Journal Paper Digests 2026 #12
- When spectral libraries are too complex to search: Evolutionary subset selection for domain-adaptive calibration
- Soil-health frameworks in agri-food systems. A review
- Soil carbon markets for climate change mitigation? Pragmatic economists and matters of concern
- Demystifying Geographic “Laws” for Soil Mapping via Interactive Geovisualization
- Causal network construction and quantification in complex ecosystems
- A global synthesis of spectroscopy-based prediction accuracy for soil carbon fractions: A systematic review
- Development and validation of physically constrained machine learning for improving remote sensing-based evapotranspiration estimation
- Evapotranspiration Everywhere, All the Time: Towards a Unified View From Earth Observation
- Soil Organic Carbon Improves Crop Yield and Yield Resilience
Soil Organic Carbon Improves Crop Yield and Yield Resilience
Increasing soil organic carbon (SOC) has been proposed as a strategy to promote crop yield resilience under extreme hydroclimatic stress, particularly drought, due to its positive effect on soil available water-holding capacity. We analyze how SOC mediates the relationship between US rainfed crop yields (1981-2020) and growing season hydroclimatic conditions-defined by soil water supply and atmospheric water demand-over 67,000 county-years across three major crops. We show that higher SOC is consistently associated with increased crop yields and yield resilience, evidenced through reduced interannual variability. Contrary to prevailing expectations, the largest yield gains from SOC occur during moderate water supply conditions-not drought. Under drought, water supply to crops may be more limited by water inputs to soil than by soil water-holding capacity, constraining the benefit provided by SOC regeneration. Furthermore, because moderate conditions are more frequent than drought or wet extremes, the largest production gains from SOC accumulate under moderate conditions. These findings indicate that SOC regeneration can enhance drought resilience to some degree but cannot compensate for extreme water scarcity; the services SOC provides to crops, including water storage, require water for their effective delivery.
Evapotranspiration Everywhere, All the Time: Towards a Unified View From Earth Observation
Scientists want to know everything, everywhere, and all the time. This is particularly true in Earth science, where we seek to understand processes that span from the molecular to the planetary scale in how the world works, how it affects us, and how we impact it—especially the water cycle. Evapotranspiration (ET) was the last component to be measured in closing the water cycle: for decades, closing the water budget meant adding up all the measurable components, then inferring ET as the residual. Early measurements relied on water loss from pans and weighing lysimeters, followed by sensors inserted into plants to monitor sap flow and leaf chambers capturing transpiration. Scaling up to ecosystems became possible through eddy-covariance flux towers and further across landscapes through proximal sensing with drones, aircraft, and, ultimately, with satellites. While enormous progress has been made to measure or estimate ET everywhere and all the time, no single approach has yet achieved both simultaneously. Flux towers help with all the time, but not everywhere. Satellites can do everywhere, but not all the time (except, in part, for geostationary satellites, though with insufficient spatial coverage and resolution). A new advent of smallsat constellations is moving us to everywhere and all the time in detail, though we are only in the beginning of that era. This paper discusses the evolution and revolution of Earth observation for ET, as we advanced from the first Landsat and development of ET models through the progression of increasingly higher spatiotemporal resolution across international space agencies and commercial industry with increasing ET model sophistication, cloud computing, and machine learning. We continue to march ahead towards ET everywhere, all the time, and use that knowledge to better manage water and sustain our planet.
Development and validation of physically constrained machine learning for improving remote sensing-based evapotranspiration estimation
Accurate estimation of terrestrial evapotranspiration (ET) is vital for understanding water and carbon cycles. Advances in satellite remote sensing (RS) techniques have greatly prompted the development of ET models, yet their performance varies inconsistently across biomes due to structural and parameterization errors. Machine learning (ML) offers data-driven alternatives, but often lacks physical mechanism-based generalization. Here, we employed automatic ML (AutoML) to develop three model categories: unconstrained ML and deep learning (DL) models, input data-constrained models integrating physical ET estimates into training data, and loss function-constrained models incorporating physical equations into DL loss functions. Validation with site-based observations in the Heihe River Basin (HRB) (75% training, 25% validation) showed all models achieved root mean square errors (RMSEs) below 0.6 mm/d, with physical constraints enhancing generalization under extreme conditions. Out-of-sample tests indicated physical constraints improved spatial extrapolation ability, with DL models benefiting most with RMSEs reduction from 0.32 mm/d to 0.47 mm/d. Utilizing the best unconstrained (MLU), input data-constrained (MLI_SGC), and loss function-constrained (DLL_PM) models, we generated daily ET for the HRB spanning 2000–2021. Water balance-based validation revealed DLL_PM reduced mean absolute percent errors (MAPEs) by 53.9%, 5.2%, and 4.1% in the upper, middle, and lower reaches versus MLU, while outperforming MLI_SGC and two mainstream RS ET products (ETMonitor, PML-V2). Furthermore, we demonstrate that input data constraints enhance physical consistency by increasing the importance of physically based ET features, whereas loss function constraints reshape the DL learning process by modifying network weights and biases. The Akaike Information Criterion (AIC) further indicates that physically constrained models achieve accuracy gains with negligible increases in model complexity during training. These findings represent a meaningful step toward understanding how to effectively integrate physical constraints into ML models for ET estimation, and hold promise for advancing large scale water and energy cycle assessments under changing environmental conditions.
