Cloud Computing — Improving access to satellite imagery
Project by Kate Kuehl and Rahul Bhojwani
As we step into an era of high-definition satellite imagery and advanced graphical processing units (GPUs), a key question emerges - how do we store and process such massive volumes of data? With satellite imagery sets now reaching into the hundreds of petabytes, relying on personal computers for storage quickly becomes impractical, if not impossible. Enter cloud computing - a transformative solution for handling the colossal amount of satellite data, providing a feasible, efficient, and scalable approach.
With the growing accessibility and application of satellite imagery in various sectors such as agriculture, climate science, urban planning, and disaster management, selecting the best source for your satellite imagery needs can be challenging. Three major platforms — Google Earth Engine, Amazon Web Services (AWS), and NASA — provide comprehensive satellite imagery data. This blog post aims to compare these platforms in terms of data availability, accessibility, cost, and user-friendliness.
Google Earth Engine
Google Earth Engine is a cloud-based platform for planetary-scale geospatial analysis. It boasts a multi-petabyte archive of satellite imagery and geospatial datasets, which includes NASA’s Landsat and the European Space Agency’s Sentinel collections.
The Earth Engine hosts public datasets and provides APIs and a code editor for users to analyze and visualize the data. What makes Google Earth Engine stand out is its high-level APIs and ability to leverage Google’s computational power to run complex geospatial analyses, often reducing processing time from months to seconds. However, while using the platform is free, there are potential costs if you want to export a significant amount of processed data.
Amazon Web Services
AWS offers a broad array of cloud-based services, including storage and computation. Among these, the AWS Earth program provides a variety of satellite imagery data through the AWS Public Dataset Program, such as Landsat, Sentinel-2, and high-resolution aerial imagery from the NAIP.
One of the main advantages of AWS is its scalability, allowing it to handle large-scale geospatial data workloads. Moreover, AWS provides several tools and SDKs to interact with the data. The primary cost factor for AWS comes from data storage, transfer, and computation, but there is a free tier for beginners, and public datasets sometimes do not incur data transfer costs.
NASA Earth Exchange
NASA, being at the forefront of space exploration and earth science, offers an extensive array of satellite data. The NASA Earthdata Search portal allows users to access numerous satellite datasets, including those from missions like Landsat, MODIS, and more.
NASA’s data is free to access, and it provides several user-friendly tools, such as Worldview for data visualization and AppEEARS for data extraction and transformation. The level of pre-processing can vary between datasets, and users may need to do more data cleaning and preparation compared to the other platforms. While NASA’s data is comprehensive and free, it lacks the built-in, large-scale analytical capabilities of Google Earth Engine and AWS.
Data Sources
Google Earth Engine, AWS, and NASA offer distinct satellite image options for diverse applications. Google Earth Engine provides a user-friendly platform that combines a vast collection of satellite imagery with powerful geospatial analysis tools. It offers seamless access to a wide range of high-resolution imagery, including Landsat and Sentinel data, enabling users to explore and analyze Earth’s surface in great detail. AWS, on the other hand, provides cloud-based infrastructure and services for satellite data storage, processing, and analysis. It offers a scalable and customizable solution for handling massive datasets, making it suitable for large-scale projects. NASA, renowned for its space exploration missions, provides an extensive collection of satellite imagery and remote sensing data from various Earth-observing missions. These datasets, such as MODIS and VIIRS, allow researchers to study global environmental changes over time. NASA also provides specialized tools and resources for analyzing satellite imagery, supporting scientific research and applications. Each option has its own strengths and user base, catering to different needs within the satellite imagery domain.
Applications: Semantic Segmentation
Semantic segmentation involves classifying each pixel in an image into distinct categories, enabling detailed land cover analysis, urban planning, and environmental monitoring. By leveraging satellite imagery, researchers and decision-makers can identify and map various land cover types like vegetation, water bodies, buildings, and roads, providing valuable insights for resource management and land-use planning.
Applications: Object Detection
Object detection in satellite imagery entails identifying and locating specific objects of interest within an image, such as buildings, vehicles, or infrastructure. This application is vital for urban development, disaster response, and security monitoring. Governments and organizations can use satellite imagery and object detection techniques to detect changes in built-up areas, assess damage after natural disasters, and monitor construction progress.
Applications: Road Detection
Road detection is another crucial application of satellite imagery. Accurate road networks are essential for transportation planning, infrastructure development, and navigation systems. Satellite imagery can be employed to automatically extract road networks, enabling efficient route planning, traffic management, and urban development. By leveraging machine learning algorithms and satellite imagery, road detection can be automated on a large scale, saving time and resources.
Conclusion
Cloud computing offers a transformative solution to the challenges of storing and processing satellite imagery. By leveraging the scalable storage, powerful processing capabilities, and advanced analytics tools offered by cloud platforms, we can unlock the full potential of satellite data, opening up new possibilities for scientific discovery, policy-making, and commercial applications.
If you’d like more information, below is our video presentation on the subject: