Transfer learning in environmental remote sensing

Remote Sensing of Environment, Journal Year: 2023, Volume and Issue: 301, P. 113924 - 113924

Published: Nov. 28, 2023

Language: Английский

---

Regional and global hotspots of arsenic contamination of topsoil identified by deep learning

Communications Earth & Environment, Journal Year: 2024, Volume and Issue: 5(1)

Published: Jan. 3, 2024

Abstract Topsoil arsenic (As) contamination threatens the ecological environment and human health. However, traditional methods for As identification rely on on-site sampling chemical analysis, which are cumbersome, time-consuming, costly. Here we developed a method combining visible near infrared spectra deep learning to predict topsoil content. We showed that optimum fully connected neural network model had high robustness generalization (R-Square values of 0.688 0.692 validation testing sets). Using model, relative content at regional global scales were estimated populations might potentially be affected determined. found China, Brazil, California As-contamination hotspots. Other areas, e.g., Gabon, although also great risk, rarely documented, making them potential Our results provided guidance regions require more detailed detection or timely soil remediation can assist in alleviating topsoil-As contamination.

Language: Английский

---

Soil organic matter content prediction based on two-branch convolutional neural network combining image and spectral features

Computers and Electronics in Agriculture, Journal Year: 2024, Volume and Issue: 217, P. 108561 - 108561

Published: Jan. 21, 2024

Language: Английский

---

An advanced soil organic carbon content prediction model via fused temporal-spatial-spectral (TSS) information based on machine learning and deep learning algorithms

Remote Sensing of Environment, Journal Year: 2022, Volume and Issue: 280, P. 113166 - 113166

Published: July 20, 2022

Language: Английский

---

Satellite Imagery to Map Topsoil Organic Carbon Content over Cultivated Areas: An Overview

Remote Sensing, Journal Year: 2022, Volume and Issue: 14(12), P. 2917 - 2917

Published: June 18, 2022

There is a need to update soil maps and monitor organic carbon (SOC) in the upper horizons or plough layer for enabling decision support land management, while complying with several policies, especially those favoring storage. This review paper dedicated satellite-based spectral approaches SOC assessment that have been achieved from satellite sensors, study scales geographical contexts past decade. Most relying on pure models carried out since 2019 dealt temperate croplands Europe, China North America at scale of small regions, some hundreds km2: dry combustion wet oxidation were analytical determination methods used 50% 35% satellite-derived studies, which measured topsoil contents mainly referred mineral soils, typically cambisols luvisols lesser extent, regosols, leptosols, stagnosols chernozems, annual cropping systems value ~15 g·kg−1 range 30 median. prediction limited preprocessing based bare pixel retrieval after Normalized Difference Vegetation Index (NDVI) thresholding. About one third these partial least squares regression (PLSR), another random forest (RF), remaining included machine learning such as vector (SVM). We did not find any studies either deep all-performance evaluations uncertainty analysis spatial model predictions. Nevertheless, literature examined here identifies information, derived under conditions, an interesting approach deserves further investigations. Future research includes considering simultaneous imagery acquired dates i.e., temporal mosaicking, testing influence possible disturbing factors mitigating their effects fusing mixed incorporating non-spectral ancillary information.

Language: Английский

---

Airborne hyperspectral imaging of cover crops through radiative transfer process-guided machine learning

Remote Sensing of Environment, Journal Year: 2022, Volume and Issue: 285, P. 113386 - 113386

Published: Dec. 8, 2022

Language: Английский

---

Mapping of soil organic matter in a typical black soil area using Landsat-8 synthetic images at different time periods

CATENA, Journal Year: 2023, Volume and Issue: 231, P. 107336 - 107336

Published: July 5, 2023

Language: Английский

---

A comparison of multiple deep learning methods for predicting soil organic carbon in Southern Xinjiang, China

Computers and Electronics in Agriculture, Journal Year: 2023, Volume and Issue: 212, P. 108067 - 108067

Published: July 21, 2023

Language: Английский

---

Estimation of soil inorganic carbon with visible near-infrared spectroscopy coupling of variable selection and deep learning in arid region of China

Geoderma, Journal Year: 2023, Volume and Issue: 437, P. 116589 - 116589

Published: July 7, 2023

Soil inorganic carbon (SIC) is the primary component of soil pool in arid and semiarid regions strongly impacts global cycle, ecosystem services, functions. The climate change intensify human activities, could substantially SIC, which highlights importance monitoring SIC. Rapid accurate estimation SIC concentration critical for monitoring. Currently, visible near-infrared (Vis-NIR) spectroscopy a promising technique estimating via rapid cost-effective manner. Thus, this study, we collected 315 topsoil samples from Alar Reclamation Area South Xinjiang, China, measured their Vis-NIR spectra content. Then, used deep learning algorithms, including one-dimensional convolutional neural network (1D-CNN), two-dimensional (2D-CNN), long short-term memory (LSTM), belief (DBN), combined with variable selection algorithms (particle swarm algorithm (PSO), interval random frog (IRF), competitive adaptive reweighting (CARS), ant colony (ACO), iteratively retaining informative variables (IRIV) to estimate Results showed that all five effectively extract featured spectral information reduce number by >97%, simplifying model structure. markedly improve accuracy, corresponding accuracy follows order: IRF > IRIV PSO CARS ACO. All four models have high prediction modeling each method follow LSTM 1D-CNN 2D-CNN DBN. achieved highest (R2 = 0.93, RMSE 1.26 g kg−1 calibration dataset; R2 0.92, 1.37 validation dataset). This study demonstrated can detect content quickly accurately.

Language: Английский

---

Spectral fusion modeling for soil organic carbon by a parallel input-convolutional neural network

Geoderma, Journal Year: 2023, Volume and Issue: 437, P. 116584 - 116584

Published: June 29, 2023

Visible-to-near-infrared (vis–NIR) and mid-infrared (MIR) spectroscopy have been widely utilized for the quantitative estimation of soil organic carbon (SOC). The fusion vis–NIR MIR data can be hypothesized to provide accurate reliable prediction SOC because spectral within a specific range each individual sensor may lack important absorptive features associated with SOC. In this study, six strategies, principally direct concatenation-partial least squares regression (DC-PLSR), outer product analysis-PLSR (OPA-PLSR), OPA-competitive adaptive reweighted sampling-PLSR (OPA-CARS-PLSR), sequentially orthogonalized-PLSR (SO-PLSR), DC-convolutional neural network (DC-CNN), parallel input-CNN (PI-CNN), were compared estimations models developed using samples collected from Zhejiang Province, East China, scanned under laboratory conditions both spectrophotometers. validation results (validation coefficient determination [R2] = 0.63–0.73) generally better than those R2 0.45–0.59). For fusion, best accuracy was achieved by PI-CNN 0.84), followed in descending order DC-CNN 0.78), SO-PLSR 0.73), OPA-CARS-PLSR 0.69), OPA-PLSR 0.66), DC-PLSR 0.64). performance over demonstrates necessity different sizes convolutional kernels before feeding into fully connected layers CNN fusing data. deep-learning method based on considered an efficient tool integrating multiple sensors estimating properties field modeling.

Language: Английский

---