Substantial terrestrial carbon emissions from global expansion of impervious surface area

The global urban land area has more than doubled in the past three decades and is projected to continue expanding significantly by the end of the century. Impervious surface area (ISA), encompassing pavements, roads, and built-up areas, is a key characteristic of urban land, easily identifiable via remote sensing. ISA expansion profoundly impacts the environment, causing heat islands, biodiversity loss, and disruptions to carbon, nitrogen, and hydrological cycles. Replacing carbon-rich ecosystems with ISA results in carbon losses from both biomass and soil. Current global carbon budget assessments by the Global Carbon Project (GCP) and the Intergovernmental Panel on Climate Change (IPCC) neglect these carbon losses from ISA expansion, using bookkeeping models and dynamic global vegetation models (DGVMs) that either don't include ISA transitions or represent ISA inconsistently. While some studies have assessed the loss of net ecosystem productivity (NEP) or net primary productivity (NPP) due to ISA expansion, this represents unrealized carbon uptake, not direct atmospheric CO2 addition. Therefore, the terrestrial carbon losses from historical ISA expansion (ISA-driven carbon emissions) remain largely unaddressed. National greenhouse gas inventories (NGHGIs) from Annex I (AI) UNFCCC countries report carbon emissions from 'settlements,' which include ISA and associated vegetated areas. However, AI country data suggest that carbon losses are mainly due to ISA expansion, justifying independent satellite-based estimates of ISA-driven emissions to validate NGHGIS data. This study aims to quantify terrestrial carbon losses from the direct land-use effects of global ISA expansion between 1993 and 2018.

Previous studies have quantified the loss of terrestrial net ecosystem productivity (NEP) or net primary productivity (NPP) caused by global ISA or urban expansion. However, these studies focused on unrealized carbon uptake and did not address the physically tangible CO2 emissions resulting from the conversion of carbon-rich ecosystems to impervious surfaces. The terrestrial carbon losses from historical ISA expansion, a potentially significant component of anthropogenic carbon emissions, have been largely ignored by the global carbon-cycle science community. National greenhouse gas inventories (NGHGIs) from Annex I (AI) UNFCCC countries report carbon emissions from 'settlements,' which include ISA and associated vegetated areas. However, AI country data suggest that carbon losses are mainly due to ISA expansion, justifying independent satellite-based estimates of ISA-driven emissions to validate NGHGIS data. This study aims to quantify terrestrial carbon losses from the direct land-use effects of global ISA expansion between 1993 and 2018.

This study used four state-of-the-art global remote-sensing products of impervious surface area: three 30-m resolution products (GAUD, GAIA, GISA) and one 300-m resolution product (ESA CCI). The analysis assumed 100% ISA coverage for mapped ISA pixels, ignoring sub-pixel green areas. This approach, while imperfect, is consistent with land cover mapping assumptions and unlikely to overestimate global ISA expansion. Annual global ISA expansion was derived from the individual ISA products, with source land covers obtained from the ESA CCI land cover product. To quantify soil organic carbon (SOC) loss following ISA establishment, two alternative loss ratios were used: the IPCC default value (20%) and a value derived from a literature synthesis (59.5%). Carbon losses due to conversion from vegetated land to ISA were quantified by overlaying spatial maps of ISA expansion, source land covers, and carbon stock maps of above- and below-ground biomass, surface litter, dead wood, and SOC. Biomass losses were limited to forest, shrubland, wetland, cropland, and grassland, while SOC losses were assumed for all source land covers. Quantified losses represent committed emissions, assuming immediate carbon loss upon conversion to ISA. The derived ISA-driven carbon emissions were compared with those from settlement expansion reported by NGHGIS for AI countries. A structural decomposition framework attributed emission dynamics to socioeconomic and urban factors, both globally and separately for AI and Non-Annex I (NAI) country groups.

