The study uses the Remote Sensing Ecological Index, or RSEI, to represent ecological quality through four components: greenness, wetness, dryness and heat. Because the index is calculated from observations acquired at the target date, it is retrospective: it identifies ecological conditions and changes after observations become available but does not itself forecast the following period. Its conventional construction through principal component analysis is also linear and does not explicitly capture nonlinear temporal interactions between ecological indicators and external environmental drivers. The proposed framework was designed to address these limitations by assigning feature extraction, temporal prediction and multivariable fusion to separate neural networks.

Within the staged convolutional neural network–long short-term memory–backpropagation neural network framework, or CLB, a convolutional neural network extracts features from multivariable time series. A long short-term memory network then models temporal dependence and predicts target-period variables. Finally, a backpropagation neural network fuses the predicted remote-sensing indicators with climatic, topographic and spatial variables to estimate the target-period RSEI. The architecture therefore produces a forecast from earlier observations and supporting environmental information rather than simply calculating an index directly from imagery for the date being evaluated. This design also allows the three networks to represent different parts of the prediction process: pattern extraction, time-dependent forecasting and nonlinear integration.

Landsat-8 imagery from 2015 to 2020 was used to quantify ecological change in the Wangpo Mine Field. Across that observed period, the mean RSEI increased by 12.1%, from 0.6136 to 0.6879. Areas classified as excellent quality expanded, while poor-quality areas contracted. These results describe a measured improvement in the ecological index for the study area and period. They do not show that the forecasting framework caused the landscape change, and the reported values remain specific to the mine field, the selected indicators and the six-year Landsat-8 record.

For the held-out 2020 data, the RSEI grades predicted by the CLB framework were consistent with grades calculated from Landsat-8 imagery, with reported grade-specific accuracies ranging from 87.3% to 94.1%. The test demonstrates agreement between one-step-ahead predictions and direct image-based classifications for this dataset. The supplied study abstract does not report uncertainty intervals, error distributions or comparisons with alternative forecasting models, and it does not establish how the framework would perform in other mining regions, under different time scales or under future climatic and land-use conditions. The authors therefore present CLB as a methodological reference for one-step-ahead ecological-quality prediction in mining areas rather than as a universally validated forecasting system.

Subject of Research: One-step-ahead ecological-quality prediction in a Chinese mining area using Landsat-8 data and a multi-neural-network framework

Article Title: Mine ecological assessment and prediction based on RSEI and a multi-neural-network framework

Article References: Li, P., Wang, B., Guo, Y., Liang, Y., & Yan, J. (2026). Mine ecological assessment and prediction based on RSEI and a multi-neural-network framework. Scientific Reports. https://doi.org/10.1038/s41598-026-73044-2

Image Credits: AI Generated

DOI: 10.1038/s41598-026-73044-2

Keywords: mine ecology, ecological forecasting, Remote Sensing Ecological Index, Landsat-8, convolutional neural network, long short-term memory, backpropagation network, Wangpo Mine Field, Shanxi, machine learning, Mine, ecological

The study area spans 34,484 square kilometers across the cities of Linfen and Yuncheng, a region long dominated by coal mining, chemical production, and heavy industry. During the winter heating season, cold, dry Siberian air masses settle over the basin, wind speeds drop, and the surrounding mountains trap pollutants close to their sources. Coal-fired heating and open straw burning add to the burden, making winter the season with the most severe and complex pollution episodes. It is precisely this combination of topography, meteorology, and intensive human activity that made southern Shanxi an ideal natural laboratory for asking whether the arrangement of urban landscapes can measurably shape air quality and carbon output.

To answer that question, the researchers assembled an unusually detailed dataset. Carbon emissions came from the Open Data Inventory for Anthropogenic Carbon Dioxide, a 1-kilometer gridded product built by disaggregating nighttime lights, power plant emissions, and sectoral energy use. Particulate matter concentrations for PM2.5 and PM10 were drawn from 66 ground monitoring stations run by Shanxi’s environmental authorities between November 2022 and March 2023, then interpolated into continuous pollution surfaces using Bayesian kriging—a method that quantifies its own uncertainty and copes well with unevenly spaced stations. Cross-validation showed the interpolation performed reliably, with coefficients of determination of 0.714 for PM2.5 and 0.705 for PM10.

