Geoderma最新文献

{"title":"Does digital soil mapping prediction performance of soil texture improve when adding uncertain field texture estimates? A study based on clay content","authors":"Anne C. Richer-de-Forges ,&nbsp;Songchao Chen ,&nbsp;Gerard B.M. Heuvelink ,&nbsp;Stephan van der Westhuizen ,&nbsp;Thomas G. Orton ,&nbsp;Hocine Bourennane ,&nbsp;Dominique Arrouays","doi":"10.1016/j.geoderma.2025.117277","DOIUrl":"10.1016/j.geoderma.2025.117277","url":null,"abstract":"<div><div>Hand-feel soil texture observations (HFST) are less accurate, yet much more numerous, than laboratory measurements of soil texture (LAST). Therefore, it is tempting to incorporate both LAST and HFST information as calibration data in digital soil mapping (DSM) of particle-size distribution. We used about 1000 LASTs and 15,000 HFSTs over an area of about 6,800 km². We incorporated the uncertainties of HFST and LAST calibration data in DSM and compared it with a case where measurement errors were ignored. We added progressively HFST calibration data to LAST data and ran predictions and k-fold validations keeping the same validation set for all experiments. We added HFSTs according to different strategies: either based on the most uncertain predicted areas from the LAST-only model, or those from the preceding LAST + nHFST model, or randomly. We discuss the pros and cons of these different strategies. Adding HFST data brought useful information for model calibration, but only if the uncertainty was accounted for. Various strategies for adding HFSTs led to different unbalanced samplings, maps and prediction intervals. We explain how these various unbalanced samplings sharpened or enlarged the predictive distribution of various clay content ranges. Adding a too large number of HFSTs led to an over-optimistic estimation of the 90 % prediction interval and to large homogeneous patterns, smoothing the spatial variation of clay content. Adding HFSTs using weights to acknowledge their uncertainty substantially improved DSM predictions, but the number of HFSTs and the strategy to add them must be carefully adapted.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117277"},"PeriodicalIF":5.6,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143739883","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Corrigendum to “Empirical estimation of saturated soil-paste electrical conductivity in the EU using pedotransfer functions and Quantile Regression Forests: A mapping approach based on LUCAS topsoil data” [Geoderma 454 (2025) 117199]","authors":"Calogero Schillaci ,&nbsp;Simone Scarpa ,&nbsp;Felipe Yunta ,&nbsp;Aldo Lipani ,&nbsp;Fernando Visconti ,&nbsp;Gábor Szatmári ,&nbsp;Kitti Balog ,&nbsp;Triven Koganti ,&nbsp;Mogens Greve ,&nbsp;Giulia Bondi ,&nbsp;Georgios Kargas ,&nbsp;Paraskevi Londra ,&nbsp;Fuat Kaya ,&nbsp;Giuseppe Lo Papa ,&nbsp;Panos Panagos ,&nbsp;Luca Montanarella ,&nbsp;Arwyn Jones","doi":"10.1016/j.geoderma.2025.117253","DOIUrl":"10.1016/j.geoderma.2025.117253","url":null,"abstract":"","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117253"},"PeriodicalIF":5.6,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143791131","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Kinetics and dynamics of dissolved organic nitrogen, and nitrogen cycling genes in the soil profile of an irrigated vineyard with surface-applied organic amendments","authors":"Mehdi Sharifi ,&nbsp;Lori Phillips ,&nbsp;Seanna Zintel ,&nbsp;Brent Seuradge","doi":"10.1016/j.geoderma.2025.117246","DOIUrl":"10.1016/j.geoderma.2025.117246","url":null,"abstract":"<div><div>Understanding the dynamics of dissolved organic nitrogen (DON) in the vineyard’s soil is crucial for nitrogen (N) management in this agroecosystem. This study investigated the kinetics and dynamics of DON, including underlying microbial mechanisms, in the soil profile of an irrigated Merlot vineyard in the Okanagan Valley, BC, Canada. Biennial treatments from 2011 to 2018 included wood-based mulch with or without agricultural compost (26.7 and 13.7 Mg ha⁻¹ fresh weight, respectively) and a non-amended control. In the Fall of 2018, composite soil samples were collected at 15 cm intervals down to 60 cm depth. Microbial N cycling functional potential was determined, and initial DON concentrations, and potentially releasable DON (N₀-_DON), and the DON release rate constant (k-_DON) were assessed via 100-day aerobic incubation. Pre-incubation DON levels (0.89–44.1 mg N kg⁻¹) were higher in mulch + compost ≥ mulch &gt; control and decreased with depth. DON release followed first-order kinetics, with N₀-_DON ranging from 12.1 to 188 mg N kg⁻¹ and being unaffected by amendments, but decreased with