Data centres near Tokyo: satellite study of heat impact

Jiawei Jing of Makuhari Development Corporation (Chiba, Japan) published a satellite study of the Inzai and Tama data-centre clusters near Tokyo in Environmental Research Letters on 4 November 2026 (doi.org/10.1088/1748-9326/aead02). doi.org The work covers twelve hyperscale facilities built in 2011–2025 and tests whether they warm their surroundings. doi.org According to the author, commissioning a facility cools daytime land surface temperature (LST) by 1.8–2.2 °C. doi.org

Data and method

The staggered build-out gave a three-phase event study with replication. The author combined daytime Landsat images (2013–2026), 70 m day and night ECOSTRESS data and 1 km MODIS data. The results were compared with matched pre-2013 controls, facility-level energy disclosures and a minimum-detectable-effect analysis. doi.org

Surface results were bounded by an air-temperature energy-balance ceiling of about 0.35 °C (cluster scale, night). doi.org

The study responds to conflicting evidence. A preprint that has not been peer-reviewed (Marinoni et al., 2026) reports mean LST warming of +2.07 °C around thousands of facilities, decaying out to about 10 km, with heat exposure claimed for roughly 340 million people. Peer-reviewed vehicle transects around four air-cooled facilities in Phoenix (Sailor et al., 2026) found near-ground air warming of +0.7–0.9 °C within a few hundred metres downwind. doi.org

Night temperatures and energy balance

Deep-night LST changes at facility scale were indistinguishable from zero, with a bound below about 0.6–1.2 °C. doi.org The cluster-scale night LST bound was 0.24–0.29 °C, corrected for autocorrelation and measured on two platforms, which is below the air-temperature ceiling. doi.org By the author's energy-balance estimate, +2 °C even in air would require 6–43 times the available waste heat. doi.org

Consistency checks in Phoenix (arid climate) and DeKalb (continental climate) reproduced the pattern. In both, night surface warming was largely in place during construction, before any server load, and matched neighbouring non-data-centre development. doi.org The author concludes that near-source air warming is real where heat-rejection density is high and the climate is dry, while kilometre-scale surface warming from waste heat is unattainable. doi.org

The paper also notes that none of the major data-centre markets (Singapore, Ireland, the Netherlands, the United States, China, Japan) requires an assessment of local temperature impact at the national permitting level before construction. doi.org The author presents the protocol as an ex-post verification tool for siting regimes. doi.org

Study of 12 Tokyo-area data centres: daytime surface 1.8–2.2 °C cooler, cluster night warming under 0.3 °C

Data and method

Night temperatures and energy balance

Sources · 2

Read also

Flow microreactors: review cites linezolid in 27 minutes and 780+ industrial installations