The landscape itself was characterized at remarkable resolution. Sentinel-2A imagery at 10 meters, processed with atmospheric correction and object-based classification, divided the land into green land, wetland, cropland, water body, and impervious surface, achieving an overall classification accuracy of 83.2 percent. For the vertical dimension of the city, the team turned to Gaofen-2 imagery at 0.8 meters, exploiting the geometric relationship between building shadows and true building heights—validated with field measurements—to calculate metrics such as floor area ratio, building coverage ratio, average height, and a newly constructed building rugosity index. Green space was described with eight patch-level indices capturing coverage, patch density, shape, contiguity, and connectivity, supplemented by satellite-derived net primary productivity as a measure of vegetation’s carbon-sequestration capacity.

The resulting maps revealed a striking and somewhat counterintuitive geography. Carbon emissions clustered intensely in the industrial heartlands—Yaodu District, Yanhu District, and Hejin City—where maximum emission intensity reached 168,486 tonnes per square kilometer. Particulate matter, by contrast, followed a southwest-to-northeast gradient, highest around Yanhu District and lowest in the greener, better-ventilated northeastern counties such as Gu, Anze, and Fushan. Crucially, the two pollution types overlapped locally but diverged overall: Yanhu District was a hotspot for both, yet Yaodu District, a core carbon-emission zone, did not show correspondingly peak particulate levels. County-level Pearson analysis confirmed only weak positive correlations between emissions and PM2.5 (r = 0.422) or PM10 (r = 0.362), a finding the authors attribute in part to Yaodu’s service-heavy economy, higher penetration of natural gas and electricity, centralized heating, and strict emission controls in the urban core.

When the team correlated seventeen landscape indices against emissions and pollution, a clear pattern emerged. Architectural intensity was consistently bad news: floor area ratio showed a strong positive correlation with carbon emissions (r = 0.716 in winter), and the building rugosity index correlated strongly with both PM2.5 (r = 0.845) and PM10 (r = 0.759). The mechanism is twofold. Denser, taller construction directly increases energy demand and emissions, while rougher urban canopies modify airflow, restrict ventilation pathways, create recirculation zones, and prolong the time that particles linger in the air people breathe. Green landscapes told the opposite story. The percentage of vegetated land correlated negatively with PM2.5 at a remarkable r = −0.930, and patch cohesion, largest patch index, and net primary productivity all tracked lower pollution and lower emissions. Vegetation intercepts and deposits particles, stabilizes carbon uptake, and forms ecological corridors that aid dispersion. Notably, carbon emissions proved more sensitive to vegetation productivity, while particulate concentrations responded more strongly to the spatial morphology of green patches and the structural form of buildings.

Because averages can hide local variation, the researchers applied geographically weighted regression, which fits a separate set of coefficients at every location rather than one global average. The models explained the spatial variation well—adjusted R-squared values of 0.769 for carbon emissions, 0.680 for PM2.5, and 0.789 for PM10—and the local coefficients revealed pronounced spatial heterogeneity, with strong effects concentrated in the northern, southwestern, and northeastern parts of the region and weaker effects in the center and southeast. Spatial autocorrelation analysis using Moran’s I with 999 Monte Carlo simulations confirmed significant clustering for both variables, with particulate pollution showing stronger spatial clustering than carbon emissions, consistent with the regional transport of aerosols across county boundaries.

One of the study’s most elegant analyses focused on Yuncheng Salt Lake, a unique inland saline ecosystem ringed by thirty concentric 1-kilometer buffer zones extending 30 kilometers from the lake’s edge. Carbon emissions and particulate concentrations both declined steadily outward from the lake, tracking the fading intensity of building development. In the near zone, where floor area ratios and building coverage remained high, the link between construction and pollution was strongest; in the far zone, it essentially vanished, apart from a localized emissions peak at 29 kilometers traced to the industrial built-up area of Yongji City rather than a data artifact. Meanwhile, the negative regulatory effect of green space grew stronger with distance from the lake: fragmented green patches near the shore offered limited protection, while the large, connected, highly productive green landscapes of the outer suburbs delivered the strongest pollution-blocking and carbon-sinking benefits.

