Mapping of Clay Montmorillonite Abundance in Agricultural Fields Using Unmixing Methods at Centimeter Scale Hyperspectral Images DOI Creative Commons
Etienne Ducasse, Karine Adeline, Audrey Hohmann

и другие.

Remote Sensing, Год журнала: 2024, Номер 16(17), С. 3211 - 3211

Опубликована: Авг. 30, 2024

The composition of clay minerals in soils, and more particularly the presence montmorillonite (as part smectite family), is a key factor soil swell–shrinking as well off–road vehicle mobility. Detecting these topsoil quantifying abundance are challenge since they usually intimately mixed with other minerals, organic carbon moisture content. Imaging spectroscopy coupled unmixing methods can address issues, but quality estimation degrades coarser spatial resolution due to pixel heterogeneity. With advent UAV-borne proximal hyperspectral acquisitions, it now possible acquire images at centimeter scale. Thus, objective this paper evaluate accuracy limitations retrieve from very-high-resolution (1.5 cm) acquired camera installed on top bucket truck over three different agricultural fields, Loiret department, France. Two automatic endmember detection based assumption that materials linearly mixed, namely Simplex Identification via Split Augmented Lagrangian (SISAL) Minimum Volume Constrained Non-negative Matrix Factorization (MVC-NMF), were tested prior unmixing. Then, two linear methods, fully constrained least square method (FCLS) multiple spectral mixture analysis (MESMA), nonlinear ones, generalized bilinear (GBM) multi-linear model (MLM), performed images. In addition, several preprocessings applied order improve performances. Results showed our selected not suitable context. However, endmembers taken available libraries successfully. method, MLM, without preprocessing or application first Savitzky–Golay derivative, gave best accuracies for using USGS library (RMSE between 2.2–13.3% 1.4–19.7%). Furthermore, significant impact estimations scale was majority (i) high variability composition, (ii) roughness inducing large variations illumination conditions surface scatterings (iii) volume coming intimate mixture. Finally, results offer new opportunity mapping expansive soils imaging very resolution.

Язык: Английский

Monitoring and Modeling the Soil‐Plant System Toward Understanding Soil Health DOI Creative Commons
Yijian Zeng, Anne Verhoef, Harry Vereecken

и другие.

Reviews of Geophysics, Год журнала: 2025, Номер 63(1)

Опубликована: Янв. 25, 2025

Abstract The soil health assessment has evolved from focusing primarily on agricultural productivity to an integrated evaluation of biota and biotic processes that impact properties. Consequently, shifted a predominantly physicochemical approach incorporating ecological, biological molecular microbiology indicators. This shift enables comprehensive exploration microbial community properties their responses environmental changes arising climate change anthropogenic disturbances. Despite the increasing availability indicators (physical, chemical, biological) data, holistic mechanistic linkage not yet been fully established between functions across multiple spatiotemporal scales. article reviews state‐of‐the‐art monitoring, understanding how soil‐microbiome‐plant contribute feedback mechanisms causes in properties, as well these have functions. Furthermore, we survey opportunities afforded by soil‐plant digital twin approach, integrative framework amalgamates process‐based models, Earth Observation data assimilation, physics‐informed machine learning, achieve nuanced comprehension health. review delineates prospective trajectory for monitoring embracing systematically observe model system. We further identify gaps opportunities, provide perspectives future research enhanced intricate interplay hydrological processes, hydraulics, microbiome, landscape genomics.

Язык: Английский

Процитировано

1
Developing a near-infrared spectroscopy calibration algorithm for soil organic carbon content in South Africa DOI Creative Commons
Willie Herman Cloete, Gerhard Du Preez, George van Zijl

и другие.

Soil Advances, Год журнала: 2025, Номер 3, С. 100039 - 100039

Опубликована: Фев. 21, 2025

Язык: Английский

Процитировано

0
Federated learning applications in soil spectroscopy DOI Creative Commons

Giannis Gallios,

Nikolaos Tsakiridis, Nikolaos Tziolas

и другие.

Geoderma, Год журнала: 2025, Номер 456, С. 117259 - 117259

Опубликована: Март 25, 2025

Язык: Английский

Процитировано

0
Development of soil spectroscopy models for the Western Highveld region, South Africa: Why do we need local data? DOI Creative Commons
Andrea de Kock, P. Dimakatso Ramphisa, George van Zijl

и другие.

