Ensemble and transfer learning of soil inorganic carbon with visible near-infrared spectra

Geoderma, Journal Year: 2025, Volume and Issue: 456, P. 117257 - 117257

Published: March 15, 2025

Language: Английский

---

Integration of multimodal data for large-scale rapid agricultural land evaluation using machine learning and deep learning approaches

Geoderma, Journal Year: 2023, Volume and Issue: 439, P. 116696 - 116696

Published: Oct. 25, 2023

Rapid and accurate agricultural land evaluation provides essential guidance for the supervision allocation of resources; it also helps to ensure food security. Previous work has mainly evaluated quality at county level by using field sampling data based on a factor approach. However, is difficult achieve uniform, large-scale via conventional approaches because its spatial heterogeneity, as well large temporal economic costs associated with acquisition. In this study, we integrated publicly available multimodal (i.e., satellite remote sensing, environmental, socioeconomic data) into Google Earth Engine (GEE) platform, selected best indicators from each modality geodetector, basis which different combinations input models were designed. And then developed machine learning (random forest, RF) deep (deep neural network, DNN) evaluate in paddy dry systems 2013 throughout Guangdong Province, China. The results showed that performance our combination variables decreased following order: > bimodal unimodal. With combination, RF model (R2 = 0.91, RMSE 97.56, CCC 0.95) outperformed DNN 0.89, 108.72, 0.94) terms predicting field. 0.90, 104.27, 0.86, 124.38, 0.93) land. estimates obtained more than greater homogeneity fields. This research proposed simple, low-cost rapid provincial scale data, can help control grade multiple scales.

Language: Английский

---

Improving model performance in mapping cropland soil organic matter using time-series remote sensing data

Journal of Integrative Agriculture, Journal Year: 2024, Volume and Issue: 23(8), P. 2820 - 2841

Published: Jan. 9, 2024

Faced with increasing global soil degradation, spatially explicit data on cropland organic matter (SOM) provides crucial for carbon pool accounting, quality assessment and the formulation of effective management policies. As a spatial information prediction technique, digital mapping (DSM) has been widely used to map at different scales. However, accuracy SOM maps is typically lower than other land cover types due inherent difficulty in precisely quantifying human disturbance. To overcome this limitation, study systematically assessed framework "information extraction-feature selection-model averaging" improving model performance using 462 samples collected Guangzhou, China 2021. The results showed that dynamic extraction, feature selection averaging could efficiently improve final predictions (R2: 0.48 0.53) without having obviously negative impacts uncertainty. Quantifying environment was an efficient way generate covariates are linearly nonlinearly related SOM, which improved R2 random forest from 0.44 extreme gradient boosting 0.37 0.43. FRFS recommended when there relatively few environmental (<200), whereas Boruta many (>500). granger-ramanathan approach average When structures initial models similar, number did not have significantly positive effects predictions. Given advantages these selected strategies over great potential high-accuracy any scales, so can provide more reliable references conservation policy-making.

Language: Английский

---

Enhanced Estimation of Rice Leaf Nitrogen Content via the Integration of Hybrid Preferred Features and Deep Learning Methodologies

Agronomy, Journal Year: 2024, Volume and Issue: 14(6), P. 1248 - 1248

Published: June 9, 2024

Efficiently obtaining leaf nitrogen content (LNC) in rice to monitor the nutritional health status is crucial achieving precision fertilization on demand. Unmanned aerial vehicle (UAV)-based hyperspectral technology an important tool for determining LNC. However, intricate coupling between spectral information and remains elusive. To address this, this study proposed estimation method LNC that integrates hybrid preferred features with deep learning modeling algorithms based UAV imagery. The approach leverages XGBoost, Pearson correlation coefficient (PCC), a synergistic combination of both identify characteristic variables estimation. We then construct models using statistical regression methods (partial least-squares (PLSR)) machine (random forest (RF); neural networks (DNN)). optimal model utilized map spatial distribution at field scale. was conducted National Agricultural Science Technology Park, Guangzhou, located Baiyun District Guangdong, China. results reveal combined PCC-XGBoost algorithm significantly enhances accuracy inversion compared standalone screening approach. Notably, built DNN exhibits highest predictive performance demonstrates great potential mapping This indicates role enhancement utilization efficiency cultivation. outcomes offer valuable reference enhancing agricultural practices sustainable crop management.

