Widespread soil contamination threatens living standards and weakens global efforts towards the Sustainable Development Goals (SDGs). Detailed soil mapping is needed to guide effective countermeasures and sustainable remediation operations. Here, we review visible and infrared reflectance spectroscopy (VIRS) based detection methods in combination with machine learning. To date, proximal, airborne and spaceborne carrier devices have been employed for soil contamination detection, allowing large areas to be covered at low cost and with minimal secondary environmental impact. In this way, soil contaminants can be monitored remotely, either directly or through correlation with soil components (e.g. Fe-oxides, soil organic matter, clay minerals). Observed vegetation reflectance spectra has also been proven an effective indicator for mapping soil pollution. Calibration models based on machine learning are used to interpret spectral data and predict soil contamination levels. The algorithms used for this include partial least squares regression, neural networks, and random forest. The processes underlying each of these approaches are outlined in this review. Finally, current challenges and future research directions are explored and discussed.

Dissolved oxygen (DO) is an effective indicator for water pollution. However, since DO is a non-optically active parameter and has little impact on the spectrum captured by satellite sensors, research on estimating DO by remote sensing at multiple spatiotemporal scales is limited. In this study, the support vector regression (SVR) models were developed and validated using the remote sensing reflectance derived from both Landsat and Moderate Resolution Imaging Spectroradiometer (MODIS) data and synchronous DO measurements (N = 188) and water temperature of Lake Huron and three other inland waterbodies (N = 282) covering latitude between 22–45 °N. Using the developed models, spatial distributions of the annual and monthly DO variability since 1984 and the annual monthly DO variability since 2000 in Lake Huron were reconstructed for the first time. The impacts of five climate factors on long-term DO trends were analyzed. Results showed that the developed SVR-based models had good robustness and generalization (average R² = 0.91, root mean square percentage error = 2.65%, mean absolute percentage error = 4.21%), and performed better than random forest and multiple linear regression. The monthly DO estimates by Landsat and MODIS data were highly consistent (average R² = 0.88). From 1984 to 2019, the oxygen loss in Lake Huron was 6.56%. Air temperature, incident shortwave radiation flux density, and precipitation were the main climate factors affecting annual DO of Lake Huron. This study demonstrated that using SVR-based models, Landsat and MODIS data could be used for long-term DO retrieval at multiple spatial and temporal scales. As data-driven models, combining spectrum and water temperature as well as extending the training set to cover more DO conditions could effectively improve model robustness and generalization.

The surface organic horizons in forest soils have been affected by air and soil pollutants, including potentially toxic elements (PTEs). Monitoring of PTEs requires a large number of samples and adequate analysis. Visible–near infrared (vis–NIR: 350–2500 nm) spectroscopy provides an alternative method to conventional laboratory measurements, which are time-consuming and expensive. However, vis–NIR spectroscopy relies on an empirical calibration of the target attribute to the spectra. This study examined the capability of vis–NIR spectra coupled with machine learning (ML) techniques (partial least squares regression (PLSR), support vector machine regression (SVMR), and random forest (RF)) and a deep learning (DL) approach called fully connected neural network (FNN) to assess selected PTEs (Cr, Cu, Pb, Zn, and Al) in forest organic horizons. The dataset consists of 2160 samples from 1080 sites in the forests over all the Czech Republic. At each site, we collected two samples from the fragmented (F) and humus (H) organic layers. The content of all PTEs was higher in horizon H compared to F horizon. Our results indicate that the reflectance of samples tended to decrease with increased PTEs concentration. Cr was the most accurately predicted element, regardless of the algorithm used. SVMR provided the best results for assessing the H horizon (R² = 0.88 and RMSE = 3.01 mg/kg for Cr). FNN produced the best predictions of Cr in the combined F + H layers (R² = 0.89 and RMSE = 2.95 mg/kg) possibly due to the larger number of samples. In the F horizon, the PTEs were not predicted adequately. The study shows that PTEs in forest soils of the Czech Republic can be accurately estimated with vis–NIR spectra and ML approaches. Results hint in availability of a large sample size, FNN provides better results.