Analysis of the four ISA products revealed that global ISA expanded at a rate of 15,913 ± 3331 km²/yr between 1993 and 2018, representing a 112% increase. Approximately 65% of this expansion occurred at the expense of cropland. The global-scale meta-analysis of SOC loss under urban green areas and ISA showed an average loss ratio of 59.5 ± 16.6% following ISA establishment. Using this value as the upper boundary and the IPCC default of 20% as the lower boundary, terrestrial carbon losses due to ISA expansion reached 74.9 ± 13.7 Tg C/yr (upper boundary) and 45.8 ± 8.2 Tg C/yr (lower boundary). Loss of soil organic carbon accounted for 32–59% of total emissions, with the remainder coming from biomass. Concurrent with accelerating global ISA expansion, annual carbon loss showed a slight increasing trend (0.36–0.78 Tg C/yr²). Fifteen countries/regions accounted for 81.1–82.0% of global emissions, with China, Brazil, India, and Indonesia driven primarily by urban population growth, while the USA, EU27, and others were driven by ISA growth. AI countries accounted for 10.1% of global urban population growth but 39.2% of ISA expansion and 51.1–54.5% of carbon emissions. The spatial distribution of emissions closely followed ISA expansion patterns. Comparison with NGHGIS for AI countries showed that annual biomass carbon losses (18.2 ± 3.9 Tg C/yr) were close to NGHGIS reports (22.9 ± 0.9 Tg C/yr), while NGHGIS SOC loss (14.8 ± 2.2 Tg C/yr) fell between the study's lower (6.8 ± 1.5 Tg C/yr) and upper (20.1 ± 4.3 Tg C/yr) boundaries. An 'ISA-driven Emissions Identity' framework attributed emission dynamics to total population, urbanization rate, residential ISA intensity, ISA expansion speed-up factor, and carbon emission intensity. The small growth rate in global ISA-driven carbon emissions (0.92%) resulted from opposing effects of population growth, urbanization, and residential ISA intensity, offset by decelerating ISA expansion and decreasing emission intensity. Contrasting trends in AI and NAI countries reflected changes in these driving factors. Analysis revealed a clear pattern of ISA-driven emissions dynamics related to economic development, with high growth rates in emissions found in countries with low GDP per capita, and declining rates in countries with higher GDP per capita.

This study highlights the substantial, previously overlooked contribution of global ISA expansion to anthropogenic carbon emissions. The estimated ISA-driven emissions (1.19–1.95 Pg C over 1993–2018) account for 3.7–6.0% of global land-use change emissions, potentially higher if considering SOC loss up to 1 m depth. The study intentionally excluded urban vegetation due to inconsistent definitions and the absence of robust estimates of urban vegetation carbon sinks. While enhanced urban vegetation productivity might offset some ISA-driven losses, the limited evidence from national inventories of AI countries suggests minimal carbon sequestration from urban vegetation. The study's estimates for Annex I countries are comparable to NGHGIS data, offering independent validation. This bottom-up approach can enhance transparency in greenhouse gas inventories, particularly for NAI countries. The observed pattern of ISA-driven emissions linked to economic development provides a predictive framework. Projected future urban growth will lead to further carbon losses unless sustainable urban planning practices are adopted. Global urbanization discussions should account for the land sector alongside energy systems.

This research quantifies the significant and previously unaccounted for contribution of global impervious surface area expansion to terrestrial carbon emissions. The findings highlight the need to incorporate ISA expansion into global carbon budget assessments and national greenhouse gas inventories, particularly for non-Annex I countries. The study's economic development-related emission patterns offer insights for future projections and the importance of sustainable urban planning in climate change mitigation strategies. Further research is needed to refine estimates of soil organic carbon loss and to explore mitigation strategies.

The study's assumption of 100% ISA coverage in mapped pixels ignores potential sub-pixel green areas, potentially underestimating ISA expansion. The limited number of observations for soil organic carbon loss analysis and the reliance on existing carbon stock maps introduce uncertainties. The study’s focus on direct land-use effects neglects indirect impacts of urbanization on carbon cycling.