A positioning network built for commercial customers is giving researchers a closer view of Japan’s moving ground. By combining SoftBank’s proprietary reference stations with the national GEONET network, a team from Tohoku and Hokkaido universities has mapped crustal deformation at finer scales. The development shows how infrastructure can acquire scientific value beyond the service it was built to deliver.[1]

The study, published in Earth, Planets and Space on September 22 and announced on September 25, uses 3,606 stations selected after quality checks. Its authors are Tohoku graduate student and Japan Society for the Promotion of Science research fellow Miku Ohtate, Tohoku professor Yusaku Ohta, and Hokkaido professors Mako Ohzono and Hiroaki Takahashi. Their subject is strain rate: how the shape of the ground changes over time.[1]

Measuring movement is only the beginning

GNSS, the collective term for satellite navigation systems, makes it possible to estimate a receiver’s position repeatedly. Japan’s Geospatial Information Authority operates GEONET, whose permanent stations observe continuously and send data to a central facility. A sequence of positions provides a record of movement rather than a single location fix.[3]

But displacement and deformation are different. Imagine two points moving east by exactly the same amount. Both have changed position, yet the distance between them has stayed constant. If one moves farther than the other, the distance changes. Strain describes relative deformation; strain rate describes how quickly it changes. Researchers therefore need to compare motion across a landscape, not simply identify which receiver moved the most.

That distinction explains why more stations can be valuable. Two widely separated observations may conceal important differences between them. Extra measurements give an analyst more evidence about where those differences occur. They do not remove the need to distinguish a real ground signal from a problem in the measurement.

Thirty years of public observation

GEONET began operating in 1996. Its roughly 1,300 reference stations support surveying and positioning services as well as crustal monitoring. The authority has retained observation data at 30-second intervals since the network’s launch. In 2026, that record spans three decades—a kind of infrastructure whose value includes both the instruments and the history they preserve.[3]

An additional receiver can improve geographical coverage today, but it cannot manufacture yesterday’s observations. This is why a denser commercial network and a long-running public system offer different strengths. The useful question is how they complement one another, and how consistently their measurements can be compared.

SoftBank launched its ichimill positioning service in November 2019. According to the joint announcement establishing the research consortium, Tohoku began receiving proprietary observation data in June 2021. In August 2022, the university’s Graduate School of Science established the consortium with cooperation from SoftBank and ALES, creating an institutional route for using the network in earth and space science.[4]

The commercial service uses reference stations to support high-precision positioning. That is a different task from estimating slow deformation across a region. A positioning service’s advertised accuracy cannot be read as the resolution of a geological map: the object being measured, the time interval and the analysis are different.[8]

The hidden work behind combining networks

A 2026 paper by Ohta and Ohtate examined the quality of daily coordinates from the proprietary network. It found broadly comparable short-term scatter in the two networks under the tested processing. It also addressed discontinuities associated with non-tectonic changes, including station maintenance. Such a jump in a coordinate series must not automatically be interpreted as sudden movement of the crust.[5]

For businesses considering research partnerships, this is a consequential part of the story. Data do not become useful simply because there are many of them. Researchers need to know when observations were missing, when equipment changed and whether an apparent trend survives appropriate checks. Operational records can be as important as the measurements themselves.

The new study retained 1,224 GEONET and 2,382 SoftBank stations. Its observations cover November 1, 2019, through December 31, 2023. It is therefore neither a live map of September 2026 nor a demonstration that the January 2024 Noto Peninsula earthquake was predicted.[2]

What the sharper map actually means

A central comparison concerns the distance-decay constant, a measure of the locality used in estimating strain. Its average fell from about 39 kilometers with GEONET alone to about 21 kilometers with the combined network under the same analytical settings. That is not a uniform 21-kilometer resolution or the spacing between stations.[2]

The underlying problem is familiar beyond geophysics. Averaging observations over a broad area can hide local differences; relying on too small an area can make a result sensitive to noise. A detailed-looking color map may suggest greater certainty than the observations justify. The scale over which information is combined matters as much as the number of colored cells drawn on screen.

The researchers also checked sensitivity by removing 10% of stations repeatedly in 100 runs. That tests stability against changes in the station set. It does not, by itself, account for every possible source of error.[2]

Readers comparing maps should therefore ask three questions together: what was observed, over what period, and how much spatial detail the observations support. An apparently sharper image is informative only if those questions have satisfactory answers. The improvement here concerns the ability to investigate smaller geographical differences, rather than a guarantee that every small feature is independently resolved.

Connecting the surface to the subsurface

The universities highlight the Niigata–Kobe Tectonic Zone, northern Hokkaido, the San-in Shear Zone and the Ou Backbone Range. They report clearer deformation patterns that can be compared with earthquake activity, volcano locations and underground seismic-velocity structure. They present the work as a basis for investigating inland earthquake mechanisms and improving future hazard assessment.[6]

The importance of such comparison is explanatory. A surface pattern can pose a more precise question about the rocks below it. Why does one part of a region deform differently from another? Does that difference correspond to an independently mapped geological structure? Agreement between different observations can strengthen an interpretation; disagreement can expose an incomplete explanation.

Time remains part of that interpretation. The observation period still includes deformation following the 2011 Tohoku earthquake. The resulting field cannot be treated as a pure picture of unchanging background deformation.[2]

This qualification is more than a footnote. A map records a period in the life of a physical system. Comparing it with an earlier map requires attention to what happened between the two periods, as well as to changes in instruments or methods. Better geographical detail does not make the history of the ground disappear.

A research foundation, not an earthquake timetable

The Japan Meteorological Agency says current scientific knowledge cannot reliably specify the time, place and magnitude of a future earthquake with high precision. Understanding where deformation concentrates and predicting when a fault will rupture are different tasks. The new study should be assessed for the former contribution.[7]

For a local authority, factory operator or logistics business, Japan.co.jp’s assessment is that this work is best understood as an improved input to expert investigation. Turning a strain map directly into a ranking of individual sites would skip the additional evidence needed to assess their circumstances. A more informative map can sharpen the questions asked during an assessment without supplying every answer.

The article provides gridded outputs in its supplementary material. Access to the underlying proprietary observations was through the consortium. Publication of the research products should not be confused with unrestricted public access to every original observation.[2]

The lasting value is in the record

There is an industrial lesson here that does not depend on treating the study as a product endorsement. A commercial network can contribute to public knowledge when access, quality checks and scientific scrutiny make its observations usable for another purpose. That contribution deserves to be measured by the evidence it enables, rather than by the ambition of a partnership announcement.

Japan.co.jp sees continuity as the next practical test. Keeping records, documenting changes and sustaining access are necessary if a useful collaboration is to remain useful years later. More stations improve coverage; reliable stewardship makes observations comparable through time. The achievement is a closer view of a changing landscape—and a stronger basis for deciding what researchers should examine next.