Language: Английский

---

Precision agriculture technologies for soil site-specific nutrient management: A comprehensive review

Artificial Intelligence in Agriculture, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 1, 2025

Language: Английский

---

National-scale mapping topsoil organic carbon of cropland in China using multitemporal Sentinel-2 images

Geoderma, Journal Year: 2025, Volume and Issue: 456, P. 117272 - 117272

Published: March 30, 2025

Language: Английский

---

Estimating Soil Attributes for Yield Gap Reduction in Africa Using Hyperspectral Remote Sensing Data with Artificial Intelligence Methods: An Extensive Review and Synthesis

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(9), P. 1597 - 1597

Published: April 30, 2025

Africa’s rapidly growing population is driving unprecedented demands on agricultural production systems. However, yields in Africa are far below their potential. One of the challenges leading to low productivity Africa‘s poor soil quality. Effective fertility management an essential key factor for optimizing while ensuring environmental sustainability. Key properties—such as organic carbon (SOC), nutrient levels (i.e., nitrogen (N), phosphorus (P), potassium (K), moisture retention (MR) or content (MC), and texture (clay, sand, loam fractions)—are critical factors influencing crop yield. In this context, study conducts extensive literature review use hyperspectral remote sensing technologies, with a particular focus freely accessible data (e.g., PRISMA, EnMAP), well evaluation advanced Artificial Intelligence (AI) models analyzing processing spectral map attributes. More specifically, examined progress applying technologies monitoring mapping properties over last 15 years (2008–2024). Our results demonstrated that (i) only very few studies have explored high-resolution sensors satellite sensors) property Africa; (ii) there considerable value AI approaches estimating attributes, strong recommendation further explore potential deep learning techniques; (iii) despite advancements AI-based methodologies availability sensors, combined application remains underexplored African context. To our knowledge, no yet integrated these Africa. This also highlights adopting encompassing both imaging spectroscopy) enhance accurate Africa, thereby constituting base addressing question yield gap.

Language: Английский

---

A two-dimensional bare soil separation framework using multi-temporal Sentinel-2 images across China

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2024, Volume and Issue: 134, P. 104181 - 104181

Published: Sept. 30, 2024

Language: Английский

---

Improving the Accuracy of Soil Classification by Using Vis–NIR, MIR, and Their Spectra Fusion

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(9), P. 1524 - 1524

Published: April 25, 2025

Soil spectroscopy offers a rapid, cost-effective alternative to traditional soil analyses for characterization and classification. Previous studies have mainly focused on predicting categories using single sensors, particularly visible–near-infrared (vis–NIR) or mid-infrared (MIR) spectroscopy. In this study, we evaluated the performance of vis–NIR, MIR, their combined spectra classification by partial least-squares discriminant analysis (PLSDA) random forest (RF). Utilizing 60 typical profiles’ data four classes from global spectral library (GSSL), our results demonstrated that in PLSDA models, direct combination (optimal overall accuracy: 70.6%, kappa coefficient: 0.60) outer product (OPA) fused 68.1%, 0.57) outperformed vis–NIR 62.2%, 0.49) but underperformed compared MIR 71.4%, 0.62). RF accuracy was inferior ranges, with achieving highest 89.1%, 0.85). Therefore, alone remains most effective range accurate class discrimination. Our findings highlight potential enhancing efficiency, important implications resource management agricultural planning across diverse environments.

Language: Английский

---

Which and How Many Soil Sensors are Ideal to Predict Key Soil Properties: A Case Study with Seven Sensors

Published: Jan. 1, 2024

Soil sensing enables rapid and cost-effective soil analysis. However, a single sensor often does not generate enough information to reliably predict wide range of properties. Within case-study, our objective was identify how many which combinations sensors prove be suitable for high-resolution mapping. On subplot an agricultural field showing high spatial variability, six in-situ proximal (PSSs) next remote (RS) data from Sentinel-2 were evaluated based on their capabilities set properties including: organic matter, pH, moisture as well plant-available phosphorus, magnesium potassium. The PSSs consisted ion-selective pH electrodes, capacitive sensor, apparent electrical conductivity measuring system passive gamma-ray-, X-ray fluorescence- near-infrared spectroscopy. All possible exhaustively ranked prediction performances. Over all properties, fusion demonstrated considerable increase in accuracy. Five out predicted with R2 ≥ 0.80 the best model. Nonetheless, improvement derived fusing increasing number subject diminishing returns. Sometimes adding more even decreased performances specific Gamma-ray spectroscopy most effective, both or combination other sensors. As RS outperformed three PSSs. showed especially potential but limited benefit when multiple fused.

Language: Английский

---