depth. Cumulative DON averaged 33.8 mg N kg⁻¹ and was also unaffected by amendments, but it decreased with depth. The k-_DON (0.016–0.224 wk⁻¹) was influenced by amendments and depth, with higher values in mulch + compost compared to mulch-only and control, and it declined with depth. Organic amendments doubled bacterial proteolytic capacity, positively correlating with N parameters. Bacillota-related proteolytic (<em>npr</em>) and nitrate reduction (<em>nrfA</em>) genes were associated with cumulative DON and nitrate release during the incubation, suggesting that mineral N released by proteolytic activity was re-immobilized into microbial biomass, maintaining plant-available N throughout the soil profile. First-order kinetics further support the key role of soil organisms in DON dynamics. In low-N systems like vineyards, DON supply rivals mineral N and increases with organic amendments. While DON supplementation supports mineral N supply, its potential capacity diminishes over the long term, necessitating periodic applications to maintain effectiveness. The relatively fast release of N_min and DON from mulch + compost makes it ideal for early spring applications to synchronize N release with grapevine N uptake. These findings are particularly relevant for organic vineyards with limited access to N sources.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117246"},"PeriodicalIF":5.6,"publicationDate":"2025-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143734967","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The single-layer equivalent soil mass method for the evaluation of soil organic carbon stocks: Sources of errors, simplification, and associated detectable change","authors":"Pascal Boivin,&nbsp;Téo Lemaître,&nbsp;Julien Clark,&nbsp;Marie Guittonneau,&nbsp;Cédric Deluz","doi":"10.1016/j.geoderma.2025.117279","DOIUrl":"10.1016/j.geoderma.2025.117279","url":null,"abstract":"<div><div>It has been well established that for consistent soil organic carbon stock (<em>SOCS</em>) monitoring over time, the <em>SOCS</em> must be determined in an equivalent soil mass (ESM) layer rather than a depth layer. This work focused on the single-layer ESM method, its cost, accuracy and simplification. With the objective of simplifying ESM method and reducing costs, we sampled 393 fields in western Switzerland and experimentally questioned the sampling methods, the sources of variance and the minimum detectable change (<em>MDC</em>) of <em>SOCS</em>, and possible simplification. The layer mass was accurately determined using gouge-augers, thus overcoming a major drawback in performing ESM <em>SOCS</em> evaluation. The relationships between number of aliquots and estimated SOCS variance was determined. Sampling at fixed depth and layer bulk density resulted in unacceptable errors and MDCs exceeding several decades when converted to years before a change was detectable. We introduced a new procedure to fully consider the coarse fraction volume in the layer, which remains a major source of error on the <em>SOCS</em> when the coarse fraction volume exceeds 10 % of the layer volume, while not impacting the detectability of the change. The single-layer ESM method provided an <em>MDC</em> corresponding to less than 10 years before <em>SOCS</em> change detectability under the average regional conditions, and mass correction results in negligible increase in <em>MDC</em> compared to that of the 0–30 cm layer. Simplifying this method by using the average soil organic carbon content of the layer mass correction only slightly increased the <em>MDC</em>, thus providing an opportunity to decrease the cost of <em>SOCS</em> monitoring significantly.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117279"},"PeriodicalIF":5.6,"publicationDate":"2025-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143734968","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Soil organic carbon stocks and stabilization mechanisms in tidal marshes along estuarine gradients","authors":"Friederike Neiske ,&nbsp;Maria Seedtke ,&nbsp;Annette Eschenbach ,&nbsp;Monica Wilson ,&nbsp;Kai Jensen ,&nbsp;Joscha N. Becker","doi":"10.1016/j.geoderma.2025.117274","DOIUrl":"10.1016/j.geoderma.2025.117274","url":null,"abstract":"<div><div>Tidal marshes in estuaries store large amounts of soil organic carbon (SOC) and are dominated by the interaction of tidal inundation and salinity with biotic ecosystem components, leading to strong spatio-temporal heterogeneity. Little is known how these estuarine conditions affect SOC stabilization. Our aim was to assess (i) SOC stocks, (ii) SOC stabilization mechanisms (aggregation and mineral-association), and (iii) their environmental