Session 11.2: New technologies and innovations to enhance soil health

Pollution with trace metals is a serious concern for global soil quality. Trace metals are characterized not only by their concentration but also by their distribution among different chemical species, known as speciation. Conventionally, the concentration and speciation of trace metals are determined using expensive and time-consuming chemical analysis and sequential extraction procedures. As a complementary approach, this study investigated the capability of visible–near-infrared–shortwave infrared (VNIR–SWIR) reflectance spectroscopy for quantifying nickel (Ni) and cobalt (Co) levels in soil samples collected from the Sarcheshmeh copper mine waste dump in southern Iran. The potential of the technique was also studied for the speciation of those elements. Three machine learning (ML) algorithms (SVR, RF, and EGB) were employed to model Ni and Co concentrations of the samples using soil spectral responses. The highest prediction performances (Ni: RMSEp values of 11.02, 10.32, 9.88 mg·kg−1, and Co: RMSEp values of 9.22, 8.03, 7.82 mg·kg−1) were yielded by the models developed using SVR, RF, and EGB approaches, respectively, on the first derivative (FD) pre-processed spectra. Comparison between the spatial distribution maps of the elements revealed relatively similar trends between the observed and predicted values. Correlation analysis and ML-based techniques revealed that the most important wavelengths for predicting Ni and Co were those associated with clay minerals and iron oxides/hydroxides, two main soil constituents that control their speciation. This study provided inspiration for implementing VNIR–SWIR spectroscopy as a rapid and cost-effective tool for speciation of trace metals in heterogeneous soil environments.

Soil is a complex environment where numerous biotic and abiotic processes operate across spatial and temporal scales, continuously modifying its structure. These dynamics influence key processes such as water flow, nutrient and pollutant leaching, microbial activity, etc. Despite this, most current soil-plant-atmosphere models still assume a static soil structure and therefore overlook its potential influence.
In this contribution, we present an extended version of the recently published USSF model featuring a more physically based and dynamic representation of soil hydraulic properties (SHP). The model was used to generate time‑dependent SHP driven by various biotic and abiotic processes, which were then provided into the widely used HYDRUS 1D model. This setup enables HYDRUS to simulate dual‑permeability (matric and macropore) water flow with dynamic domains and hydraulic properties for each domain.
We applied the model to a long-term agricultural scenario consisting of 20 years of conservation agriculture following 20 years of conventional ploughing. Comparisons between simulations with dynamic and static SHP show substantial differences in the water fluxes, underscoring the importance of accounting for evolving SHP in hydrological models. 

Tailor-made fertilizers (TMFs) derived from biobased fertilizers (BBFs) represent a promising strategy to improve nutrient use efficiency while reducing reliance on conventional mineral fertilizers. In this study, two TMFs (TMF1 and TMF2) were formulated based on site-specific soil conditions, crop nutrient requirements, and alignment with the EU fertilizing products regulation. TMF1 promotes biologically driven carbon–nitrogen coupling and sustained nitrate formation, whereas TMF2 enhances carbon availability and nutrient solubility, reflecting distinct formulation-driven biogeochemical regulation strategies.
A controlled 90-day soil incubation experiment was conducted to assess their functional behavior prior to field deployment. Carbon mineralization was monitored through CO2 emissions and modeled to evaluate mineralizable carbon pools and turnover dynamics. In parallel, nitrogen transformations (NH4+ and NO3-) and phosphorus fractionation (Olsen, water-soluble, and CaCl2-extractable) were quantified. A CaCl2 extraction protocol was applied as a proxy for plant-available nutrient pools.
Results revealed distinct formulation-dependent responses. TMF2 increased initial respiration intensity and expanded the mineralizable carbon pool, whereas TMF1 prolonged carbon turnover, resulting in stronger carbon–nitrogen coupling and higher nitrate formation. Both TMFs enhanced labile phosphorus, with TMF2 showing higher immediately soluble P and greater ionic strength, indicating enhanced nutrient solubility.
These findings demonstrate that TMFs can act as engineered regulators of soil biogeochemical processes, controlling the temporal dynamics of carbon turnover and nutrient availability. The laboratory-based approach provides a robust pre-field validation framework, reducing uncertainty before agronomic testing. Ongoing field trials under wheat cultivation will further assess nutrient uptake, crop performance, and environmental impacts.

