Potential of Hyperspectral Data Combined With Optimal Band Combination Algorithm for Estimating Soil Organic Carbon Content in Lakeside Oasis

Land Degradation and Development, Journal Year: 2024, Volume and Issue: unknown

Published: Oct. 9, 2024

ABSTRACT Accurate estimation of soil organic carbon (SOC) content is essential for promoting regional sustainable agriculture and improving land quality. Visible near‐infrared (Vis‐NIR) near‐Earth remote sensing spectroscopy has become an effective alternative to the traditional time‐consuming costly methods due its high‐resolution nondestructive application, but it vulnerable redundancy spectral information overlap between bands. This study delves into potential optimal parameters estimating SOC in arid lakeside oases, using Bosten Lake Xinjiang, China, as a focal point. Soil samples (0–10 cm, 10–20 20–30 30–40 cm) were collected, their hyperspectral reflectance measured. The data underwent preprocessing techniques, including continuum removal (CR), standard normal variate (SNV), continuous wavelet transform (CWT). was predicted back propagation neural network models constructed based on one‐dimensional (1D), two‐dimensional (2D), three‐dimensional (3D) correlation coefficients. Results showcased effectiveness CWT method accentuating enhancing variable correlation. Among indices, 3D exhibited highest performance ( R 2 = 0.82, RPD 2.02 TDI‐1 at 0–10 cm; 0.85, 2.28 TDI‐2 0.83, 2.24 0.86, 2.53 TDI‐4 cm), followed by 2D then 1D. These insights offer guidance future strategies index determination, facilitating spatial distribution mapping advancing agricultural planning. They also have implications determining interpolation, which would contribute planning development.

Language: Английский

---

Fine Resolution Mapping of Soil Organic Carbon in Croplands with Feature Selection and Machine Learning in Northeast Plain China

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(20), P. 5033 - 5033

Published: Oct. 20, 2023

Unsustainable human management has negative effects on cropland soil organic carbon (SOC), causing a decrease in health and the emission of greenhouse gas. Due to contiguous fields, large-scale mechanized operations are widely used Northeast China Plain, which greatly improves production efficiency while decreasing quality, especially for SOC. Therefore, an up-to-date SOC map is needed estimate after long-term cultivation inform better land management. Using Quantile Regression Forest, total 396 samples from 132 sampling sites at three depth intervals 40 environmental covariates (e.g., Landsat 8 spectral indices, WorldClim 2 MODIS products) selected by Boruta feature selection algorithm were spatial distribution Plain 90 m resolution. The results showed that increased overall southern area northern area, with average 17.34 g kg−1 plough layer (PL) 13.92 compacted (CL). At vertical scale, decreased, depths getting deeper. PL CL was 3.41 kg−1. Climate (i.e., temperature, daytime nighttime surface mean temperature driest quarter) dominant controlling factor, followed position oblique geographic coordinate 105°), organism variance net primary productivity non-crop period). uncertainty 1.04 1.07 CL. high appeared relatively scattered altitudes, complex landforms. This study updated resolution maps scales, clarifies influence provides reference conservation policy-making.

Language: Английский

---

An integrated feature selection approach to high water stress yield prediction

Frontiers in Plant Science, Journal Year: 2023, Volume and Issue: 14

Published: Dec. 4, 2023

The timely and precise prediction of winter wheat yield plays a critical role in understanding food supply dynamics ensuring global security. In recent years, the application unmanned aerial remote sensing has significantly advanced agricultural research. This led to emergence numerous vegetation indices that are sensitive variations. However, not all these universally suitable for predicting yields across different environments crop types. Consequently, process feature selection index sets becomes essential enhance performance models. study aims develop an integrated method known as PCRF-RFE, with focus on selection. Initially, building upon prior research, we acquired multispectral images during flowering grain filling stages identified 35 yield-sensitive indices. We then applied Pearson correlation coefficient (PC) random forest importance (RF) methods select relevant features set. Feature filtering thresholds were set at 0.53 1.9 respective methods. union selected by both was used recursive elimination (RFE), ultimately yielding optimal subset constructing Cubist Recurrent Neural Network (RNN) results this demonstrate model, constructed using obtained through (PCRF-RFE), consistently outperformed RNN model. It exhibited highest accuracy stages, surpassing models or subsets derived from single method. confirms efficacy PCRF-RFE offers valuable insights references future research realms studies.

Language: Английский

---

Soil Organic Carbon Prediction Based on Vis–NIR Spectral Classification Data Using GWPCA–FCM Algorithm

Sensors, Journal Year: 2024, Volume and Issue: 24(15), P. 4930 - 4930

Published: July 30, 2024

Soil visible and near-infrared reflectance spectroscopy is an effective tool for the rapid estimation of soil organic carbon (SOC). The development spectroscopic technology has increased application spectral libraries SOC research. However, direct prediction remains challenging due to high variability in types soil-forming factors. This study aims address this challenge by improving accuracy through classification. We utilized European Land Use Cover Area frame Survey (LUCAS) large-scale library employed a geographically weighted principal component analysis (GWPCA) combined with fuzzy c-means (FCM) clustering algorithm classify spectra. Subsequently, we used partial least squares regression (PLSR) Cubist model prediction. Additionally, classified data land cover compared classification results those obtained from showed that (1) GWPCA-FCM-Cubist yielded best predictions, average R2 = 0.83 RPIQ 2.95, representing improvements 10.33% 18.00% RPIQ, respectively, unclassified full sample modeling. (2) modeling based on GWPCA-FCM was significantly superior type Specifically, there 7.64% 14.22% improvement under PLSR, 13.36% 29.10% Cubist. (3) Overall, models better than PLSR models. These findings indicate GWPCA FCM conjunction technique can enhance libraries.

Language: Английский

---