Ho Jun Jang

Machine learning (ML) is increasingly used in biochar research to predict yield, material properties, and adsorption performance. Yet, this review reveals that much of the reported success in ML‐driven biochar studies may be overstated due to data leakage, the inadvertent transfer of information between training and testing datasets. We systematically examined 64 of the most‐cited studies published between 2015 and 2025 and found that all studies exhibited potential leakage, commonly arising from hierarchical or grouped data structures, multi‐source compilations, or temporal and concentration dependencies. Nearly all studies (98%) employed random point‐level data splits for training and testing, disregarding dependencies among samples, while only one adopted a group‐based split approach. Re‐analysis of five representative datasets showed that random point‐level splitting can inflate predictive accuracy by up to 86%, often yielding test‐set R ² (coefficient of determination) values exceeding 0.90 even when model generalisation fails. Leakage risks were particularly severe in adsorption studies and in those using very small datasets (< 35 samples), where overfitting further distorted model performance. To ensure methodological rigour and reliable generalisation, biochar ML research must adopt grouped or study‐level data‐splitting strategies, maintain strict separation of preprocessing steps, and incorporate hierarchical or domain‐aware validation frameworks. Future progress requires automated leakage‐detection tools, benchmark datasets with explicit metadata on dependencies, and hybrid models that integrate physical constraints. Addressing data leakage systematically is essential to restore credibility, enhance reproducibility, and realise the full potential of ML as a reliable tool for environmental research.

The long-term sustainability of Australian agriculture is fundamentally constrained by the capacity, condition, availability, and governance of soil resources. Australian soils are among the oldest and most weathered globally, highly heterogeneous, and often slow or effectively irreversible to recover once degraded. Traditional approaches centred on soil health, while valuable at paddock scale, are insufficient to address national-scale challenges related to spatial variability, data continuity, economic valuation, and policy integration. This paper examines soil security as a policy-relevant framework for supporting more sustainable Australian agriculture. Building on the dimensions of soil security (capacity, condition, capital, connectivity, and codification), we synthesise recent Australian case studies to show how soil security extends beyond soil health to integrate biophysical properties, digital soil infrastructure, socio-economic value, and governance mechanisms. Drawing on recent Australian case studies, this review identifies advances in digital soil mapping, national soil assessments, economic valuation of soil capital, stakeholder connectivity, and emerging policy frameworks, while also identifying persistent gaps in regulation, data standardisation, and institutional coordination. The paper argues that soil security can help operationalise 3-N agriculture—Net-Zero, Nature-Positive, and Nutrient-Balanced systems—by translating sustainability goals into spatially explicit, place-based decisions grounded in soil realities. By explicitly accounting for soil capacity limits, condition trajectories, capital value, information flows, and codified rules, soil security can support more realistic climate mitigation strategies, targeted nature-positive interventions, and durable nutrient security outcomes. We conclude that embedding soil security more explicitly within Australian agricultural research, policy, and governance would strengthen efforts to deliver productive, resilient, and socially legitimate food and fibre systems. Without soil security, sustainability frameworks may remain difficult to operationalise consistently; with soil security, they can be translated more effectively into measurable, place-based, and durable decisions.

Infrared (IR) spectroscopy has emerged as a rapid, cost-effective, and reliable alternative to traditional methods, enabling real-time, indirect monitoring of nutrients. Most reviews have discussed visible-near-infrared (Vis-NIR) and mid-infrared (MIR) spectroscopy individually for soil analysis. This review highlights the application of IR spectroscopy, particularly Vis-NIR, MIR spectroscopy, and their data fusion, coupled with chemometrics and spectral preprocessing for estimating soil attributes. Additionally, the crucial functions of assessing model accuracy and validating model estimates of soil properties are discussed. Partial least squares regression (PLSR) was used in more than 100 studies in 2022. Based on the literature published from 2020 to 2025, the data fusion method predicts soil properties more accurately. This review also sheds light on recent advances in spectroscopic methods, including improvements in speed (e.g., MIR spectroscopy is up to 12 times faster than traditional methods), instrument miniaturization, and integration with portable devices, which can make field analysis more affordable. However, the sensitivity of IR spectroscopy to soil moisture, sample heterogeneity, vegetation cover, and calibration transfer issues remains a significant challenge in certain studies. Therefore, a discussion on the challenges in implementing this technique is included in this review, and future perspectives, such as integration of various sensors and portable devices for real-time soil assessment, are successively discussed.