In this study, size-fractionated particulate matters (PM) down to ultrafine (PM₀.₁) particles were collected using a cascade air sampler with a PM₀.₁ stage, in Hat Yai city, Songkhla province, southern Thailand during the year 2018. The particle-bound carbonaceous aerosols (CA) as elemental carbon (EC) and organic carbon (OC) were quantified with the thermal/optical reflectance method following the IMPROVE_TOR protocol. The concentrations of different temperature carbon fractions (OC1-OC4, EC1-EC3 and PyO) in the size-fractionated PM were evaluated to discern OC and EC correlations as well as those between char-EC and soot-EC. The results showed that biomass burning, motor vehicle, and secondary organic aerosols (SOC) all contributed to the size-fractionated PM. The OC/EC ratios ranged from 2.90 to 4.30 over the year, with the ratios of PM₂.₅₋₁₀ being the highest, except during the open biomass burning period. The concentration of CA was found to increase during the pre-monsoon season and had its peak value in the PM₀.₅₋₁.₀ fraction. The long-range transport of PMs from Indonesia, southwest of Thailand toward southern Thailand became more obvious during the pre-monsoon season. Transported plumes from biomass burning in Indonesia may increase the concentration of OC and EC both in the fine (PM₀.₅₋₁.₀ and PM₁.₀₋₂.₅) and coarse (PM₂.₅₋₁₀ and PM>₁₀) fractions. The OC fraction in PM₀.₁ was also shown to be significantly affected by the transported plumes during the pre-monsoon season. Good OC and EC correlations (R² = 0.824–0.915) in the fine particle fractions indicated that they had common sources such as fossil fuel combustion. However, the lower and moderate correlations (R² = 0.093–0.678) among the coarser particles suggesting that they have a more complex pattern of emission sources during the dry and monsoon seasons. This indicates the importance of focusing emission control strategies on different PM particle sizes in southern Thailand.

Growing evidence suggests that anthropogenic noise has deleterious effects on the behavior and physiology of free-living animals. These effects may be particularly pronounced early in life, when developmental trajectories are sensitive to stressors, yet studies investigating developmental effects of noise exposure in free-living populations remain scarce. To elucidate the effects of noise exposure during development, we examined whether noise exposure is associated with shorter telomeres, duller carotenoid-based coloration and reduced body mass in nestlings of a common urban bird, the great tit (Parus major). We also assessed how the noise environment is related to reproductive success. We obtained long-term measurements of the noise environment, over a ∼24-h period, and characterized both the amplitude (measured by LAₑq, LA₉₀, LA₁₀, LAₘₐₓ) and variance in noise levels, since more stochastic, as well as louder, noise regimes might be more likely to induce stress. In our urban population, noise levels varied substantially, with louder, but less variable, noise characteristic of areas adjacent to a highway. Noise levels were also highly repeatable, suggesting that individuals experience consistent differences in noise exposure. The amplitude of noise near nest boxes was associated with shorter telomeres among smaller, but not larger, brood members. In addition, carotenoid chroma and hue were positively associated with variance in average and maximum noise levels, and average reflectance was negatively associated with variance in background noise. Independent of noise, hue was positively related to telomere length. Nestling mass and reproductive success were unaffected by noise exposure. Results indicate that multiple dimensions of the noise environment, or factors associated with the noise environment, could affect the phenotype of developing organisms, that noise exposure, or correlated variables, might have the strongest effects on sensitive groups of individuals, and that carotenoid hue could serve as a signal of early-life telomere length.

The concept of the smart city raised several technological and scientific issues including light pollution. There are various negative impacts of light pollution on economy, ecology, and heath. This paper deals with the census of the colour of light emitted by lamps used in a city environment. To this end, we derive a light bulb colour estimator based on Bayesian reasoning, directional data, and image formation model in which the usual concept of reflectance is not used. All choices we made are devoted to designing an algorithm which can be run almost in real-time. Experimental results show the effectiveness of the proposed approach.

