The National Geographic Society, working with partners at Google and World Resources Institute, is building a living map of the world.
National Geographic is calling on the community of land cover and photo interpretation experts by September to help annotate and curate Sentinel-2 satellite imagery needed by machine learning algorithms. Experts who can spend 20-40 hours on this task in the next 6-8 weeks should send their resumes to [email protected].
“Our vision is to produce the world’s first open global time series map of land cover and land use at 10-meter resolution with annual updates using public satellite imagery,” the society said in a statement.
A living map of the world is a foundational dataset for knowledge products driving understanding and forecasting of the world as a system and enabling data-driven conservation, resource management and policy making for sustainable development.
Simultaneous advances in global satellite imagery, super-computing on demand in commercial cloud, and powerful open source machine learning algorithms in high-performance software frameworks, combine to enable production of a global time series map of land cover and land use at a scale, speed and cost that is within reach for large NGOs and global governments.
The major roadblock to production of a global time series map is availability of a large quantity of high-quality annotated data (hundreds of millions of labeled pixels) required to train algorithms to automate production of the map time series.
National Geographic is aiming to create an initial training dataset of densely annotated tiles of Sentinel-2 imagery before September, following an expert-defined land cover taxonomy. This expert-labeled tile set will be used to train a large non-expert crowd to produce tens of thousands of additional labeled scenes, which will then be used to train the machine learning algorithms that produce maps.
A major use of remote sensing data is to compare images of an area taken at different times and identify the changes it underwent. With a wealth of long-term satellite imagery in open use, detecting such changes manually would be time-consuming and most likely inaccurate.
To address this, EOS Data Analytics has introduced an automated Change Detection tool to its flagship product LandViewer, a cloud tool for satellite imagery search and analysis in today’s market.
Unlike the methods involving neural networks that identify changes in the previously extracted features, the change detection algorithm implemented by EOS is using a pixel-based strategy, meaning that changes between two raster multi-band images are mathematically calculated by subtracting the pixel values for one date from the pixel values of the same coordinates for another date.
This new signature feature is designed to automate a change detection task and deliver accurate results in fewer steps and in a fraction of the time needed for change detection in most image-processing software.
Change detection interface: Images of Beirut city coastline selected for tracing the developments of the past years. (Image: LandViewer)Change detection interface: Images of Beirut city coastline selected for tracing the developments of the past years. (Image: LandViewer)
Applications from farming to environmental monitoring
One of the main goals set by EOS team was to make the complex process of change detection in remote sensing data equally accessible and easy for non-expert users coming from non-GIS industries.
With Land Viewer’s change detection tool, farmers can quickly identify the areas on their fields that were damaged by hail, storm or flooding. In forest management, satellite image detection of changes will come in handy for estimation of the burned areas following the wildfire and spotting the illegal logging or encroachment on forest lands.
Observing the rate and extent of climate changes occurring to the planet (such as polar ice melt, air and water pollution, natural habitat loss due to urban expansion) is an ongoing task of environmental scientists, who may now have it done online in a matter of minutes. By studying the differences between the past and present using the change detection tool and years of satellite data in Land Viewer, all these industries can also forecast future changes.
Top change detection use cases: Flood damage and deforestation
A picture is worth a thousand words, and the capabilities of satellite image change detection in Land Viewer can be best demonstrated on real-life examples.
Forests that still cover around a third of the world’s area are disappearing at an alarming rate, mostly due to human activities such as farming, mining, grazing of livestock, logging, and also the natural factors like wildfires. Instead of massive ground surveying of thousands of forest acres, a forestry technician can regularly monitor the forest safety with a pair of satellite images and the automated change detection based on NDVI (Normalized Difference Vegetation Index).
How does it work? NDVI is a known means of determining vegetation health. By comparing the satellite image of the intact forest with the recent one acquired after the trees were cut down, Land Viewer will detect the changes and generate a difference image highlighting the deforestation spots, which can further be downloaded by users in JPG, PNG or TIFF format. The surviving forest cover will have positive values, while the cleared areas will have negative ones and be shown in red hues indicating there’s no vegetation present.
A difference image showing the extent of deforestation in Madagascar between 2016 and 2018; generated from two Sentinel-2 satellite images. (Image: LandViewer)
Another widespread use case for change detection would be agricultural flood damage assessment, which is of most interest to crop growers and insurance companies. Whenever flooding has taken a heavy toll on your harvest, the damage can be quickly mapped and measured with the help of NDWI-based change detection algorithms.
Results of Sentinel-2 scene change detection: The red and orange areas represent the flooded part of the field,; the surrounding fields are green, meaning they avoided the damage. California flooding, February 2017. (Image: LandViewer)
How to run change detection in Land Viewer
There are two ways you can launch the tool and start finding differences on multi-temporal satellite images: by clicking the right menu icon “Analysis tools” or from the Comparison slider ‒ whichever is more convenient. Currently, change detection is performed on optical (passive) satellite data only; addition of the algorithms for active remote sensing data is scheduled for future updates.
Esri is releasing Sentinel-2 Image Services to all Esri users for no additional cost.
According to the company, Sentinel-2 is an Earth Observation Satellite that provides multi-spectral imagery for any location in the world at 10-meter resolution. Currently in beta, the service is updated daily with new imagery for all ground locations every five to seven days.
The Sentinel-2 Image Services provide temporal, multi-spectral imagery of the entire globe for improved monitoring of agricultural and forest conditions, monitoring of land cover changes, and to assist with natural disaster management.
Sentinel-2 is part of Copernicus, the world’s largest single Earth observation program directed by the European Commission in partnership with the European Space Agency.
Esri makes the multi-spectral data quickly accessible using ArcGIS Image Server and publishes an image service through the ArcGIS Living Atlas of the World, hosted on the Amazon Web Services Infrastructure.
The service includes all Sentinel-2 imagery going back 14 months, enabling change to be easily reviewed. Image analysis can be run directly on the service to create indexes displaying properties such as vegetation health or soil moisture as well as quantifying the changes over time, for better understanding of the environment.
“We are committed to helping our users discover, explore, and better understand our changing planet,” said Jack Dangermond, Esri founder and president. “Pairing Sentinel-2 imagery with our ArcGIS Image Server provides a powerful platform for in-depth analysis which can inform meaningful action.”
Sentinel-2 multi-spectral imagery can provide better visualization and understanding of catastrophic events such as Hawaii’s Kilauea volcano, the company said. The ability to use imagery of the volcano along with other spatial data, such as digital elevation models, provides an unprecedented opportunity to help predict lava flow direction and provide advanced notice to those who may be in danger.
Sentinel-2 can also help provide understanding of the conditions that lead to fires such as this past winter’s Thomas Fire, which is California’s largest wildfire on record. The Thomas Fire burned more than 280,000 acres and triggered massive mudslides. Visualizing factors, such as periods of increased moisture contributing to more lush vegetation followed by hot and dry weather, can help predict future wildfires and mitigate their effects in the future.
The Sentinel-2 imagery is available through the Living Atlas, the foremost collection of geographic information from around the globe. The Living Atlas is included with all ArcGIS online subscriptions. It is comprised of maps, apps, and data layers that support the work of thousands of Esri users around the world. Full service access, including a rolling 14-month archive of the Sentinel-2 data, is now available to all Esri ArcGIS users.
The National Geographic Society, working with partners at Google and World Resources Institute, is building a living map of the world.