Soil analysis is fundamental to sustainable agriculture; however, traditional laboratory methods are time-consuming, expensive, and environmentally hazardous. Spectroscopy techniques, particularly Near-Infrared (NIR), have gained considerable attention because they require minimal sample preparation, no chemicals, and predict multiple soil properties with a single scan. However, selecting appropriate equipment remains critical, as previous studies have reported inconsistent performance between conventional Near-Infrared (NIR) spectroscopy and advanced Fourier-Transform Near-Infrared (FT-NIR) spectroscopy. Therefore, this study aimed to compare the predictive performance of conventional NIR and advanced FT-NIR spectroscopy for sixteen soil properties. Soil samples (n = 567) were collected from different land-use types across South Korea at a depth of 0–20 cm and analyzed using laboratory methods and spectroscopy techniques. Five models, including partial least squares regression (PLSR), Cubist, support vector machine (SVM), random forest (RF), and memory-based learning (MBL), were evaluated using 15-fold cross-validation to assess prediction accuracy. Overall, conventional NIR spectroscopy yielded consistently higher accuracy for all soil properties than FT-NIR. Strong predictive accuracy was achieved for EC (R² = 0.84), OM (R² = 0.84), avl. P (R² = 0.77), TN (R² = 0.84), and CEC (R² = 0.76). In contrast, FT-NIR provided good prediction accuracy only for Ex. K (R² = 0.72) and TN (R² = 0.84). The average performance of NIR (R² = 0.67) outperformed FT-NIR spectroscopy (R² = 0.63) across all soil properties. These findings demonstrate that, despite their lower spectral resolution, NIR spectra provide robust predictive capability across a wide range of soil properties, which can significantly reduce the investment cost of advanced equipment such as FT-NIR for routine soil analysis.

Soil organic carbon (SOC) plays a critical role in key soil functions, yet SOC is highly vulnerable to human activities, which can shift soil from acting as a net carbon sink to becoming a net carbon source. Despite considerable efforts to monitor soil conditions, traditional evaluations often focus on temporal comparisons within similar locations, which can limit the understanding of broader changes. To address these challenges, this study employs the pedogenon map framework to systematically compare SOC changes in soils under natural conditions (genosoils) and those affected by human activity (phenosoils) across Australia. By analysing SOC fractions with different mean residence times – mineral‐associated organic carbon (MAOC) and particulate organic carbon (POC) – the study aims to assess the durability of SOC changes over time. The methodology includes selecting soil profiles within pedogenons, evaluating SOC changes, and estimating the impact of these changes on SOC's residence time using the soil tonne‐year carbon (STYC) metric. The findings reveal significant spatial variations in SOC changes and their impact on residence time. Areas where SOC increased are mainly due to genosoils with lower SOC values, while genosoils with high SOC values have significantly decreased when converted to phenosoils, reflecting potentially unsustainable agricultural practices. Similar results have been found concerning the residence time of SOC regarding the genosoils SOC values. The study demonstrates that the pedogenon map is an effective tool for detecting SOC changes due to human activities, providing a robust framework for assessing soil carbon storage and informing targeted remediation strategies.

Despite the critical role of soil microbial communities in biomass production and ecosystem functioning, previous research primarily focussed on microbial structure without functional insights, especially for rare species. This study addresses this gap by exploring the functional potential of both abundant and rare bacterial communities across various land uses and soil groups in the Lower Namoi Valley, Australia. By integrating plant‐beneficial bacteria (PBB) and Functional Annotation of Prokaryotic Taxa (FAPROTAX) databases, we show that rare species exhibited higher alpha diversity and multifunctionality than abundant species. Cropping enhanced the biodiversity of abundant functional bacteria in fine‐textured soils, which promoted crop growth through increased PBB and carbon cycling. Conversely, rare functional bacteria exhibited consistently lower biodiversity in croplands. Random forest models and linear regression analyses identified land use as a significant factor influencing the biodiversity of rare functional bacterial communities, likely through plant–soil feedback systems. These findings underline the importance of land use in shaping bacterial community functionality and call for conservation strategies to protect soil biodiversity, especially rare species, to ensure sustainable soil ecosystems and support future food production.

In this study, a map of soil parent material is created to support the delineation of soil properties and classes of the Narrabri Shire, NSW. Currently, available information in this study area is geological and lithological maps at a scale of 1:250 000 to 1:1 000 000. These maps are not detailed, and the description in some areas is not accurate. Thus, this study created a new parent material map using information from the geological and lithology information, barest earth satellite imagery, gamma radiometric, topography, prior soil map and digital soil texture maps (clay and sand content). Based on interpretation and parent material observations, 18 parent material classes were delineated in the area. The 18 classes were then modelled using Linear Discriminant Analysis using Digital Elevation Model, slope, topographic wetness index, Gamma potassium (K) and thorium (Th), Ratio K to Th and soil visible and near infrared (NIR) reflectance (created using RGB and NIR bands) as covariates. This modelling process was iterated 50 times, and the most frequently predicted class was assigned to each of the 90 m × 90 m pixels throughout the study area. A map of the frequency of the predicted classes was also created to assess modelling uncertainty. The new parent material map consists of sedimentary residuals (sandstone), volcanic materials (basalt), alluvium and colluvium. The alluvium can be distinguished into six classes according to slope, soil information from satellite images and soil texture. The colluvium consists of three classes with a characteristic of high clay content (smectitic) and brown in colour (kaolinitic). Using similar approaches, such soil parent material or substrate maps could be developed for different regions in Australia. This method generated unique soil parent material classes combining stratigraphy, lithology and geomorphology.