A global synthesis of spectroscopy-based prediction accuracy for soil carbon fractions: A systematic review
Infrared spectroscopy combined with machine learning (ML) offers a rapid and cost-effective approach for quantifying particulate organic carbon (POC) and mineral-associated organic carbon (MAOC) on large spatial scales. However, its predictive accuracy has not been systematically evaluated across diverse ML models, climate zones, and soil types. A systematic search of Web of Science and Google Scholar (before December 2024) identified 108 observations from 27 eligible studies that applied mid-infrared (MIR) or near-infrared (NIR) spectroscopy to mineral soil samples using ML modeling and reported predictive performance and sample size for POC and/or MAOC. This study employed a three-level random-effects model with Fisher’s Z-transformed effect sizes, with no evidence of publication bias detected (Egger’s test, P = 0.71). Our synthesis revealed robust predictive capacities for both fractions, yielding pooled correlation coefficients (r) = 0.88 (95% CI:0.85–0.90) and 0.83 (95% CI:0.78–0.87) for POC and MAOC, respectively. MIR spectroscopy and partial least squares regression (PLSR) achieved the highest prediction performance for POC and MAOC among spectroscopy types and model families. Meta-regression revealed that soil texture and latitude were the dominant drivers of POC prediction, indicating that the prediction performance increased with increasing latitude. No single dominant driver emerged for MAOC, suggesting that the multiple interacting factors governed its predictive performance. The identification of these drivers remains challenging due to the predominance of agricultural studies and the sparse reporting of secondary minerals (e.g., Fe/Al oxides). MIR showed predictive advantage specific to MAOC, while POC prediction performance was less constrained by spectroscopy type and showed a more consistent trend across soil textures. MAOC prediction was highly dependent on soil texture, with higher accuracy in coarse or sandy soils than in medium-textured soils. These findings support fraction-specific spectral strategies for improving large-scale SOC stock estimates.
Causal network construction and quantification in complex ecosystems
Understanding and identifying causal relationships within complex, dynamic ecosystems is essential for elucidating ecological mechanisms and guiding effective ecosystem management. Conventional statistical approaches predominantly quantify correlations among multiple variables, yet fall short of capturing causality. Existing causal-discovery tools in ecology either focus on pairwise interactions, thereby overlooking emergent effects arising from the simultaneous influence of multiple drivers, or fail to provide a direct metric for the strength of causal control. Consequently, there is an urgent need for a framework that can simultaneously reconstruct causal networks among numerous variables and furnish quantitative assessments of causal importance. To address this challenge, we developed a dual-strategy ecological causal-discovery (DS-ECD) model founded on Granger causality that integrated a forward local-search strategy with a backward global-search strategy to detect causal links in long-term time-series data. The local strategy employed a forward greedy search to construct an information set capturing significant individual-level causal relationships, while the global strategy utilized a backward one-step search to uncover group-level causal interactions. In addition, we introduced the Relative Causality-Driven Intensity (RCDI) metric to quantify causal strength by decomposing direct and indirect effects, which complemented existing causal-discovery tools. Simulation experiments demonstrated robust model performance under high noise and high dimensionality. Accordingly, DS-ECD was deployed in ecosystems across different scenarios, rapidly revealing intuitive causal networks together with their associated RCDI values. Being purely data-driven, the approach promptly delivers transparent causal graphs and quantitative intensity values, facilitating cross-validation with experimental or process-model results and offering a reference for ecosystem-management practices.