drivers along estuarine gradients. We analyzed SOC stocks and SOC density fractions in topsoil (0–10 cm) and subsoil (10–30 cm) of three marsh zones representing three flooding regimes (daily, monthly, yearly) in three marsh types along the salinity gradient (salt, brackish, freshwater) of the Elbe Estuary, Germany. The SOC stocks in 0–30 cm ranged between 9.3–74.6 t ha⁻¹, and decreased with increasing salinity and flooding. This was linked to decreasing plant biomass and soil fine texture. Organic matter (OM) associated with minerals (C_MAOM) constituted the largest SOC fraction (59 % of SOC), followed by aggregate-occluded OM (C_oPOM) (24 %) and free particulate OM (C_fPOM) (16 %). The C_MAOM amount in topsoils decreased with increasing salinity, reflecting decreasing soil fine texture. Amounts of C_oPOM were higher in topsoils and high marshes, indicating negative effects of flooding on aggregation. The contribution of C_fPOM to total SOC increased with increasing flooding, likely related to its preservation under reducing soil conditions. Our results emphasize that increasing marine influence (rising salinity and flooding frequency) leads to a decrease in SOC content and stabilization. Therefore, sea–level rise has the potential to negatively impact SOC storage in estuarine marshes.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117274"},"PeriodicalIF":5.6,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143725366","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"National-scale mapping topsoil organic carbon of cropland in China using multitemporal Sentinel-2 images","authors":"Jie Xue ,&nbsp;Xianglin Zhang ,&nbsp;Songchao Chen ,&nbsp;Zhongxing Chen ,&nbsp;Rui Lu ,&nbsp;Feng Liu ,&nbsp;Bas van Wesemael ,&nbsp;Zhou Shi","doi":"10.1016/j.geoderma.2025.117272","DOIUrl":"10.1016/j.geoderma.2025.117272","url":null,"abstract":"<div><div>Precise monitoring of soil organic carbon (SOC) is urgently needed in agricultural regions to tackle global challenges like food security, water regulation, land degradation, and climate change. Remote sensing technology has emerged as a powerful method for detecting variations in SOC at localized scales. However, its application on a broader, national scale faces limitations, especially in countries like China, where soil landscapes exhibit significant diversity. This study aimed to couple bare soil reflectance and conventional environmental covariates to map Chinese cropland SOC content at a 10-m spatial resolution. First, a new time-series bare soil extraction method, the Two-Dimensional Bare Soil Separation Algorithm, was applied, utilizing Sentinel-2 images from 2018 to 2022 to generate a continuous spectral reflectance composite. Then, nine indices with the strongest correlation to SOC were selected. Additionally, a list of environmental covariates was prepared based on SCORPAN model. Finally, bootstrapping random forest models were fitted using the covariates selected through forward recursive feature selection (FRFS), and the spatial prediction SOC map was created. The results indicated that the framework was suitable for mapping SOC in croplands of China, with the best model using remote sensing indices and environmental covariates selected through FRFS achieving an R² of 0.62, an RMSE of 4.84 g kg⁻¹, and an uncertainty depicted by a 90 % prediction interval range of 17.88 g kg⁻¹. The final map showed that the Northeast China had the highest SOC content in cropland. Climatic conditions, position, and remote sensing indices are key covariates in national-scale SOC estimation. This study can be easily implemented across broad areas for the prediction of SOC with computational efficiency. The 10-m spatial resolution SOC map of China contributes to land management and the development of policies for precision agriculture.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117272"},"PeriodicalIF":5.6,"publicationDate":"2025-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143725365","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Soil gas and solute diffusivity as predictors of N2O emissions after spring fertilization: An incubation study with intact soil","authors":"Winnie Ntinyari,&nbsp;Loraine ten Damme,&nbsp;Jianbo Cheng,&nbsp;Lars J. Munkholm,&nbsp;Søren O. Petersen","doi":"10.1016/j.geoderma.2025.117267","DOIUrl":"10.1016/j.geoderma.2025.117267","url":null,"abstract":"<div><div>Agricultural soils are the main source of anthropogenic N₂O emissions to the atmosphere. To effectively mitigate these emissions there is a need to understand the interactions between potential drivers under field conditions. We conducted an incubation experiment with intact soil cores adjusted to one of four matric potentials (−15, −30, −50 or −100 hPa). We measured the relative gas diffusivity <em>(Dp/Do</em>) and effective air-filled porosity (<em>ε_eff</em>) before incubation with four surface-applied fertiliser treatments: NS26-13 (NS) at 50 kg N ha⁻¹ (NS50; starter); NS at 50 + 100 kg N ha⁻¹ (NS150); NS at 50 + 100 kg N ha⁻¹, the main dose of 100 kg N ha⁻¹ containing the nitrification inhibitor 3,4-dimethylpyrazol phosphate (NS150 + DMPP); and an unfertilized Control. We monitored N₂O fluxes for 35 days and found a positive relationship with matric potential and a weak inverse relationship with <em>Dp/Do</em> and <em>ε_eff..