2. What are Earth observations?
Earth observations are data and
information collected about our
planet, whether atmospheric, oceanic
or terrestrial.
This includes space-based or remotely-
sensed data, as well as ground-based
or in situ data.
Earth Observations
3. Assessing climate change impact
Older, thicker Arctic sea ice declines, September 1984 - September 2016
Source: NASA Images of Change
Earth Observations for Temporal Monitoring
4. Tracking urbanization
Urban expansion in New Delhi, India, March 14, 1991 - March 2, 2016
Source: NASA Images of Change
Earth Observations for Temporal Monitoring
5. Flood monitoring
Storms swell Ceder River, Iowa, July 8 2016 - September 26, 2016
Source: NASA Images of Change
Earth Observations for Temporal Monitoring
6. Wildfire monitoring
Wildfires near Fort McMurray in Alberta, Canada, April 17, 2016 - May 3, 2016
Source: NASA Images of Change
Earth Observations for Temporal Monitoring
7. Drought and water resources management
Drying Lake Poopó, Bolivia, April 12, 2013 - January 15, 2016
Source: NASA Images of Change
Earth Observations for Temporal Monitoring
8. What is GEO?
GEO is an intergovernmental
partnership that improves the
availability, access, understanding
and use of Earth observations for
the benefit of society.
Group on Earth Observations
9. Focus areas are the UN 2030 Agenda for Sustainable
Development, the Paris Agreement on Climate and
the Sendai Framework for Disaster Risk Reduction.
Group on Earth Observations
GEO Engagement Priorities
10. The big questions:
1. What are we going to do
with these innovations?
2. How are we going to use disruptive
technologies to build better societies?
Blockchain+AI+Human: Whitepaper (A. Pentland, J. Werner, C. Bishop)
13. Artificial Intelligence for Earth observations
Supervised AI and Machine Learning tools are
already being used across the GEO community to
automate analysis of big data, but this can only be
built on solid human and open data foundations.
Example: An study by Planet used crowdsourced building footprints
and height information to train an ML model to be used for urban
growth monitoring in Dar es Salaam.
14. Artificial Intelligence for Earth observations
https://www.gfdrr.org/en/publica
tion/machine-learning-disaster-
risk-management
15. GEO-AMAZON WEB SERVICES –
EARTH OBSERVATION CLOUD CREDITS PROGRAMME
Helping countries realize the potential of
Earth observations for sustainable
development.
Application deadline: 31 March 2019
www.earthobservations.org/aws.php
Non-commercial projects in developing countries that support
environmental and international development goals can apply
for up to $100,000 of AWS cloud credits to help host, process
and analyse large geospatial data sets.
16. GEO – UN ENVIRONMENT
WORKING GROUP ON NEW TECHNOLOGIES
Making best use of global environmental
data and information
GEO and UN Environment work with countries around the world on
the management and application of global environmental data. As
such, the partnership is working towards building and governing a
global ecosystem of environmental data and analytics. This is
relevant for new and emerging technologies, such as blockchain
and AI, as well as insights for evidence informed policy.
17. Digital Earth Africa will build on [Data Cube] technology
developed by Geoscience Australia and initial efforts
through the Africa Regional Data Cube and CEOS to deliver
a unique continental-scale platform that democratizes
capacity to process and analyze satellite data.
It will track changes across Africa in unprecedented detail,
and provide data on a vast number of issues, including soil
and coastal erosion, forest and desert development, water
quality and changes to human settlements.
DE Africa currently supported by Kenya, Ghana and South Africa
The Global Risks Landscape 2016 highlighted the failure of climate change mitigation and adaptation, extreme weather events, water crises, as well as biodiversity loss and ecosystem collapse as the top 10 risks in terms of likelihood and impact. As we know these events are all interconnected, for example an increase in flooding or drought leads to food crises.
Geography plays a key role in underpinning the decisions and actions taken. In this short talk I will discuss how Earth observations provide Insights for a changing world.
However, it is important to note that Earth observations add most value when integrated with other socioeconomic data, for example relating to population growth.
In terms of Earth observations we talk about the ability to measure and monitor anything in, on or around the Earth. Many of you will recognize that space-based technology is used for looking at the planet from space, it’s also important that we acknowledge the importance and need for in situ observations, such as those measuring ocean salinity, rainfall or crop conditions.
This is a NASA image, but Sentinel 1 from Copernicus is also very useful for flood monitoring.
If you watched any news in the last 18 months you would be aware of the water shortage situation in South Africa. This happening in other parts of the world, including other areas in Africa, Australia and South America. This is where policy decisions informed by Earth observations can make an impact.
The Group on Earth Observations has been in existence for more than a decade to support countries around the world to make decisions relating to societal benefit areas, including agriculture, biodiversity, climate, disasters, energy, forestry, oceans and others. Our goal is to help national governments understand the science and technology, as well as the value of open Earth observations data and information for policy, decision making and action. We are lowering the barrier to entry for national governments while working on institutional strengthening.
With more than 70 activities across numerous societal benefit areas, there are three engagement priority areas.
Looking at how EO intersects each area and can inform respective policy actions is a key element of our work.
In the MIT whitepaper that was shared with all of us, there were numerous questions, but these were two of the main ones.
Ensuring that data is accredited and has not been manipulated is critical if policy decisions are being based on them.
Blockchain can create trust in the provenance and integrity of satellite images, individual measurements and in situ data feeds.
Example: Index-based farm insurance requires crop health measurements to be trusted by both the farmers and the insurers.
The emergence of index based insurance where a farmer could take out insurance that when crop greenness falls below a certain threshold, the insurance company pays out requires the measurement to be trusted by both parties and it must be verifiable that it has not been manipulated by either party.
http://www.earthobservations.org/geo_blog_obs.php?id=322
EO imagery will provide key information to diverse applications, helping to reinforce trust between the involved stakeholders (by providing evidence). It could also validate the conditions of execution of smart contracts. In the insurance sector, EO imagery could for instance be used to assess the damage from floods and fires and then to automatically trigger claims processing.
EO data could provide meteorological parameters relevant to renewable solar, wind, water and other energy production. This could improve predictions of production and enable operators to adapt upcoming transactions.
AI and Machine Learning tools are already being used across the GEO community to automate analysis of big data, but this can only be built on solid human and open data foundations. In the MIT paper it talks about yottabytes of data and this is becoming increasingly difficult for humans to manage, so machines can help enormously. However, lets not forget the data drought, as well as the data deluge. Of the 93 environmental SDG indicators only 34 have agreed methodologies and enough data from most countries (per UN Environment GEO 6).
The sheer volume of data generated by Earth-observing systems often makes it impractical for humans alone to perform the analysis, so many groups are turning to AI and ML algorithms to support their analysis. Development Seed, EOS, ESA, NASA, World Bank and other GEO partners are using AI and ML to unlock the power of this planetary-scale data that is becoming increasingly more accessible in the cloud, from open-source libraries and human-in-the loop initial processing passes, to fully automated pipelines.
The GEO community is working to build capacity of countries and institutions to take advantage of and tailor these technologies to their needs, in support of sustainable development.
Example:
Study by Planet: Monitoring urban growth through floor space index
Regular, cloud-free satellite images combined with ML algorithms can be used to monitor horizontal and vertical urban growth. This study uses crowdsourced building footprints and height information to train an ML model to be used for urban monitoring in Dar es Salaam.
Reference data: OpenStreetMap building footprints and height attribute collected during the Ramani Huria project
Algorithms: Deep learning, convolutional neural networks
Results:
• The study shows how to combine OSM reference data and machine learning methods.
• Building footprints were extracted to an accuracy of 77%; the correct number of floors predicted for 23% of the buildings.
• Difficulties were caused by densely built (informal) areas.
This report provides a useful overview of ML, terminology, methods, shortfalls and other considerations relating to Machine Learning. In this case it is applied to disaster risk management, but some of the concepts are broadly applicable in numerous other thematic areas.