Global Data Center Environmental Footprint and Heat Island Dataset

Description & Abstract: This dataset compiles geographic coordinates, aggregated facility operational metrics, surrounding demographic distributions, and satellite-derived biophysical indicators for global data centers. Designed to evaluate the microclimatic and heat island (HI) impacts of hyperscale and enterprise computing infrastructure, this archive couples data center location and power capacity estimates with multi-temporal population density layers from the Global Human Settlement Layer (GHSL) and multi-scale thermal/vegetation anomalies derived from satellite observations (e.g., Surface HI intensity and Normalized Difference Vegetation Index [NDVI] gap metrics). Provenance & Attribution Notice: Reprint & Source Notice: Copyright © 2026 S&P Global and/or its Affiliates. All Rights Reserved. Reprinted with permission of S&P Global. Certain foundational facility metrics, specifically data center geographic locations (latitude/longitude coordinates) and multi-year averaged IT power/UPS capacity indicators, are derived from proprietary facility databases licensed through 451 Research by S&P Global. In compliance with the licensing agreement and statement of work governing the use of licensed excerpts: All raw commercial entity names, proprietary commercial identifiers, vendor comparisons, and granular quarterly temporal telemetry have been completely excluded. Only spatial coordinates and derived/aggregated analytical attributes are published. Disclaimer of Liability: Neither the authors, the Pacific Northwest National Laboratory (PNNL), nor S&P Global and its affiliates make any representation or warranty, express or implied, as to the accuracy, completeness, or timeliness of such information, nor do they assume any liability arising from any errors, omissions, or analysis contained herein or derived from the use of this dataset. License & Usage Rights: Derived Dataset & Environmental Indicators: Distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license. Users are free to share and adapt the material for any purpose, provided appropriate credit is given to the original authors and the underlying source attribution. Underlying Source Attribution: Any academic publication, report, or derivative product utilizing the location or power data must also include the 451 Research by S&P Global source info. Variable Descriptions & Data Dictionary: The archive contains 9,726 facility records across 116 variables, organized into five thematic groups covering location and power metrics, demographic context, and multi-temporal/multi-scale satellite biophysical anomalies. 1. Geographic & Facility Operational Attributes (Columns 1–4): latitude: Geographic latitude of the data center facility in decimal degrees. longitude: Geographic longitude of the data center facility in decimal degrees. Country_Name: Full sovereign country or territory name where the facility is located. Averaged Datacenter IT Power: Multi-year averaged facility IT/UPS power capacity in kilowatts (kW), derived from 451 Research by S&P Global records. Missing values (NaN) indicate facilities without disclosed power capacity telemetry. 2. Demographic & Settlement Context (Columns 5–17): GHSL_Pop_2015, GHSL_Pop_2020, GHSL_Pop_2025, GHSL_Pop_2030: Total historical and projected population counts associated with the host Global Human Settlement Layer (GHSL) urban cluster for the years 2015, 2020, 2025, and 2030 (0 for facilities located outside defined urban clusters). In_GHSL_Cluster: Binary spatial indicator denoting whether the facility falls within a delineated GHSL settlement cluster (1 = inside cluster, 0 = outside cluster). Pop_Center_300m_2015, Pop_Center_300m_2020, Pop_Center_300m_2025, Pop_Center_300m_2030: Local population count estimates within the immediate 300 m core circular buffer around the data center for each respective epoch. Pop_RuralRef_425m_2015, Pop_RuralRef_425m_2020, Pop_RuralRef_425m_2025, Pop_RuralRef_425m_2030: Local population count estimates within the surrounding 300 m to 425 m equal-area reference buffer ring around the facility for each respective epoch. 3. Primary Seasonal & Annual Biophysical Anomalies — 300 m / 425 m Reference Scale (Columns 18–41): These 24 variables follow the naming convention {Season}_{Metric}_{Adjustment} and quantify local thermal and vegetation differences between the immediate facility core (300 m radius) and the surrounding reference ring (300 m to 425 m outer radius): Season Prefix (Annual, Summer, Winter): Multi-year composite aggregation window representing full-year (Annual), warm-season (Summer), or cold-season (Winter) satellite observations. Metric Component (HI, NDVI_Gap): HI denotes Surface Heat Island intensity (land surface temperature anomaly in °C, where positive values indicate the facility core is warmer than the reference zone). NDVI_Gap denotes the Normalized Difference Vegetation Index anomaly (unitless, ranging from -1 to 1, where negative values indicate reduced vegetation cover at the facility core relative to the reference zone). Adjustment Suffix (Base, Rural_Unadj, Urban_Adj, Both_Adj): Methodological baseline and pixel-filtering variants used to isolate facility impacts from background land cover: Base: Unfiltered baseline difference between the inner facility core and the outer reference ring. Rural_Unadj: Background contrast evaluated using unadjusted non-urban/rural reference pixels. Urban_Adj: Urban-adjusted contrast controlling for surrounding built-up/impervious surface fractions. Both_Adj: Combined adjustment incorporating both urban and background rural reference corrections. 4. Multi-Temporal Baseline Thermal Subsets (Columns 42–52): Rolling multi-year and single-year temporal subsets of baseline Surface Heat Island intensity (°C) at the primary 300 m / 425 m scale, used to evaluate recent thermal trends and post-construction ramp-up effects: Annual_2021_2024_HI_Base, Summer_2021_2024_HI_Base, Winter_2021_2024_HI_Base: Baseline HI intensity averaged over the 2021–2024 observation window. Annual_2022_2024_HI_Base, Summer_2022_2024_HI_Base, Winter_2022_2024_HI_Base: Baseline HI intensity averaged over the 2022–2024 observation window. Annual_2023_2024_HI_Base, Summer_2023_2024_HI_Base, Winter_2023_2024_HI_Base: Baseline HI intensity averaged over the 2023–2024 observation window. Annual_2024_HI_Base, Winter_2024_HI_Base: Baseline annual and winter HI intensity for the single year 2024. 5. Multi-Scale Spatial Sensitivity Indicators (Columns 53–116): These 64 variables follow the naming convention {Season}_{Metric}_{Adjustment}_{InnerRadius}_{OuterRadius} and replicate the Summer and Winter HI (°C) and NDVI_Gap (unitless) metrics across all four adjustment formulations (Base, Rural_Unadj, Urban_Adj, Both_Adj) at four alternative concentric buffer scales to test footprint sensitivity: 100_150: Micro-scale footprint comparing a 100 m inner facility core against a 100 m to 150 m outer reference ring. 200_300: Intermediate-small footprint comparing a 200 m inner facility core against a 200 m to 300 m outer reference ring. 400_575: Intermediate-large footprint comparing a 400 m inner facility core against a 400 m to 575 m outer reference ring. 500_725: Campus-scale footprint comparing a 500 m inner facility core against a 500 m to 725 m outer reference ring.