Demystifying Geographic “Laws” for Soil Mapping via Interactive Geovisualization
“Laws” of geography such as Tobler’s First Law (spatial autocorrelation) and Zhu’s Third Law (environmental similarity) offer fundamental principles for spatial prediction and mapping, yet their implications for digital soil mapping (DSM) are often opaque because the underlying principles and mechanisms of DSM models are rarely inspectable in typical DSM workflows. This study presents an interactive geovisualization portal that demystifies Tobler’s Law, Zhu’s Law, and a combined formulation in spatial prediction processes, using soil organic matter (SOM) concentration prediction in Xuancheng, China, as a case study. The portal integrates multiple DSM frameworks that operationalize two geographic laws—inverse distance weighting (IDW), individual predictive soil mapping (iPSM), an iPSM-IDW hybrid, ordinary kriging (OK), and regression kriging (RK)—and couples them with user-configurable parameters such as neighborhood size, distance-decay factor, and variogram model. The portal provides coordinated, interactive views that link SOM predictions to dynamic map and diagnostic statistical charts for explaining location-level predictions, visualizing the manifestation of geographic laws in constructing local predictions, examining weight allocation patterns, and assessing overall prediction accuracy. Additionally, a built-in sensitivity analysis enables users to investigate and understand the effects of varying the geographic law, modeling framework, and modeling parameters on prediction results. This geovisualization portal advances interpretable DSM by rendering its underlying geographic principles, model mechanics, and parameter influences visually inspectable.
When spectral libraries are too complex to search: Evolutionary subset selection for domain-adaptive calibration
Background: The increasing availability of large spectral libraries offers new opportunities to reduce the costs and efforts required to develop spectroscopy-based sensing techniques for rapidly and non-destructively estimating key properties across environmental and agricultural matrices. These libraries can provide training samples for developing models adapted to specific target domains. However, identifying which samples are most relevant for a given target domain remains challenging. This study introduces gesearch, a non-linear evolutionary algorithm for selecting target-domain-relevant training samples from complex spectral libraries to build accurate and interpretable quantitative models. Results: The gesearch, method was used to extract a subset of relevant samples from a large North American infrared soil spectral library to build predictive models of total carbon for an independent target area in the Democratic Republic of the Congo In this challenging cross-domain test case, simple linear models built with the training samples found by gesearch achieved superior accuracy compared with established modelling approaches, including LOCAL, Cubist, convolutional neural networks, and global partial least squares regression. Significance: The proposed method provides a practical framework for exploiting large heterogeneous spectral libraries when only limited target-domain reference data are available. By selecting samples that are spectrally and compositionally coherent with the target domain, gesearch supports accurate, compact, and interpretable calibration models. It can also operate when only unlabelled target-domain spectra are available.
Soil-health frameworks in agri-food systems. A review
Soil health is central to agroecological transitions, yet guidance for integrating it into agri-food system design and monitoring remains fragmented. Institutions increasingly use frameworks to define indicators, guide interventions, and report progress against climate, biodiversity, and food-security agendas. However, to our knowledge, there is no integrative soil health framework which coherently links biophysical diagnostics, socio-institutional enablers, and multiscale accountability. This leaves critical gaps in design, sequencing, and measurement of agroecological transitions. Here we review how soil health is operationalized within agroecology and agri-food systems and translate these patterns into an actionable programming guide. We reviewed 64 frameworks and extracted 652 indicators across 12 agroecological principles to build a framework-by-principle evidence matrix. Frameworks were classified by use-orientation (theory, practice, analysis), and indicator thematic profiles were analyzed using hierarchical clustering with adaptive branch detection. The major findings are as follows: (1) framework evolution exhibits four chronological waves with shifts from conceptual foundations to operational measurement and outcome reporting, alongside changes in global and regional agenda setting and a rising demand for comparable indicators; (2) clustering identified five soil health design domains separating biophysical and socio-economic principles and revealing stable micro-constellations beyond earlier pathway framing. These include soil management and input stewardship, soil-health assessment, agroecological and ecosystem-based, integrated landscape and livelihood, and policy- and outcome-based. These findings were translated into a sequenced, multi-domain programming architecture that operationalizes complementarity across diagnostics, stewardship implementation, ecosystem safeguards, landscape–livelihood embedding, and iterative learning, thereby closing the gaps between farm practices, governance mechanisms, and outcome monitoring for soil health.
Soil carbon markets for climate change mitigation? Pragmatic economists and matters of concern
Soil carbon markets are increasingly promoted as climate mitigation instruments, yet their emergence is uneven, contested, and shaped by complex socio-technical configurations. Through comparative research in Taiwan and the United Kingdom, this paper examines how these markets are actively constructed rather than pre-given, highlighting the critical but often overlooked role of mediators whom we conceptualise as pragmatic economists. Drawing on interviews, workshops, and stakeholder mapping, we show how pragmatic economists translate scientific and metrological knowledge, assemble infrastructures for measurement and certification, and align agricultural, policy, and commercial interests. Their practices extend but also complicate Callon’s concept of economists in the wild, revealing marketisation as a situated, relational, and performative process. Across both sites, we identify key matters of concern, including scientific simplification, fragmented governance, unequal power relations, and new dependencies between farmers and mediators. By foregrounding pragmatic economists, the paper advances debates on the political economy of environmental markets and underscores the need for more reflexive, ecologically attentive, and socially just approaches to governing soil carbon within wider decarbonisation strategies