A Cloud-Based Solution to a Radar Data Deluge
Since the late 1970s, scientific satellites equipped with radar have been orbiting Earth and providing invaluable insights into surface topography, atmospheric conditions, ocean currents, ecosystem health, system change, and more.
Various satellite-borne radar technologies exist, each with its own strengths. Among these, synthetic aperture radar (SAR) has been used for an especially wide range of applications—from monitoring forest dynamics to measuring surface displacements caused by earthquakes, volcanic activity, or groundwater extraction—because of its ability to collect data independent of weather and lighting conditions.
SAR systems transmit microwave signals that can penetrate clouds to reach the surface below; energy reflected back to the system is then recorded by onboard sensors. Thus, unlike instruments that measure visible or infrared light, these systems do not require daylight or cloud-free skies to provide high-resolution images of Earth’s surface.
Once sparse and difficult to obtain, synthetic aperture radar (SAR) data from satellites (and aircraft) are now plentiful.
Once sparse and difficult to obtain, SAR data from satellites (and aircraft) are now plentiful—and a revolution is underway in their availability and ease of use for research and teaching. Roughly 15 SAR-equipped Earth-observing satellites are actively collecting publicly accessible data, many on multiyear missions that provide regular and frequent global coverage. The dense data these satellites capture allow researchers to better understand Earth’s systems and changes on the ground that affect millions of people. And whereas processing these data was once the domain of experts in SAR and computer science, powerful new tools are allowing nonexperts to get in on the act.
The Alaska Satellite Facility (ASF), a NASA Distributed Active Archive Center (DAAC), has developed one such tool through its OpenScienceLab, an online platform hosting open-source data analysis solutions. At the heart of OpenScienceLab is OpenSARLab (OSL), a cloud-based environment that can be configured for everything from basic SAR data exploration to developing cutting-edge SAR processing algorithms.
From the Satellite to the CloudSAR data are acquired by an active sensor that transmits radar pulses and then receives signals backscattered, or returned, from Earth’s surface. As a satellite or aircraft carries a (relatively short) SAR antenna forward, the antenna’s location changes relative to points on the ground, approximating a much longer antenna. This synthetic lengthening of the antenna aperture, combined with measurements of the time delay between when pulses are transmitted and received back, allows scientists to focus the signal in the azimuth direction along the flight track. Frequency modulation of the transmitted pulses allows for focusing in the range direction orthogonal to the flight track.
Synthetic aperture radar (SAR) data are acquired by an active sensor that transmits microwave pulses and then receives signals backscattered, or returned, from Earth’s surface. SAR sensors can be very small while also providing high spatial resolution. Credit: NASA/JPL-CaltechTogether these techniques mean that SAR sensors can be very small while also being able to map Earth’s surface accurately, oftentimes with a resolution of just a few meters. Further, by regularly repeating measurements, scientists can produce long time series that allow the detection of changes at the surface.
Owing to their high resolution and dense information content, SAR sensors produce large volumes of data that must be processed and interpreted to be useful for practical applications and decisionmaking.
OpenSARLab (OSL) began in 2017 with the realization that a cloud-based SAR analysis platform was needed to enable the use of data from current—and future—satellite missions.
OSL began in 2017 with the realization that a cloud-based SAR analysis platform was needed to enable the use of data from current—and future—satellite missions. Planning was already underway for the joint NASA-ISRO (Indian Space Research Organisation) SAR, or NISAR, satellite, which promises a tenfold increase in downlinked data compared with Sentinel-1, the current principal SAR mission for ASF. In addition, scientists recognized that a cloud-based platform was needed to teach the research community how to work with NISAR data so that they can meaningfully leverage the data set.
At a 2018 workshop about NISAR applications, OSL was identified as an ideal platform on which the NISAR Science Team could access data and codevelop data processing, calibration, and validation workflows. Until 2021, OSL was available only to the NISAR Science Team and to users enrolled in specific, ASF-run training activities. After that, when it was made available to the public, it quickly grew in popularity because it is accessible through a web browser, free, and powerful enough for SAR analysis.
At the height of the COVID-19 pandemic, when much of the world was in lockdown, the need for a virtual SAR lab grew substantially. Many students and researchers no longer had access to the institutional computing resources they had been using to download and process vast data volumes used for SAR analysis, which accelerated the adoption of OSL by users around the world. OSL has now been used by thousands of individuals across 68 countries as part of an ecosystem of tools provided by ASF.
A Tool for Teaching and ResearchOSL is well suited for teaching new users how to analyze and interpret SAR data while still being powerful enough to serve the needs of advanced users.
