How AI Can Help Cities Conserve Resources and Plan for the Future
The world is changing swiftly, and nowhere are those changes felt more acutely than in our cities. Over half of the world’s population lives in urban areas, a trend only expected to accelerate in the coming decades.
The rapid growth of urban populations is concentrating environmental stress in key regions of the world, increasing demand for resources such as reliable energy and clean water, and pushing urban growth boundaries farther and farther into surrounding forests and farmlands. This tension between our need for urban development and our responsibility to the environment is strained further by factors like global climate change and increasingly severe weather events, which have devastated both our cities and the natural ecosystems that sustain them. Together, these factors present a complex challenge for city managers who need tools that can help them better understand, address and even predict environmental risks to their communities.
Fortunately, there are technology trends emerging today that hold tremendous promise for helping cities address these issues. Advances in artificial intelligence (AI) have the potential to exponentially increase our ability to collect and analyze enormous amounts of data on the state of our planet and empower land use planners with the insights and tools they need to build a more resilient and sustainable future.
One of the many challenges that urban planners face is a lack of current, detailed and easily accessible data about how their municipalities are changing. Land cover maps can help by showing what is where in a particular landscape, and how fast it is changing over time—visualizing everything from natural phenomena (like forests, rivers, lakes, and prairies) to human-made surfaces (such as cities, farms, buildings, roads, and parking lots). The better the picture of our lands, the more informed policymakers and communities will be as they make critical decisions for the future, such as how to grow sustainably, how to manage their natural resources effectively, and where to implement critical environmental protections and interventions like green infrastructure.
In the video below, from a demonstration at this year’s Esri User Conference, I explain how AI and deep learning can be used to develop enhanced mapping software for more sustainable land management.
Unfortunately, many local decision-makers do not have easy access to accurate and timely data. The most accessible land cover map for the United States is currently produced at 30 meter resolution, and is now 5 years out of date. Products like these are useful when asking very broad questions at a national scale, but the lack of spatial and temporal resolution minimizes the utility of these data for city-level planning. Other free, higher resolution imagery, such as the National Agriculture Imagery Program’s (NAIP) 1 meter-resolution data collected annually for the United States government, provides the detail required to spot trends and manage lands with a degree of resolution relevant to urban planning. However, analyzing these increasingly data-dense images to identify every new road, house, and other land use using standard approaches is a resource-intensive process. For example, a land cover map of the entire United States would require training an algorithm to recognize a large array of land uses across a wide variety of geographies and analyzing nearly 10 trillion pixels all in one go, a process requiring access to a level of computing resources and technical expertise that is far beyond what most nonprofits and government agencies can, or should, afford.
AI for Urban Sustainability
Fortunately, land cover mapping is an application perfectly suited for modern AI. Over the past year, Microsoft has partnered with the Chesapeake Conservancy, and the geospatial mapping and technology company Esri, to improve access to high resolution land cover maps across the United States. We started with data created by the Chesapeake Conservancy, who worked with Microsoft and ESRI technologies, along with partners throughout the Chesapeake Bay watershed, to map the roughly 100,000 square miles of land in the region. The new map provides 900 times the amount of information as conventional 30-meter resolution land cover data and enables land managers to see specific details like buildings or trees within larger plots of development or forest land, and pinpoint exactly where the bay’s ecosystems are becoming fragmented or damaged.
Now, we are using those data to train a single, powerful AI algorithm for mapping land cover at 1 meter resolution across the United States, with the ambition to enable on the fly land cover mapping through a simple user interface. Early results are promising, and should we succeed, land use managers at organizations of all sizes will be able to develop accurate, actionable, and sharable insights about what is happening on their lands more quickly than ever before.
In the video below, you can see how the Chesapeake Land Cover Mapping Project is already being used to help those that live and work on the Chesapeake Bay protect and restore their watershed.
Microsoft’s AI for Earth Initiative
This project showed us just how effective AI can be in solving pressing environmental and resource challenges facing cities. Land cover mapping helps understand the physical landscape of the world around us, which is a prerequisite to then make smarter decisions about overall resource management, smart transportation and grid solutions, and a host of other areas critical to building more sustainable cities.
That’s where Microsoft comes in, through existing programs like CityNext, focused on cities, and a newer program focused on AI applications for sustainability called AI for Earth. AI for Earth is built to help address some of the most significant barriers to harnessing AI for sustainability, and provides mechanisms to overcome them. These include access and education, so we have allocated $2 million to grants, trainings and projects to showcase the transformative potential of Microsoft’s AI technologies.
Ultimately, Microsoft’s AI for Earth program exists because of what cities around the world already know: the scale and speed of the changes we see in our physical and natural world require new solutions, and quickly. As Microsoft’s President and Chief Legal Officer Brad Smith wrote at the launch of AI for Earth, “There is no shortage of work to be done to chart a healthier, more sustainable future for our planet. Time is not our friend on this endeavor, and the stakes are high.” By working with organizations of all kinds, we hope to empower communities everywhere with the insights they need to solve today’s complex environmental challenges, and build a more resilient and sustainable future.
To learn more, please visit: https://www.microsoft.com/en-us/aiforearth
Leave your comment below, or reply to others.
Please note that this comment section is for thoughtful, on-topic discussions. Admin approval is required for all comments. Your comment may be edited if it contains grammatical errors. Low effort, self-promotional, or impolite comments will be deleted.
Read more from MeetingoftheMinds.org
Spotlighting innovations in urban sustainability and connected technology
Social distancing is becoming the new normal, at least for those of us who are heeding the Center for Disease Control’s warnings and guidelines. But if you don’t have reliable, high-speed broadband, it is impossible to engage in what is now the world’s largest telecommunity. As many schools and universities around the world (including those of my kids) are shut down, these institutions are optimistically converting to online and digital learning. However, with our current broadband layout, this movement will certainly leave many Americans behind.
Accenture analysts recently released a report calling for cities to take the lead in creating coordinated, “orchestrated” mobility ecosystems. Limiting shared services to routes that connect people with mass transit would be one way to deploy human-driven services now and to prepare for driverless service in the future. Services and schedules can be linked at the backend, and operators can, for example, automatically send more shared vehicles to a train station when the train has more passengers than usual, or tell the shared vehicles to wait for a train that is running late.
Managing urban congestion and mobility comes down to the matter of managing space. Cities are characterized by defined and restricted residential, commercial, and transportation spaces. Private autos are the most inefficient use of transportation space, and mass transit represents the most efficient use of transportation space. Getting more people out of private cars, and into shared feeder routes to and from mass transit modes is the most promising way to reduce auto traffic. Computer models show that it can be done, and we don’t need autonomous vehicles to realize the benefits of shared mobility.
The role of government, and the planning community, is perhaps to facilitate these kinds of partnerships and make it easier for serendipity to occur. While many cities mandate a portion of the development budget toward art, this will not necessarily result in an ongoing benefit to the arts community as in most cases the budget is used for public art projects versus creating opportunities for cultural programming.
Rather than relying solely on this mandate, planners might want to consider educating developers with examples and case studies about the myriad ways that artists can participate in the development process. Likewise, outreach and education for the arts community about what role they can play in projects may stimulate a dialogue that can yield great results. In this sense, the planning community can be an invaluable translator in helping all parties to discover a richer, more inspiring, common language.