</em> After 35 days, the cores were sectioned for determination of mineral N pools and potential ammonia oxidation (PAO). A strong NH₄⁺ gradient, declining with depth, was observed which indicated that nitrification took place at some distance from the surface-applied fertiliser, and this was consistent with PAO and pH distributions. The flux of NH₄⁺ in the soil solution between 0–2 and 2–4 cm depth after 35 days showed a consistent relationship with N₂O emissions at −30 to −100 hPa indicating nitrification was the main source of N₂O in this range. The results of this study suggest that both gas and solute diffusivity have to be considered for predicting N₂O emissions from surface-applied fertilisers.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117267"},"PeriodicalIF":5.6,"publicationDate":"2025-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143716149","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Biogeographic patterns and adaptive strategies of microbial carbon metabolic profiles in paddy soils in the Chinese Mollisol region","authors":"Xiaojing Hu ,&nbsp;Haidong Gu ,&nbsp;Mikhail Semenov ,&nbsp;Yongbin Wang ,&nbsp;Jinyuan Zhang ,&nbsp;Zhenhua Yu ,&nbsp;Yansheng Li ,&nbsp;Junjie Liu ,&nbsp;Jian Jin ,&nbsp;Xiaobing Liu ,&nbsp;Guanghua Wang","doi":"10.1016/j.geoderma.2025.117265","DOIUrl":"10.1016/j.geoderma.2025.117265","url":null,"abstract":"<div><div>Microbial adaptive evolution and regulation strategies response to carbon (C) heterogeneity in paddy fields at large spatial scales remain poorly understood. Here, we employed gene- and genomic-centric metagenomic approaches to explore soil microbial biogeographic patterns and functional traits regulating C metabolisms across 120 soils derived from 30 paddy fields in Chinese Mollisols. Our findings revealed that significant distance-decay relationships (DDRs) were observed at both microbial C functional and genomic taxonomic levels. Microbial C cycling profiles were clustered into two groups. HCS (including sites R1-R10) represented soils with high total carbon (TC) at relatively high latitudes, whereas LCS (including sites R11-R30) had low soil C content distributed at low latitudes. Compared with HCS, LCS presented higher abundances of C cycling pathways involving aerobic respiration, C fixation, and methanogenesis pathways, as well as higher levels of carbohydrate esterase (CE) and glycosyl transferase (GT) classes. 211 metagenome-assembled genomes (MAGs) with diverse C metabolic functions were constructed. Among these, high-quality MAG292, assigned to the Nanopelagicales order, had a significantly positive correlation with TC and was more abundant in HCS. Contrarily, MAG153, assigned to the Chitinophagales order, exhibited an opposite trend. Additionally, 133 novel vMAGs were retrieved, and the abundances of phage11, phage16, phage26, and phage120 were higher in LCS than in HCS, containing <em>chiA</em> and GH19 that involved in chitin degradation. HCS had a relatively high abundance of phage89, containing <em>slt</em> and GH23 genes that regulate peptidoglycan lysis. These results indicated that soil viruses potentially lyse bacteria by encoding peptidoglycan lyase, releasing nutrients, and increasing the amount of dead microbial debris that facilitates soil C accumulation at relatively high latitudes. In contrast, at low latitudes, the phages together with microbes may indirectly decrease the soil TC by potentially expressing auxiliary metabolic genes (AMGs) involved in chitin degradation. Our findings indicate the divergent microbial adaptive evolution and soil C regulation strategies response to soil C heterogeneity in paddy soils of Chinese Mollisols.