Ten Rules Could Unlock the Power of NASA’s Earth Observation Data

Every day, satellites orbiting hundreds of kilometers above our heads generate torrents of data about the planet: the height of ice sheets, the extent of forests, the temperature of oceans, and the structure of the atmosphere. NASA missions alone are projected to produce more than 320 petabytes of data by 2030, a volume so vast that simply storing it strains conventional systems. Yet according to a new review published in Earth Science Informatics, the greatest obstacle to turning this flood of information into scientific discovery and societal benefit is not the data itself. It is the way missions are designed, archived, and delivered to the researchers, policymakers, and communities who need them. A team of scientists and software engineers, drawing on years of experience with NASA’s ICESat-2 laser altimetry mission, has distilled their lessons into ten rules for making publicly funded Earth observation datasets genuinely usable.

The central argument of the paper, led by Jessica Scheick of the University of New Hampshire and Anthony Arendt of the University of Washington’s eScience Institute, is that open access alone is not enough. Making data freely downloadable does not guarantee that anyone can efficiently analyze it. The authors describe a paradigm shift in mindset: instead of a data provider building systems in isolation and then handing them to users, mission organizations should co-create data formats, software tools, and communities with the researchers they serve. This means involving external research and software specialists from the very beginning of mission planning, not after a satellite has already launched and its data formats are locked in.

The ICESat-2 mission illustrates both the problem and the promise of this approach. The satellite’s Advanced Topographic Laser Altimeter System, or ATLAS, is a photon-counting instrument that measures surface height by timing the flight of individual laser photons from the spacecraft to Earth. Its most basic geophysical product, ATL03, contains nearly a thousand variables per file organized across six beams, with roughly 300 gigabytes of data spread across more than 200 files each day. The mission archives most products in Hierarchical Data Format 5, or HDF5, a format prized for its metadata and compression but poorly matched to modern cloud-based access patterns. Researchers who wanted to study a specific region were forced to download enormous swaths of data organized by satellite flight path and then subset them locally, wasting storage, processing time, and money.

The turning point came through community gatherings. At the first hands-on ICESat-2 workshop in 2019, participants discovered that the available data access tools were rigid and difficult to customize, and they spent days writing their own scripts just to reach the data. That friction sparked a deeper collaboration between users and the National Snow and Ice Data Center, the NASA archive responsible for ICESat-2. With mission funding for annual hackathon-style events, the partnership produced open-source libraries such as earthaccess, which simplifies NASA data search and authentication, and icepyx, tailored to ICESat-2 workflows. The community also built SlideRule Earth, a cloud-based system that processes raw laser returns on demand, and CryoCloud, a shared computing platform for the cryosphere research community.

Perhaps the most striking technical payoff came from rethinking file formats. During the 2023 ICESat-2 Hackweek, staff from the data center, the HDF Group, and the mission’s processing team tested a cloud-optimized version of HDF5 for the mission’s seventh data release. The result was dramatic: access times to ICESat-2 data in the cloud dropped by at least an order of magnitude in every case tested, and in the worst scenarios fell from roughly twenty minutes to about thirty seconds. The innovation was quickly adopted by the NASA-ISRO Synthetic Aperture Radar mission, launched in July 2025, showing how improvements born in one mission’s community can ripple across the entire Earth science enterprise.

Several of the ten rules address the social infrastructure that makes such technical wins possible. The authors call for regular community-building and educational gatherings, particularly hackweeks modeled on the University of Washington eScience Institute’s format, which blends short-format training with collaborative project work. They also urge missions to fund persistent shared computing environments rather than temporary workshop hubs. CryoCloud, operational since October 2022, offers users up to 16 CPUs, 128 gigabytes of RAM, GPU access, and 45 gigabytes of persistent storage, managed under a shared-responsibility model by the nonprofit 2i2c, which handles cloud engineering, security, and cost monitoring while researchers govern access. Because users keep their environments after events end, the effort invested in learning new tools continues to pay off in ongoing research rather than evaporating when a temporary hub is shut down.

Communication matters as much as computation. The paper recommends centralized, open, self-governed asynchronous spaces, such as Slack workspaces paired with the computing hubs, where data producers, software engineers, archive specialists, and science users can interact regardless of institution or time zone. Usage data from CryoCloud show that communication platform activity remained consistently high even during seasonal lulls in computing, signaling sustained collaboration. Centralizing these conversations also relieves a hidden burden: user support specialists in the ICESat-2 ecosystem previously had to monitor at least eight Slack organizations, five GitHub organizations, and countless email lists and web forms, a fragmentation that left users posting questions in the wrong places and answers buried where no one could find them.

The remaining rules tackle sustainability and standards. The authors argue that missions should fund developers and open-source contributors with dedicated staff time rather than relying on volunteer labor, which they describe as unsustainable and disproportionately burdensome to the most committed community members. They recommend building open, extensible, reusable tooling that plugs into the existing scientific software stack instead of reinventing bespoke systems for each mission, and they call for community-owned instructional resources, such as the ICESat-2 Cookbook and curated resource lists, maintained as living documents on non-institutional platforms. Crucially, they advocate formal roles for community managers and leaders, identified at the mission planning stage, to organize engagement, mentor newcomers, and coordinate development roadmaps across tools that might otherwise evolve in isolation.

The final rule addresses performance and reproducibility. The authors recommend that missions adopt cloud-native, analysis-ready standards from the outset, including consistent formats, metadata conventions such as the SpatioTemporal Asset Catalog and CF conventions, and shared benchmarks built around representative scientific use cases. Without such benchmarks, individual teams run incomparable optimization exercises on arbitrarily chosen data subsets, and their results cannot guide mission-wide decisions. The rise of artificial intelligence raises the stakes further: machine-learning workflows for automated feature extraction and large-scale model training demand fast, consistent, scalable data access that only cloud-optimized formats and rich machine-readable metadata can provide.

Taken together, the ten rules sketch a future in which Earth observation missions are designed not just for their instruments but for their users, with open collaboration codified into the mission lifecycle itself. The authors acknowledge that NASA already solicits community input through Early Adopter Programs and Science Definition Teams, but they argue these efforts have been volunteer-driven, uneven, and secondary rather than central to mission planning. Their proposal is to make co-creation a first-order planning tool, equipping Early Adopters with realistically sized synthetic data years before launch and embedding software experts in mission design decisions. If the ICESat-2 community’s experience is any guide, the payoff could be enormous: faster science, lower costs, tools that outlive their missions, and, ultimately, a greater return on the public investment in watching our changing planet from space.

Subject of Research: Improving the usability of Earth observation mission datasets through open, collaborative mission design

Article Title: Ten rules to increase the usability of earth observation mission datasets

Article References: Scheick, J., Barrett, A. P., Fair, Z., Katz, Z., Lopez, L., Neeley, A., Roberts, C., Smith, B., Snow, T., Wegener, R., & Arendt, A. (2026). Ten rules to increase the usability of earth observation mission datasets. Earth Science Informatics, 19(11), Article 186. https://doi.org/10.1007/s12145-026-02216-5

Image Credits: AI Generated

DOI: 10.1007/s12145-026-02216-5

Keywords: Earth observation, ICESat-2, NASA, open science, cloud computing, data formats, HDF5, hackweeks, community management, open-source software, CryoCloud, data usability