The practical implications are direct. Rather than treating smog and carbon as a single problem, the authors argue for differentiated landscape planning tuned to each county’s dominant burden. Areas prioritizing particulate control, such as Yonghe and Xi counties, should build protective forest belts and ventilation corridors along mountains and river valleys while strictly regulating building forms and floor area ratios. Areas prioritizing carbon reduction, such as Wenxi and Daning counties, should expand ecological green space and vegetation productivity while curbing construction intensity. Regions needing both, including Ruicheng, Yongji, Linyi, and Hejin, should weave core carbon-sink green spaces, dust-retention buffers, and ventilation corridors into a single ecological network alongside compact, low-carbon building layouts. Around the salt lake itself, the team proposes a gradient framework: strict lakeside development restrictions and greenway connections in the near zone, large ecological patches and industrial-residential separation in the middle zone, and a reinforced mountain-and-farmland ecological barrier in the far zone. The authors caution that their analysis covers only one winter heating season, includes only building and green space landscape types, and cannot fully disentangle meteorological and socioeconomic drivers—but the core message stands. In China’s coal country, and in resource-based regions across the Yellow River Basin, the cheapest combined attack on smog and carbon may already be growing in the ground: well-connected, productive green landscapes, deployed where the data say they will work hardest.

Subject of Research: Driving mechanisms of atmospheric particulate matter and carbon emissions under the influence of landscape patterns in southern Shanxi, China

Article Title: Analysis of the driving mechanisms of atmospheric particulate matter and carbon emissions under the influence of landscape patterns in southern Shanxi, China

Article References: Ren, W., Zhao, J., & Hu, F. (2026). Analysis of the driving mechanisms of atmospheric particulate matter and carbon emissions under the influence of landscape patterns in southern Shanxi, China. Frontiers in Environmental Science, 14, Article 1944157. https://doi.org/10.3389/fenvs.2026.1944157

Image Credits: AI Generated

DOI: 10.3389/fenvs.2026.1944157

Keywords: air pollution, carbon emissions, PM2.5, landscape patterns, urban green space, geographically weighted regression, remote sensing, Shanxi, Yellow River Basin, Yuncheng Salt Lake, winter heating season, spatial heterogeneity

The research team surveyed 384 male workers from a large aluminum plant in Shanxi Province between July and August 2024. Each participant provided fasting blood samples, from which the investigators measured fasting plasma glucose alongside the plasma concentrations of eight metals. Abnormally elevated blood glucose was defined according to World Health Organization diagnostic criteria as a fasting plasma glucose of at least 6.1 millimoles per liter, a threshold that captures both frank diabetes and the high-risk zone immediately beneath it. Because occupational cohorts are rarely uniform, the researchers also collected information on potential confounders, including age, body mass index, smoking, alcohol consumption, and other lifestyle and workplace characteristics, so that the statistical models could separate the metal signal from the noise of everyday habits and working conditions.

What makes this study technically interesting is not simply that it measured metals, but how it interrogated the mixture. Environmental epidemiology has long struggled with the problem of co-exposure: workers inhaling aluminum-laden dust are simultaneously exposed to a shifting ensemble of other elements, and the traditional one-metal-at-a-time analysis can miss both the joint effect and the individual contributions within it. The Shanxi team therefore deployed a battery of complementary statistical tools. Logistic regression provided the conventional single-metal estimates. Restricted cubic splines mapped the shape of each dose-response relationship without forcing it into a straight line. Least absolute shrinkage and selection operator regression, a machine-learning technique that penalizes weak predictors to zero, identified which metals mattered most. Weighted quantile sum regression estimated the overall mixture effect and apportioned it into weights for each component. Finally, Bayesian kernel machine regression modeled the mixture nonparametrically and probed for interactions between metals.

The single-metal results were striking. After adjustment for confounders, workers in the highest quartile of plasma aluminum had nearly three times the odds of abnormal fasting glucose compared with those in the lowest quartile, with an adjusted odds ratio of 2.93 and a 95 percent confidence interval of 1.16 to 7.40. Copper told a similar story: the top quartile carried an adjusted odds ratio of 2.71 (95 percent CI 1.17 to 6.27) relative to the bottom. Selenium, an element more often discussed as a protective micronutrient, was also implicated, with the highest quartile showing an odds ratio of 2.30 (95 percent CI 1.05 to 5.05). Restricted cubic spline analyses reinforced these findings, indicating positive and approximately linear dose-response relationships across the observed concentration ranges for all three metals, meaning the risk climbed steadily rather than appearing only at some threshold.

When the methods turned to the mixture as a whole, the picture sharpened rather than blurred. LASSO regression independently selected aluminum, selenium, and copper as the key metals in the panel, discarding the others as redundant. Weighted quantile sum regression then revealed a positive overall association between the mixed metal burden and abnormal blood glucose, and within that mixture aluminum contributed the largest share, with a weight of 0.434, followed closely by selenium at 0.375. In other words, roughly four-fifths of the mixture’s statistical weight in the model was carried by just two elements, one an unavoidable occupational exposure in an aluminum plant and the other an essential trace element whose excess appears anything but benign. Bayesian kernel machine regression confirmed positive exposure-response trends for the individual metals while finding no strong evidence of interactions, suggesting the metals act largely in parallel rather than amplifying one another.

