Prediction of wetland soil carbon storage based on near infrared hyperspectral imaging and deep learning

Infrared Physics & Technology, Год журнала: 2024, Номер 139, С. 105287 - 105287

Опубликована: Апрель 6, 2024

Язык: Английский

---

Spatial Distribution of Soil Heavy Metal Concentrations in Road-Neighboring Areas Using UAV-Based Hyperspectral Remote Sensing and GIS Technology

Sustainability, Год журнала: 2023, Номер 15(13), С. 10043 - 10043

Опубликована: Июнь 25, 2023

Monitoring and restoring soil quality in areas neighboring roads affected by traffic activities require a thorough investigation of heavy metal concentrations. This study examines the spatial heterogeneity copper (Cu) chromium (Cr) concentrations 0.113 km² area adjacent to Jin-Long Avenue Wuhan, China, using Unmanned Aerial Vehicle (UAV)-based hyperspectral remote sensing technology. Through this UAV-based technology, we innovatively achieve small-scale fine-grained analysis pollution related with activities, which represents major contribution research study. In our approach, generated 4375 spectral variates transforming original spectrum. To enhance result accuracy, applied Boruta algorithm correlation select optimal variates. We developed retrieval model Gradient Boosting Decision Tree (GBDT) regression method, selected from set four methods LOOCV method. The resulting yielded R-square values 0.325 0.351 for Cu Cr, respectively, providing valuable insights into Based on retrieved bare pixels (17,420 points), analyzed relationship between perpendicular distance road. Additionally, employed universal kriging interpolation method map across entire area. Our findings reveal that concentration metals exceeds background decreases as road increases. significantly contributes understanding distribution characteristics caused activities.

Язык: Английский

---

A study on hyperspectral soil total nitrogen inversion using a hybrid deep learning model CBiResNet-BiLSTM

Chemical and Biological Technologies in Agriculture, Год журнала: 2024, Номер 11(1)

Опубликована: Окт. 18, 2024

Rapid, accurate and non-destructive acquisition of soil total nitrogen (TN) content in the black zone is significant for achieving precise fertilization. In this study, types corn soybean fields Jilin Agricultural University, China, were selected as study area. A 162 samples collected using a five-point mixed sampling method. Then, spectral data obtained noisy edge initially eliminated. Subsequently, denoised underwent smoothing by Savitzky–Golay (SG) After performing first-order difference (FD) second-order (SD) transformations on data, it was input to model. hybrid deep learning model, CBiResNet-BiLSTM, designed prediction TN content. This model optimized based ResNet34, its capabilities enhanced incorporating CBAM residual module facilitate additional eigenvalue extraction. Also, Bidirectional Long Short-Term Memory (BiLSTM) integrated enhance accuracy. Besides, partial least squares regression (PLSR), random forest (RFR), support vector machine (SVR), back propagation neural network (BP), well ResNet(18, 34, 50, 101, 152) models taken comparative experiments. The results indicated that traditional PLSR achieved good performance, with R2 0.883, CBiResNet-BiLSTM had best inversion capability 0.937, being improved 5.4%, compared On basis, we present LUCAS dataset demonstrate generalisability Therefore, fast feasible hyperspectral estimation method

Язык: Английский

---

Removing DOM from chloride modified hydrochar could improve Cu2+ adsorption capacity from aqueous solution

Chemosphere, Год журнала: 2023, Номер 342, С. 140202 - 140202

Опубликована: Сен. 16, 2023

Язык: Английский

---

Machine learning for open-pit mining: a systematic review

International Journal of Mining Reclamation and Environment, Год журнала: 2024, Номер unknown, С. 1 - 39

Опубликована: Июнь 20, 2024

Язык: Английский

---

Applications and Challenges Related to the Use of Unmanned Aircraft Systems in Environment Monitoring

Опубликована: Янв. 1, 2024

This chapter gives an overview of the latest research and development activities conducted by VTT regarding environmental monitoring using unmanned aircraft systems (UAS) discusses associated challenges. An AI-based drone swarm technology in a unified framework can provide situational awareness decision support tools for wildfire monitoring. The floating waste from (UA) with optical sensors suggests that multi-imaging near-infrared (NIR) hyperspectral (HS), thermal infrared (TIR), multicolor (RGB) is promising method separating plastic organic material. Monitoring tailing ponds mines onboard multispectral indicated hints seepage or water spectral signatures vegetation ground along general structural information, particularly pond dams. Hyperspectral data acquired UAS well suited vegetation's biochemical composition, moisture content, biodiversity since it offers unprecedented spatial resolution pixel sizes comparable to basic elements, leaves flowers. demonstrated applicability novel analysis algorithms based on theory invariant such ultra-high-resolution HS imagery trait retrieval. challenges related use are multifaceted. These include connectivity technologies protocols, operational limitations UA, application artificial intelligence (AI), fusion, machine learning methods. Also, legislative demand autonomous operations, significantly beyond visual line sight (BVLOS), requires range U-space services.

