Industrial land, crucial for manufacturing and economic development, is expanding rapidly due to urbanization and industrialization. This expansion, however, significantly contributes to global warming through increased energy consumption and fossil fuel use. Industrial activities accounted for approximately 29.6% of global greenhouse gas emissions in 2016. While previous research has focused on single-country or regional analyses (primarily in China), demonstrating a positive correlation between industrial land expansion, economic growth, and CO2 emissions, a comprehensive multi-country analysis considering varying development levels is lacking. This research gap stems from the scarcity of comparable datasets for identifying industrial land across diverse regions. This study addresses this gap by developing a methodology for high-resolution mapping of urban industrial land (IND) and analyzing its impact on economic growth and CO2 emissions across ten countries with substantial industrial value-added, encompassing both developed and developing nations. The study aims to understand the differential impacts of industrial land expansion on economic growth and CO2 emissions across diverse development stages and inform sustainable industrial land management strategies. The importance of this research lies in its potential to provide policymakers with evidence-based insights for balancing economic development with environmental protection and climate change mitigation.

Most existing research on the relationship between industrial land expansion, economic growth, and CO2 emissions has limitations. Many studies focus on single countries or regions, particularly China, showing a positive relationship between industrial land expansion and both economic growth and CO2 emissions. Regional case studies hint at varying impacts depending on socioeconomic factors, but large-scale, multi-country comparisons are rare. The lack of consistent, comparable datasets across multiple countries has hindered this type of research. There's also a scarcity of studies that concurrently explore the impact of industrial land expansion on both economic growth and CO2 emissions. The need for a more comprehensive understanding of the interplay between these factors is crucial for effective land management and sustainable development.

This study developed a novel framework for high-resolution (30m) mapping of urban industrial land (IND) areas across ten countries: China, the United States, Japan, Germany, India, South Korea, Italy, France, Vietnam, and Bangladesh. The methodology employed a synergistic approach combining multiple satellite-based datasets and machine learning techniques. For the 2019 IND map, the researchers used a variety of data sources: World Settlement Footprint (WSF) and global impervious surface area dataset (GISA) for built-up areas and impervious surfaces; OpenStreetMap (OSM) land use data for reference data; Landsat-8 surface reflectance and temperature data; Sentinel-1 SAR data; VIIRS Day/Night Band (DNB) data; local climate zone (LCZ) maps; WorldPop population data; and the global Human Modification dataset (gHM). A random forest (RF) machine learning model was trained on this data to create a 30-meter resolution map of IND areas. For historical mapping (2000-2018), the 2019 IND map was overlaid onto historical GISA data, exploiting the strong correlation between IND expansion and impervious surface growth. The accuracy of the 2019 IND map was assessed and found to be approximately 91%. For the economic growth and CO2 emissions analysis, the study used subnational regions defined by the Global Data Lab (GDL). Regions were classified as developing or developed based on their 2010 Human Development Index (HDI). Data on GDP (using gridded GDP data and GNI for extrapolation), CO2 emissions (from the Open-Data Inventory for Anthropogenic Carbon Dioxide, ODIAC), and population (from WorldPop) were compiled. Additional socioeconomic variables (health and education levels from the Subnational Human Development Index Database), and environmental variables (air temperature from TerraClimate, greenness from MODIS NDVI, and soil properties from SoilGrids 2.0) were also included. A Mixed Effect Random Forest (MERF) model was used for longitudinal analysis (2000-2019) to assess the impact of per capita IND on per capita GDP and CO2 emissions in developing and developed regions separately. SHAP values were used to interpret the model's output and assess the relative importance of different variables.

The study's high-resolution mapping revealed substantial variation in IND area changes across the ten countries. China showed the most dramatic increase in IND area from 2000 to 2019, surpassing the United States. Developed countries experienced comparatively minor changes. Analysis using the MERF model showed unequal contributions of per capita IND to per capita GDP and CO2 emissions between developing and developed regions. In developing regions, per capita IND was a major driver of both economic growth (approximately 31% contribution) and CO2 emissions (approximately 55% contribution). Other factors such as education and per capita NIND (non-industrial urban land) also played roles. In developed regions, however, the impact of per capita IND was significantly reduced for both economic growth (approximately 8%) and CO2 emissions (approximately 3%). Education level emerged as the most influential factor for economic growth in developed regions (approximately 35% contribution), suggesting a shift towards human capital investment as a primary driver of economic growth in these regions. In contrast, population density was the most impactful factor for CO2 emissions in developed regions (approximately 35%). In developing regions, the positive impact of per capita IND on CO2 emissions was much more pronounced than in developed regions. Sensitivity analysis confirmed the robustness of the findings. The impact of per capita IND on both economic growth and CO2 emissions consistently outweighed that of per capita NIND, regardless of the region's development level.

The findings highlight the unequal impact of industrial land expansion on economic growth and CO2 emissions across different development levels. In developing regions, industrial land expansion is strongly linked to both economic growth and increased CO2 emissions, reflecting a carbon-intensive industrial structure and less efficient production processes. In developed regions, the decoupling of economic growth and CO2 emissions is more evident, indicating a transition to less emission-intensive industries and technological advancements. The significant role of education in driving economic growth in developed regions underscores the importance of human capital investment for sustainable economic development. The study suggests that different strategies for industrial land management are needed for developing and developed regions. For developing regions, promoting cleaner energy and low-carbon technologies is crucial to mitigate the environmental consequences of industrial expansion while facilitating economic growth. For developed regions, optimizing existing industrial land use and promoting innovation through human capital investment could maximize economic gains while minimizing further environmental impact. The study also notes that when multinational corporations from developed countries establish industrial facilities in developing regions, they should carefully consider environmental impact assessments and implement emission reduction measures. Transfer of best practices and clean technologies is essential for promoting sustainability.

This study provides crucial insights into the unequal impact of industrial land expansion on economic growth and CO2 emissions across varying levels of development. The findings underscore the need for tailored industrial land management strategies that balance economic development with environmental sustainability. Developing regions should prioritize cleaner technologies and emission reduction measures, while developed regions should focus on human capital investment and optimizing existing industrial land use. Further research could broaden the geographical and temporal scope of the study, incorporate a more diverse set of countries, and investigate the impacts of regional industrial expansion on other regions' economic growth (e.g., offshoring). The framework developed here could also be applied to other environmental factors.

Several limitations need to be considered. The historical IND mapping relied on the assumption of a strong correlation between IND/NIND expansion and impervious surface growth, which might introduce some error. The accuracy of the input datasets (e.g., GISA, WSF, WorldPop, ODIAC) also introduces uncertainties into the analysis. The study focused on CO2 emissions, and future research should expand to other pollutants. The study did not explicitly account for the impact of IND expansion in one region on other regions' economic growth (e.g., offshoring), which warrants further investigation.