OSL is well suited for teaching new users how to analyze and interpret SAR data while still being powerful enough to serve the needs of advanced users. It provides these capabilities transparently and accessibly as an open-source JupyterHub service hosting a Jupyter Notebook environment. Users can access it through any web browser, run existing open-source workflows, and develop new algorithms, all without installing software or downloading data to their computer. Furthermore, with no cost to register and use OSL, it removes a financial barrier that might hinder users, including from traditionally underrepresented groups, from engaging in SAR research.
These features are ideal for students, in part because they are not required to set up scientific computing environments on their own desktops, which can be complicated, time-consuming, and error prone. Students can start working on assignments right away, lab exercises are scalable and customizable, and because students use identical computing environments, the time teachers spend troubleshooting is minimized.
Further, because OSL’s user interface sits alongside ASF’s data archive in the cloud, users can quickly access and copy data to their individual cloud-based workspaces at minimal cost. OSL includes a library of SAR data recipes—available in an open online repository—to help, for example, with time series analyses and change detection workflows. Instructional notes with additional information on the science behind the computational code are available, and through its subject matter experts and seasoned developers, ASF provides extensive support to OSL users, from those teaching and learning SAR processing concepts to NASA scientists advancing understanding of Earth system processes.
The Science That OSL SupportsASF originally developed OSL to support SAR data analysis specifically, but it has evolved as part of the larger OpenScienceLab ecosystem of tools to support research applications across a variety of fields. This wider utility has arisen in part because many types of data can be pulled into OSL for processing, and the tool can be customized to meet different user needs. For example, optical and SAR data can be coregistered (made compatible) over an area of interest in OSL to provide rich context about conditions on the ground for decisionmakers when creating policy or executing emergency response efforts related to volcanoes, wildfires, or other hazards.
Recently, OSL has also been used to process data to study land subsidence and landslides affecting safety and infrastructure in Alaska and India, water transport through glacial aquifers, and even terrestrial analogues for sites on Mars that might have hosted life, among other applications. OSL is also supporting long-term multisite projects around the world. For example, researchers in the South American hub of SERVIR, a joint initiative of NASA and the U.S. Agency for International Development, are training to use OSL to analyze SAR data to track areas at risk of deforestation due to agricultural palm oil and cocoa production.
The NASA-ISRO (Indian Space Research Organisation) SAR (NISAR) satellite will have an orbital repeat cycle of 12 days, providing dense observations of and a wealth of insights into various Earth system processes. Credit: NASA/JPL-CaltechWith the expected launch of the NASA-ISRO (Indian Space Research Organisation) SAR satellite, OSL will be integrally involved in the next revolution in satellite SAR data acquisition.
With the expected launch in 2025 of the NISAR satellite, which will collect data over Earth’s landmasses and ice-covered surfaces, OSL will be integrally involved in the next revolution in satellite SAR data acquisition. The 3-year mission has an orbital repeat cycle of 12 days, which means that with both its ascending (northward) and descending (southward) flight directions, NISAR will provide observations of all covered locations on average every 6 days.
These dense observations will result in a wealth of insights into various Earth system processes as well as into the occurrence of—and potential mitigation strategies for—hazards like earthquake-induced or other land deformations, deforestation, and landslides. They will also result in massive amounts of data—approximately 50 petabytes per year, roughly equivalent to the entire written works of humankind through all recorded history. ASF will manage NISAR data. With OSL sitting beside ASF’s archives in the cloud, users will have low-latency access to the data soon after acquisition by the satellite sensor.
Supporting the SAR RevolutionOpenScienceLab and OSL are well situated to be adopted by scientists, students, and technicians amid the coming deluge of SAR data. These services are exemplars and enablers of open science: The tools and source code are freely and publicly available for use by virtually anyone at any time, and substantial experience in coding and computer science is not required. Users can be trained quickly to perform complex analyses on large quantities of SAR data to produce research products that are accurate, easily reproducible, and needed.
As we continue fine-tuning the tools and computing resources available from ASF to improve their value for science while also minimizing barriers to their use, we envision OpenScienceLab and OpenSARLab increasingly facilitating the SAR revolution to the benefit of our knowledge of Earth’s surface and how it is changing.
Author InformationSargent Shriver ([email protected]), Franz J. Meyer, Alex Lewandowski, Eric Lundell, and Dylan Palmieri, Alaska Satellite Facility, Fairbanks
Citation: Shriver, S., F. J. Meyer, A. Lewandowski, E. Lundell, and D. Palmieri (2024), A cloud-based solution to a radar data deluge, Eos, 105, https://doi.org/10.1029/2024EO240464. Published on 18 October 2024. Text © 2024. The authors. CC BY-NC-ND 3.0Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.