The continuous release of persistent iron-cyanide (Fe-CN) complexes from various industrial sources poses a high hazard to the environment and indicates the necessity to analyze a considerable amount of samples. Conventional flow injection analysis (FIA) is a time and cost consuming method for cyanide (CN) determination. Thus, a rapid and economic alternative needs to be developed to quantify the Fe-CN complexes. 52 soil samples were collected at a former Manufactured Gas Plant (MGP) site in order to determine the feasibility of diffuse reflectance infrared Fourier spectroscopy (DRIFTS). Soil analysis revealed CN concentrations in a range from 8 to 14.809 mg kg−1, where 97% was in the solid form (Fe4[Fe(CN)6]3), which is characterized by a single symmetrical CN band in the range 2092–2084 cm−1. The partial least squares (PLS) calibration-validation model revealed IR response to CNtot which exceeds 2306 mg kg−1 (limit of detection, LOD). Leave-one-out cross-validation (LOO-CV) was performed on soil samples, which contained low CNtot (<900 mg kg−1). This improved the sensitivity of the model by reducing the LOD to 154 mg kg−1. Finally, the LOO-CV conducted on the samples with CNtot > 900 mg kg−1 resulted in LOD equal to 3751 mg kg−1. It was found that FTIR spectroscopy provides the information concerning different CN species in the soil samples. Additionally, it is suitable for quantifying Fe-CN species in matrixes with CNtot > 154 mg kg−1. Thus, FTIR spectroscopy, in combination with the statistical approach applied here seems to be a feasible and quick method for screening of contaminated sites.

The reflectometry of leaf asymmetry is a novel approach in the bio-monitoring of tree health in urban or industrial habitats. Leaf asymmetry responds to the degree of environmental pollution and reflects structural changes in a leaf due to environmental pollution. This paper describes the boundary conditions to scale up from leaf to canopy level reflectance, by describing the variability of adaxial and abaxial leaf reflectance, hence leaf asymmetry, along the crown height gradients of two tree species. Our findings open a research pathway towards bio-monitoring based on the airborne remote sensing of tree canopies and their leaf asymmetric properties.

The objective of this paper is to give an account of the evaluation of the effect of urban habitat quality on dorsi-ventral leaf reflectance asymmetry to bio-monitor urban habitat pollution. Reflectance in the RGB bands of a reflex camera is measured at the adaxial and abaxial sides of Carpinus betulus L. leaves for two contrasting urban habitats, e.g.; suburban green and industrial habitats in the city of Gent (Belgium). Abaxial leaf reflectance is consistently higher than adaxial leaf reflectance. We quantified leaf dorsi-ventral reflectance asymmetry with a newly defined Normalized Dorsi-ventral Asymmetry Index (NDAI). The NDAI is significantly higher in industrial habitats as opposed to suburban green ones. Our optical observations indicate that changes in Carpinus betulus L. leaf morphology are related to urban habitat quality. Hence, we suggest that leaf dorsi-ventral reflectance asymmetry allows the estimation of the magnitude and spatial extent of environmental pollution in urban environments.

Intra-urban pollution monitoring requires fine particulate (PM₂.₅) concentration mapping at ultrahigh-resolution (dozens to hundreds of meters). However, current PM₂.₅ concentration estimation, which is mainly based on aerosol optical depth (AOD) and meteorological data, usually had a low spatial resolution (kilometers) and severe spatial missing problem, cannot be applied to intra-urban pollution monitoring. To solve these problems, top-of-atmosphere reflectance (TOAR), which contains both the information about land and atmosphere and has high resolution and large spatial coverage, may be efficiently used for PM₂.₅ estimation. This study aims to systematically evaluate the feasibility of retrieving ultrahigh-resolution PM₂.₅ concentration at a large scale (national level) from TOAR. Firstly, we make a detailed discussion about several important but unsolved theoretic problems on TOAR-based PM₂.₅ retrieval, including the band selection, scale effect, cloud impact, and mapping quality evaluation. Secondly, four types and eight retrieval models are compared in terms of quantitative accuracy, mapping quality, model generalization, and model efficiency, with the pros and cons of each type summarized. Deep neural network (DNN) model shows the highest retrieval accuracy, and linear models were the best in efficiency and generalization. As a compromise, ensemble learning shows the best overall performance. Thirdly, using the highly accurate DNN model (cross-validated R² equals 0.93) and through combining Landsat 8 and Sentinel 2 images, a 90 m and ∼4-day resolution PM₂.₅ product was generated. The retrieved maps were used for analyzing the fine-scale interannual pollution change inner the city and the pollution variations during novel coronavirus disease 2019 (COVID-19). Results of this study proves that ultrahigh resolution can bring new findings of intra-urban pollution change, which cannot be observed at previous coarse resolution. Lastly, some suggestions for future ultrahigh-resolution PM₂.₅ mapping research were given.