Embodied Carbon Data Construction Materials Data: Dataset Structure and Field Coverage

Three free sources in Datadory's construction-materials slice carry embodied-carbon numbers outright, and each operates at a different scale.

Building Material Stocks and Embodied Carbon — Two Chinese Urban Agglomerations works at urban scale: 26 xlsx tables totalling 132,931,068 records quantify in-use stocks of 11 materials and cradle-to-gate embodied carbon for 38 cities in the Yangtze River Delta and Greater Bay Area, annually from 2000 to 2020, at both city and 500x500 m grid resolution.

Global CO2 Emissions from Cement Production (GCP-CEM) supplies the production-side ledger: annual process CO2 computed from cement and clinker output for 214 countries, 1880 through 2025 (clinker from 1900), in about 2.5 MB across nine files. Every data point carries source attribution, which is why this 9/10-rated record doubles as the citation backbone for cement decarbonization studies.

Two supporting records convert those factors into building-scale quantities: FutuRaM's material intensities in kg/m2 and BMAT's footprint-level facade labels.

These sources are not interchangeable: one is a modeling template, one is a measured urban accounting exercise, one is a national emissions reconstruction, and one is an intensity library. Confusing them is the most common way embodied-carbon analyses go wrong.

NIST states its own scope plainly - process-level models per material or product, intended as templates rather than industry averages, for United States construction with background processes drawn from US LCI databases. There is no time series; treat each release as a static, inspectable calculation you adapt rather than a factor table you quote.

The Chinese agglomeration set separates its two concerns into distinct workbooks: 'Total MS data' (material stocks, roughly 216 KB) and 'Total ECE data' (embodied carbon emissions, roughly 202 KB), with the gridded per-material tables substantially larger. Keeping stocks and emissions apart lets you compute carbon intensity per tonne of in-use material instead of conflating the two.

GCP-CEM measures neither buildings nor products. It reconstructs process CO2 from clinker chemistry at country-year resolution, so it belongs on the production side of your ledger - the emissions embodied in a supply chain, not in a specific wall assembly.

A defensible first-pass workflow assembles in five steps, each anchored to a named dataset:

1. Set the calculation frame with NIST. Open the NIST openLCA model for your material - cement, concrete or gypsum board - and keep its process boundaries intact. Because the models bridge to Federal LCA Commons USLCI/eLCI databases, background flows resolve without extra sourcing work.

1. Attach material quantities with FutuRaM. Pull kg/m2 intensities from FutuRaM — Building Composition and Material Intensity Raw Data: 7,346 data rows across 68 columns, covering 28 mostly EU/EEA countries and building production years 1859-2012, keyed by archetype and construction year.

1. Add building-level material mix with BMAT. Where you have geography but no drawings, BMAT's facade labels classify 22.09 million individual footprints across 73 major world cities - including roughly 37 Chinese cities - into nine material classes inferred from 147 million street-view images collected 2007-2025.

1. Sanity-check against published totals. The Chinese urban agglomeration tables give you independent city-level and 500x500 m grid results for 2000-2020; if your bottom-up estimate for Shanghai or Shenzhen lands far outside their range, revisit your intensity assumptions before publishing.

1. Frame the supply side with production data. Join GCP-CEM's country-year process CO2 and, for monthly resolution, Monthly Global Cement Production Data (CICERO), whose 175 collated time series include 70 series across 24 countries linked to live national statistics.

Further than any other construction-materials series in the catalog. GCP-CEM's emissions and cement production lines run 1880-2025, clinker from 1900, and the May 2026 release (version 260526) ships nine countries with detailed national workbooks - China's reaches back to 1949 with monthly NBS series. The methodology paper, Andrew (2019) in Earth System Science Data 11, 1675-1710, documents the computation you are inheriting.

Monthly resolution is rarer than users expect. CICERO's collation packs 528 monthly rows by 176 columns into a 292.4 kB CSV (915.5 kB across four files with metadata and clinker trade), with observations from 1982 in Taiwan's case through late 2025 depending on country, refreshed quarterly on Zenodo.

Asset-level context closes the loop. Global Energy Monitor's Global Cement and Concrete Tracker lists 3,884 cement and clinker plants across 171 countries and areas in its July 2026 release - roughly 6.1 billion tonnes of annual cement capacity and 3.7 billion tonnes of clinker capacity - with kilns, ownership status and green-cement technology. The Spatial Finance Initiative's Global Database of Cement Production Assets and Upstream Suppliers adds what no other cement asset database attempts: plant-to-mine supplier attribution rows, under commercial delivery terms, with the supplier file (2.15 MB) outsizing the asset file (1.01 MB).