Accurate mapping of soil organic carbon (SOC) is essential for sustainable land management, climate mitigation, and agricultural policy, yet spatially explicit SOC data remain scarce across Europe. Earth Observation (EO) technologies beyond traditional multiband and RGB sensors, such as hyperspectral imaging (HSI), offer a scalable pathway to fill this gap, but mapping SOC requires integrating heterogeneous data sources and overcoming institutional data-sharing barriers. This study contributes to improved topsoil SOC estimation by introducing a federated learning framework that enables knowledge sharing at the regional scale, while providing preliminary evidence that data fusion can enhance spectral and spatial information at the field scale. 
At regional scale, 10 m resolution SOC maps were generated by combining Sentinel-2 multispectral imagery with LUCAS 2018 European cropland dataset and ancillary soil properties (bulk density, texture). A federated learning approach, using a CNN architecture and regional soil datasets from both study areas, enabled decentralized fine-tuning respecting data privacy, sharing only model weights rather than raw soil data and improving model generalization (MAE 2.3–3.6 g/kg). 
However, regional-scale approaches lack the detail required for field-level variability. To address this, we combine UAV-based HSI (< 1 m) with Sentinel-2 and Hyperfield-1 data through a data fusion framework that exploits complementary spatial and spectral information across sensors. Preliminary results indicate that data fusion improved SOC estimation accuracy (R² from 0.36 to 0.40; MSE from 0.22 to 0.21).  
Together, these products demonstrate a scalable, multi-source EO pipeline for SOC monitoring across spatial scales. 

Mediterranean maize production systems are characterised by a high dependence on mineral nitrogen fertilisation and a strong vulnerability to nitrogen losses, particularly through nitrate leaching between maize growing seasons. In this context, cover crops represent a promising strategy to capture residual N, reduce environmental losses, and support soil-mediated nutrient cycling.
This study is conducted within the SoilRes European project and aims to evaluate the impact of winter cover crops on nitrogen dynamics in a commercial maize-based system at Quinta da Cholda (Santarém, Portugal). The experiment follows a randomized complete block design (three treatments and four replicates: (i) bare soil control, (ii) a commercial multispecies mixture, and (iii) a farm-designed multispecies mixture. Treatments were established on 11 November 2025.
Nitrogen dynamics are assessed through complementary field measurements during the intercrop and subsequent crop period. Greenhouse gas emissions, including nitrous oxide, are measured using static chambers at regular intervals to capture temporal variability. Soil solution is sampled at 100 cm depth using ceramic suction cups to determine NO₃⁻-N and NH₄⁺-N concentrations, as well as pH and electrical conductivity.

Preliminary results show marked temporal variability in nitrous oxide fluxes (−0.22 to 0.37 g N–N₂O m⁻² day⁻¹) and substantial nitrate mobility, with NO₃⁻-N concentrations ranging from 3.78 to 26.11 mg L⁻¹. In contrast, NH₄⁺-N concentrations remained consistently low.
Overall, this study provides a robust field-based framework to assess nitrogen dynamics and highlights the potential of winter cover crops to improve nitrogen retention and support soil health in Mediterranean maize systems.

Soil degradation threatens nearly 33% of global soils, with soil organic carbon (SOC) declining by 0.3–0.5% annually in intensively cultivated regions due to excessive fertilizer application, monocropping, and poor nutrient management. In developing countries, inefficient input use results in nutrient losses of 40–60%, reduced microbial diversity, and declining productivity. This study evaluates the application of AI-driven soil intelligence systems to restore soil health across 1,000 acre farming clusters through a precision-based, soil intelligence model.
The proposed framework integrates IoT-enabled soil sensors that capture real-time data on moisture, pH, temperature, and macronutrient levels at 15-minute intervals, along with satellite-based soil variability mapping at 10 m spatial resolution. These data streams feed into machine learning–based nutrient optimization models trained on multi-season agronomic datasets. The AI-enabled models reduce synthetic fertilizer application by 20–35% while improving nutrient use efficiency by up to 25%.
The system also incorporates biochar application strategies that further improve soil water retention by 15–20% and reduce nutrient leaching losses by up to 30%. Field implementation demonstrates 8–12% yield improvement, a 25% reduction in input costs, and measurable improvements in soil aggregation and microbial biomass carbon.
Through a scalable hub-and-spoke digital advisory model, the initiative supports carbon quantification and soil health indexing, enabling participation in voluntary carbon markets. By combining artificial intelligence, remote sensing, and regenerative agriculture practices, this approach offers a measurable, climate-resilient pathway to restore soil functionality while sustaining farm profitability and long-term agricultural sustainability.