The biology behind these associations is a subject of active research, and the study’s authors situate their findings within a substantial mechanistic literature. Oxidative stress is a central theme. Excess copper can catalyze the formation of reactive oxygen species, and laboratory work has shown that copper can mediate hydrogen peroxide production from the amylin peptide, a process proposed to contribute to the degeneration of insulin-producing islet cells in type 2 diabetes. Selenium presents a more paradoxical case. It is an essential component of antioxidant enzymes, yet several human studies, including randomized supplementation trials and dose-response meta-analyses, have linked higher selenium status to an increased risk of type 2 diabetes, and animal work has shown that high selenium can impair hepatic insulin sensitivity through dysregulated reactive oxygen species signaling. Aluminum, meanwhile, has been repeatedly associated in occupational cohorts with cognitive and metabolic disturbances, and earlier longitudinal work in northern China linked occupational aluminum exposure to changes in both blood pressure and blood glucose.

Context matters for interpreting the magnitude of the risk. Diabetes has become one of China’s defining health challenges, with epidemiological analyses in Nature Metabolism describing the scale of the epidemic and the Global Burden of Disease study projecting continued growth in prevalence through mid-century. Against that backdrop, identifying modifiable environmental contributors is a public health priority. Occupational cohorts are especially informative because exposure levels can be far above those seen in the general population, and because they can be reduced through engineering controls, personal protection, and workplace surveillance. The authors argue that their results justify targeted strengthening of occupational health protection in aluminum smelting, a recommendation that echoes earlier findings from the same research group on cognitive dysfunction among aluminum-exposed workers, where blood glucose itself appeared to mediate part of the effect of aluminum on cognition.

The study’s design imposes important caveats. It is cross-sectional, meaning metals and glucose were measured at the same time, so it cannot establish whether the metal exposures caused the metabolic disturbance or whether some shared factor, such as diet, kidney function, or disease-related changes in metal handling, drives the association. Reverse causation is a genuine possibility in trace element epidemiology, since impaired glucose metabolism can alter the transport and excretion of metals. The cohort was also limited to male workers at a single plant, which strengthens internal comparability but limits generalizability to women, to other industries, and to the wider population. Confidence intervals, while excluding the null, are wide for some estimates, reflecting the moderate sample size. The authors declare no competing financial interests, and the study was approved by the Medical Ethics Committee of Shanxi Medical University with written informed consent from all participants, but the data contain sensitive personal information and are available only from the corresponding author.

Even with those limitations, the methodological rigor of the mixture approach gives the findings unusual weight for a single cross-sectional study. Four independent techniques converged on the same trio of metals, and the dose-response curves were consistent with a monotonic relationship rather than a statistical fluke at one end of the distribution. For occupational physicians, the practical message is that glucose screening may deserve a place alongside the neurological and respiratory monitoring already routine in smelters, and that biomonitoring of aluminum, selenium, and copper could help identify workers whose cumulative metal burden places them at metabolic risk. For the broader field of environmental health, the study adds to a growing consensus that human exposure is inherently multi-element and that the health effects of any single metal can only be understood in the context of the mixture that accompanies it. As aluminum remains indispensable to transport, construction, and the renewable energy transition, the challenge is to keep the benefits of the metal while limiting its metabolic costs to the people who make it.

Subject of Research: Mixed metal exposure and abnormal fasting blood glucose among aluminum factory workers

Article Title: The association between mixed exposure to multiple metals and the risk of abnormally elevated blood glucose among aluminum factory workers

Article References: Zhang, Y., Xin, Y., Li, M., Xu, H., Guo, X., Wang, L., Zhang, H., Yin, J., Lu, X., Pan, B., & Song, J. (2026). The association between mixed exposure to multiple metals and the risk of abnormally elevated blood glucose among aluminum factory workers. Environmental Geochemistry and Health, 48(14), Article 564. https://doi.org/10.1007/s10653-026-03460-w

Image Credits: AI Generated

DOI: 10.1007/s10653-026-03460-w

Keywords: aluminum, selenium, copper, heavy metals, blood glucose, occupational health, aluminum smelting, metal mixture, diabetes risk, Bayesian kernel machine regression, weighted quantile sum regression, Shanxi