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117265"},"PeriodicalIF":5.6,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143705817","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Predicting the thickness of alpine meadow soil on headwater hillslopes of the Qinghai-Tibet Plateau","authors":"Xiaole Han ,&nbsp;Jintao Liu ,&nbsp;Pengfei Wu ,&nbsp;Zhenghong Yu ,&nbsp;Xiao Qiao ,&nbsp;Hai Yang","doi":"10.1016/j.geoderma.2025.117271","DOIUrl":"10.1016/j.geoderma.2025.117271","url":null,"abstract":"<div><div>Alpine meadow soils in tectonically active High Mountain Asia (HMA) are highly vulnerable, playing a critical role in the fragile ecosystems of Earth’s “Third Pole”. Understanding the formation and distribution of these soils is essential, yet the mechanisms governing their thickness remain unclear. To address this, we applied a multi-resolution (0.25–30.00 m) stochastic approach to predict soil thickness in a nested headwater catchment in the Lhasa River Basin, namely D5K (415 sites) and its headwater sub-catchment D5KH1 (330 sites). Using a linear mixed-effects model, we quantified uncertainty through within-site and between-site variance analysis, and compared the performance of deep learning (DL) with three traditional machine learning methods—random forest (RF), support vector machine (SVM), and artificial neural networks (ANN)—under three variable selection strategies: (1) all variables, (2) stepwise selection, and (3) a novel Boruta-based recursive method. Field investigations revealed that organic matter accumulation and freeze–thaw cycles are dominant pedogenic factors, producing a root-dense turf layer beneath a dark humus horizon. Freeze-thaw dynamics also contribute to geomorphological features such as landslides and stone stripes. Soil thickness varied significantly across topographic positions—valleys, sideslopes, and ridges (p &lt; 0.001)—with within-site variance (SD = 11.06 cm) slightly exceeding between-site variance (SD = 9.58 cm). Fine-resolution digital elevation models (DEMs) effectively captured these variations, largely because they incorporate critical topographic factors such as the Topographic Position Index (TPI), which Boruta analysis identified as more influential on soil thickness than flow-related indices like the Topographic Wetness Index. Among the tested models, RF combined with the Boruta method and a 0.25 m resolution DEM provided the most accurate predictions, outperforming DL. These findings emphasize that the complexity of DL may not translate into superior performance for all applications, particularly in small dataset terrain-driven ecological studies.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117271"},"PeriodicalIF":5.6,"publicationDate":"2025-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143704533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Federated learning applications in soil spectroscopy","authors":"Giannis Gallios ,&nbsp;Nikolaos Tsakiridis ,&nbsp;Nikolaos Tziolas","doi":"10.1016/j.geoderma.2025.117259","DOIUrl":"10.1016/j.geoderma.2025.117259","url":null,"abstract":"<div><div>Soil spectroscopy has emerged as a key technique for rapid, non-destructive soil property prediction, yet the centralized nature of the training of machine learning models raises concerns around data privacy, accessibility, and transferability. This study proposes the application of Federated Learning (FL) as a decentralized approach to soil spectroscopy, enabling collaboration between multiple data contributors (or clients) without requiring the exchange of raw data. Convolutional neural networks were used to estimate key soil attributes such as soil organic carbon, texture, pH, cation exchange, and total nitrogen using VNIR or MIR spectral data from the Kellogg Soil Survey Laboratory. Three data partitioning approaches were explored involving geopolitical, bioclimatic, independent, and identically distributed (IID) splitting methods to simulate various real-world scenarios of particular interest to the soil community. Each scenario was investigated under two different averaging aggregation strategies: Federated Averaging (FedAvg) and Weighted Averaging (WgtAvg), which are used to develop a consensus model by aggregating the weights of the different contributors. The results show that the FL framework can match, and in some cases exceed, the performance of centralized models, particularly when using IID data. The WgtAvg strategy was particularly effective in boosting prediction accuracy even by 50 % for soil properties where contributors had unequal data sizes. This study highlights the potential of FL as a privacy-preserving framework across networks of soil labs, facilitating enhanced soil health monitoring with decentralized global models.</div></div>","PeriodicalId":12511,"journal":{"name":"Geoderma","volume":"456 ","pages":"Article 117259"},"PeriodicalIF":5.6,"publicationDate":"2025-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143684577","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}