Язык: Английский

---

Estimation of the dolomite content of carbonate rock outcrops based on spectral knowledge and machine learning

Frontiers in Earth Science, Год журнала: 2024, Номер 12

Опубликована: Июнь 11, 2024

Accurately estimating the dolomite content in carbonate rocks is crucial for optimizing oil and gas exploration production strategies. Hyperspectral techniques have advantages terms of efficiency, cost-effectiveness, non-destructiveness compared with traditional laboratory methods. Despite abundance hyperspectral data, feature selection extraction remain challenging. In this study, data collected from surface outcrop field using analytical spectral device (ASD) were applied to construct model content. Firstly, preprocessed via outlier analysis continuum transformation. Next, a hybrid approach integrating knowledge machine learning was proposed facilitate efficient precise data; approach, preliminary screening based on followed by further random forest algorithm. The selected features then combined support vector regression algorithm obtain estimation model. Finally, accuracy evaluated samples. To verify effectiveness method, various combinations eight input variables four algorithms compared. Among all combinations, our achieved highest test R 2 value 0.91 root-mean-square error only 0.122. method practical provides quantitative geologists identify mineral distribution outcrops. Thus, robust understanding reservoir characteristics has significant geological surveys exploration.

Язык: Английский

---

Prediction of copper contamination in soil across EU using spectroscopy and machine learning: handling class imbalance problem

Smart Agricultural Technology, Год журнала: 2024, Номер unknown, С. 100728 - 100728

Опубликована: Дек. 1, 2024

Язык: Английский

---

Synthesis and Selective Adsorption Performance of Cu(II) Imprinted SA/CMC/AM Microspheres in Multi‐Ion Coexistence Environments

Journal of Applied Polymer Science, Год журнала: 2024, Номер unknown

Опубликована: Дек. 26, 2024

ABSTRACT To achieve efficient and selective removal of Cu 2+ from a multi‐ion coexistence environment, ‐imprinted SA/CMC/AM microspheres (IMSCA) were synthesized. The adsorption capacity efficiency the material under different preparation conditions investigated. surface morphology functional groups IMSCA characterized analyzed using SEM, FTIR, XRD. SEM images revealed relatively smooth with specific pores uniform structure. Special peaks appeared in FTIR spectrum IMSCA, indicating possibility imprinting process presence sites. semicrystalline structure exhibited by imprinted XRD characterization further reinforces likelihood In terms kinetics, followed pseudo–second‐order kinetic model, suggesting that chemical was dominant during process. isotherms, Langmuir model better fitted experimental data, monolayer adsorption. mixed solutions multiple metal ions. After 10 adsorption–desorption cycles, remained at approximately 90%. Compared other materials, exhibits higher capacity, faster elution rate, superior reusability.

Язык: Английский

---

Integrated methodology to link geochemical and geophysical-lab data in a geophysical investigation of a slag heap for resource quantification

Journal of Environmental Management, Год журнала: 2023, Номер 349, С. 119366 - 119366

Опубликована: Окт. 27, 2023

The increasing need to find alternative stocks of critical raw materials drives revisit the residues generated during former production mineral and metallic materials. Geophysical methods contribute sustainable characterization metallurgical inferring on their composition, zonation volume(s) estimation. Nevertheless, more quantitative approaches are needed link geochemical or mineralogical analyses with geophysical data. In this contribution, we describe a methodology that integrates laboratory measurements interpret field data solving classification problem. final aim is estimate different types assess potential resource recovery. We illustrate slag heap composed from iron steel factory. First, carried out 3D acquisition using electrical resistivity tomography (ERT) induced polarization (IP), based which, sampling survey was designed. conducted ERT, IP, spectral (SIP), X-ray fluorescence analysis, 4 groups chemical composition were identified. Then probabilistic data, 2D kernel density estimators (for each group) fitted inverted collocated samples. estimated volumes model were: 4.17 × 103 m3 ± 12 %, 1.888 105 59.4 19 2.30 104 21% for ordered an content. uncertainty ranges derived comparing without considering probabilities associated classification. found representative definition KDE bandwidths defining elements in ultimately estimation volumes. This suitable quantitatively terms materials, integrating uncertainties both Furthermore, several crucial investigation could be applied real case study, e.g., acquisition, lab measurements.

Язык: Английский

---