborative environment that you’re talking about. One scientist may have state-of-the-art algorithm that takes the raw imagery and produces a time series of forest cover change over time. Another scientist might want to take that input into their own algorithm that estimates biomass changes or carbon changes. Another scientist might want to take these along with input such as vector and environmental factors and model potential future landcover changes based upon past time series analysis and these environmental factor vector layers.

The scientific community that we’re already working with are already talking to one another about how they can use this platform for collaboration. The big advantage is that all of the data resides in a common and easily accessible format in the cloud, and with the easy access to the analytical processing and the creation of these derived mapped products, it should be a logical next step to create the type of collaboration environment that you’re talking about.

We want to be cautious that scientists have control that they need over how their tools are used, and how their derived products are used. I think we will not be launching this sort of collaborative environment in our first release. We need time to work with the community and understand what the needs are, and what types of access control, security, collaborative capabilities make sense.

V1: The COP 16 event is scheduled for late November or early December next year in Mexico City. What is the timeframe for rolling out the rainforest monitoring capability?

Moore: We will launch this platform publicly this year in advance of COP 16. Synchronous with the launch of the platform, that has access to data, a data download service, and the computational resources, there will be one or more computational algorithms of remote sensing scientists that run on Earth Engine related to forest monitoring, reporting and verification.

We’re at the very beginning of this project, this technology and this platform, but it’s very exciting. I also want to make it very clear that this is a project of Google.org. Google.org is sponsoring all of my team’s engineering development and the hosting of the data and the costs we incur with the computational resources, which are not inconsiderable. And making this available as a not-for-profit service.

I’d like to acknowledge our close partners, the Gordon and Betty Moore Foundation. The forest monitoring application has been their focus and they’ve helped us develop our strategy and supported many of the  remote sensing scientists and tropical nations that we’re working with.

V1: I’m really excited about this, and it really seems to outline the vision of a digital earth that’s been out there for some time. I’m wondering if Al Gore is aligned with this idea, and if there was any input of his on this effort.

Moore: Al Gore is a senior advisor to Google as you may know. He is definitely an inspiration to all of us for the vision he articulated about a digital earth, and he is aware of the work we are doing on Earth Engine, and is very supportive.

V1: That speech is one that I come back to year after year for its clear vision of what a digital earth could become. It has inspired me, and certainly a lot of people in the industry. It’s interesting to see some of the pieces fall into place, to see the computational power, the software as a service approach, and the computing in the cloud.

Moore: Yes. I do want to make one thing clear as there could be a potential for misunderstanding. This is a platform that we’re putting in the hands of the world for scientists and students and developers to create these new applications. To re-implement existing applications and allow them to run at a scale and speed over data sets that has never been technically feasible before. And also to create new applications that perhaps weren’t possible before. And then to put these applications in the hands of developing world nations, NGO’s,  indigenous peoples like the Surui. To empower them to do their own forest monitoring in the cloud. We at Google will not be the ones operating this platform or the applications that are doing the monitoring. We will be providing the technology to do this in collaboration with those scientists and remote sensing application developers who chose to make their applications available.

V1: This approach seems to support the whole move toward data-driven science.

Moore: Some model-driven approaches have arisen only because the scientists haven’t had access to the data. With this kind of access to data you open up whole new possibilities.

I liken it to sitting in a room with the blinds drawn and wondering if it’s raining outside. I can build a model that analyzes 20 years of climate change data or I can just open the blinds and look outside. This level of data will be like opening the blinds.

Leave a Reply

Your email address will not be published. Required fields are marked *