European Journal of Soil Science, Год журнала: 2024, Номер 75(6)

Опубликована: Сен. 1, 2024

Abstract The increasing global demand for sustainable agriculture requires accurate and efficient soil analysis methods. Conventional laboratory techniques are often time‐consuming, costly environmentally damaging. To address this challenge, we developed validated locally calibrated mid‐infrared (MIR) spectroscopy models predicting key properties pH, phosphorus (P) exchangeable cations in samples from South Africa's Western Highveld region, using a dataset of 979 machine learning algorithms Cubist, partial least squares regression (PLSR) random forest (RF). A subset spectra was also submitted to the newly Open Soil Spectral Library's (OSSL) prediction determine whether could be used local property prediction. Accurate predictions calcium (Ca) magnesium (Mg), with coefficient determination ( R 2 ) values exceeding 0.76 were obtained calibration algorithms. P, potassium (K) sodium (Na) did not meet requirements reliability. spectroscopic soils outperformed corresponding considered. OSSL results inaccurate, RPIQ <1, consistently underpredicted all properties. Furthermore, collection does include pH (KCl) model, routinely measurement method Africa. These findings highlight importance underscore need regional representation spectral libraries. This research serves as first MIR region Africa provides foundation future inference model development. It potential starting point comprehensive African library that can contributed

Язык: Английский

Процитировано

2
Building a Near-infrared (NIR) Soil Spectral Dataset and Predictive Machine Learning Models using a Handheld NIR Spectrophotometer DOI Creative Commons

Colleen Partida,

José Lucas Safanelli,

Sadia Mannan Mitu

и другие.

Data in Brief, Год журнала: 2024, Номер 58, С. 111229 - 111229

Опубликована: Дек. 16, 2024

Язык: Английский

Процитировано

2
Comparing the handheld Stenon FarmLab soil sensor with a Vis-NIR multi-sensor soil sensing platform DOI Creative Commons
Alexander Steiger, Abdul M. Mouazen, Muhammad Qaswar

и другие.

Smart Agricultural Technology, Год журнала: 2024, Номер unknown, С. 100717 - 100717

Опубликована: Дек. 1, 2024

Язык: Английский

Процитировано

1
Mapping of Clay Montmorillonite Abundance in Agricultural Fields Using Unmixing Methods at Centimeter Scale Hyperspectral Images DOI Creative Commons
Etienne Ducasse, Karine Adeline, Audrey Hohmann

и другие.

Remote Sensing, Год журнала: 2024, Номер 16(17), С. 3211 - 3211

Опубликована: Авг. 30, 2024

The composition of clay minerals in soils, and more particularly the presence montmorillonite (as part smectite family), is a key factor soil swell–shrinking as well off–road vehicle mobility. Detecting these topsoil quantifying abundance are challenge since they usually intimately mixed with other minerals, organic carbon moisture content. Imaging spectroscopy coupled unmixing methods can address issues, but quality estimation degrades coarser spatial resolution due to pixel heterogeneity. With advent UAV-borne proximal hyperspectral acquisitions, it now possible acquire images at centimeter scale. Thus, objective this paper evaluate accuracy limitations retrieve from very-high-resolution (1.5 cm) acquired camera installed on top bucket truck over three different agricultural fields, Loiret department, France. Two automatic endmember detection based assumption that materials linearly mixed, namely Simplex Identification via Split Augmented Lagrangian (SISAL) Minimum Volume Constrained Non-negative Matrix Factorization (MVC-NMF), were tested prior unmixing. Then, two linear methods, fully constrained least square method (FCLS) multiple spectral mixture analysis (MESMA), nonlinear ones, generalized bilinear (GBM) multi-linear model (MLM), performed images. In addition, several preprocessings applied order improve performances. Results showed our selected not suitable context. However, endmembers taken available libraries successfully. method, MLM, without preprocessing or application first Savitzky–Golay derivative, gave best accuracies for using USGS library (RMSE between 2.2–13.3% 1.4–19.7%). Furthermore, significant impact estimations scale was majority (i) high variability composition, (ii) roughness inducing large variations illumination conditions surface scatterings (iii) volume coming intimate mixture. Finally, results offer new opportunity mapping expansive soils imaging very resolution.

Язык: Английский

Процитировано

0