New & Historical Building Starts: Detection of Construction from (Freemium) Satellite Imagery
Within the construction industry, data produced by construction companies and the U.S. Commerce Department can forecast the industry’s overall performance. U.S. and Global information can reveal and localize trends like economic growth, gentrification, adoption of solar, zone usage, and more.
As the satellite industry transforms itself into a rapid and responsive data-driven industry, the data they collect can shed light on the health of various aspects of the economy. Datasets can be created that impact industries including oil & gas, retail, farming, shipping, and manufacturing, and insurance.
Detecting New Construction
The NASA LandSat satellite has produced a moderate resolution map of the entire Earth every two weeks for the last forty years. These global maps provide accurate accounts of change anywhere, stretching back decades. We’re talking real ground truth, if you know to read the data. To the left, we see the construction of a new gas station and new residential development. Pennsylvanians know the Wawa regional chain.
The Starlight Aether SDK (docs) is a tool analysts use to make accessing and processing of satellite data sets ridiculously easy. The video to the left is every cloud-free LandSat satellite image of a 2 mile by 2 mile region in Bucks County, PA, from 2005 until 2018, and processed to highlight construction, concrete, and impervious surfaces. The additional processing makes this data analysis-ready. Our tools easily eliminate imagery polluted by clouds and cloud shadows, normalize coordinates, and highlight the land use character of interest. The exact coordinates of the Wawa are lat/lng (40.17, -75.07).
Characterizing Construction Projects and Dates
The analysis-ready processing of LandSat imagery gives us what we need to draw real insights and characterization of this change detection. The change pattern over time follows a well understood form that we can extract from the dataset. The time series data shows the characteristic change pattern of building construction. For even better quality, we confirm our characterization with information derived from news.
The independent validation of real ground truth is the point. Using external, existing, and historical sources triangulate information that can be determined with enough confidence to act and operate against. Ultimately, the service to business users is a model of data value that is independent, high quality, and global scale.
Global Industries, Global Datasets
The global construction market is expected to grow beyond $24 Trillion — with a T — through 2021, smashing estimates and annual rates from 2017.
Asia Pacific has a nearly 60% share of the global construction market.
The explosion of construction worldwide can be attributed to the growing economy in emerging countries, increasing government investments in large-scale infrastructure projects, and increased public-private partnerships in these infrastructure projects.
Foxconn, prominently known for its manufacturing of the Apple iPhone, made news with its announcement of a $10 Billion Wisconsin manufacturing plant, including $3 Billion of state incentives to select Wisconsin.
Foxconn, headquartered in Shenzhen, China, is the world’s largest contract electronics manufacturer. The Foxconn Zhengzhou Branch, shown to the left, emerged after 2009, and is now a “city” which employs more than 120,000 employees. The largest factory is its Longhua Science & Technology Park, and employs upwards of 400,000 people.
More questions about our analysis of satellite imagery to arrive at local or economic indices of construction? Please reach out to us at Starlight.
The Starlight Aether code to render these images is in Github and Github Gist.
Get to know us. Starlight.
David Bernat is founder and president/CEO of Starlight, a company dedicated to innovating intersectional technology with satellites and AI. Dr. Bernat is a Space Scientist and former small fry on the Goldman Sachs strategies desk.
Use the Aether SDK as part of the Starlight Network
This link will having you downloading satellite data in less than 15 minutes.
