Establishing robust geochemical baselines in the hyper-arid Atacama Desert remains challenging because of extreme climatic gradients, polymetallic mineralisation, and decades of intensive mining. To disentangle natural lithogeochemical signals from anthropogenic inputs, a region-wide, multi-institutional soil dataset (1404 samples; 32 elements) was compiled. The analytical workflow integrated compositional data analysis (CoDA) with isometric log-ratio transformation (ILR), principal component analysis (PCA), robust principal component analysis (RPCA), and consensus anomaly detection via hierarchical (HC) and spectral clustering (SC), applied both with and without spatial coordinates to capture compositional structure and geographic autocorrelation. Optimal cluster solutions differed among laboratory subsets (k = 2–17), reflecting instrument-specific biases. The dual workflows flagged 76 (geochemical-only) and 83 (geo-spatial) anomalies, of which 33 were jointly identified, yielding high-confidence exclusions. Regional baselines for 13 priority elements were subsequently computed, producing thresholds such as As = 66.9 mg · kg⁻¹, Pb = 53.6 mg · kg⁻¹, and Zn = 166.8 mg · kg⁻¹. Incorporating spatial variables generated more coherent, lithology-aligned clusters without sacrificing sensitivity to geochemical extremes (Jaccard index = 0.26). These findings demonstrate that a reproducible, compositional-aware machine learning workflow can separate overlapping geogenic and anthropogenic signatures in heterogeneous terrains. The resulting baselines provide an operational reference for environmental monitoring in northern Chile and a transferable template for other arid mining locations.

This data article describes a multi‐element soil‐geochemistry dataset collected over a six‐week field campaign (2 June–9 July 2025) across the La Negra Formation outcrop in the Antofagasta Region of northern Chile. The dataset comprises 101 primary Atacama Desert soil samples, 15 systematic field duplicates and 7 outcrop rock pulps collected in the field distributed across two parallel north–south transects with an average along‐transect spacing of ~10 km between consecutive points. The transects span six sectors in the communes of Tocopilla, Antofagasta and Taltal. Every sample was prepared following the Chilean national protocol for soil studies: air‐dried, sieved to the < 2 mm fine‐earth fraction, and pulverised for analysis, so the dataset characterises the conventional fine‐earth fraction of the soils. Each sample was analysed at AGS Análisis Geoquímicos (Coquimbo, Chile; lab batch CL25‐7655) under a single, internally consistent protocol. The AGS panel covers 62 analytes per sample, reported as aqua‐regia‐extractable concentrations quantified by ICP‐MS; because aqua regia is a partial digestion, values for elements hosted in resistant silicate phases are lower bounds on total contents. Six elements (Mn, Co, Ni, Cu, Zn, Pb) were re‐analysed by atomic‐absorption spectroscopy (AAS) on the same pulps at the Laboratorio de Geoquímica, Universidad Católica del Norte (UCN, Antofagasta, Chile), providing an independent inter‐method and inter‐laboratory check on the ICP‐MS values for these environmentally relevant trace metals. Quality assurance and quality control includes per‐element duplicate , the fraction of samples above the method detection limit, an inter‐method relative‐difference table, per‐element soil‐versus‐rock comparisons and a four‐level per‐element quality flag (11 of the 62 analytes classify as high‐quality, 40 as acceptable, 6 as low‐reproducibility and 5 as low‐detection) that identifies which analytes support quantitative reuse. We describe the sampling design, the laboratory workflow and the QA/QC outcomes. The dataset is released as a set of well‐documented CSV and GeoJSON files under CC‐BY‐4.0 and provides a single‐laboratory, single‐protocol baseline of fine‐earth soil geochemistry along the two sampled transect corridors of the La Negra Formation, against which future surveys of the soils of the Atacama Desert can be compared.

Machine learning (ML) is increasingly applied to geochemical baseline estimation and anomaly detection in soils and sediments, yet the methodological conditions under which machine learning outperforms traditional approaches—and which preprocessing and validation decisions most consequentially determine that advantage—remain incompletely characterized across environmental and mineral exploration domains. A structured systematic scoping review of 146 records from the Web of Science Core Collection applied sequential filtering to yield 78 thematically eligible studies, from which 20 were prioritized through a composite index integrating age-adjusted citation impact, platform usage, and semantic relevance. Four cross-cutting findings emerge. First, performance gains in environmental applications were driven primarily by spatial model structure rather than algorithm selection: incorporating a spatial covariate derived from geographically weighted regression raised test-set explained variance from $R^2=0.80$ to $R^2=0.96$ for cadmium mobility prediction in a geochemically heterogeneous karst setting, a gain the source study supported with a held-out test set and a Monte Carlo analysis of sensitivity to data size. Second, isometric or centered log-ratio preprocessing was applied in the majority of mineral exploration studies (three of five classical and hybrid studies and four of five deep-learning studies) but in none of the seven environmental studies, representing a systematic methodological gap with direct consequences for covariate importance estimates under compositional closure. Third, Shapley additive explanations and accumulated local effects functioned as instruments of operational value, enabling element-specific anomaly threshold derivation, training sample diagnosis, and grid-cell anomaly type classification; this evidence demonstrates that the accuracy–interpretability trade-off commonly assumed in the machine learning literature is not fundamental in geochemical applications but contingent on algorithm selection. Fourth, 90% of the 20 synthesized studies (18 of 20 by study-area location—13 in China and five in Iran) were evaluated under within-domain validation designs, and the consistently high performance metrics reported should be interpreted as interpolation estimates rather than evidence of transferable predictive capability. Geographic diversification of training datasets and spatially explicit cross-regional validation are identified as structural prerequisites for regulatory-grade applicability.

Chile, the world’s leading copper producer, accumulates hundreds of tailings deposits whose battery metals were never a recovery target. We screened cobalt, nickel and residual copper across the national tailings inventory using the public geochemical survey of the Chilean National Geology and Mining Service (SERNAGEOMIN, ∼2100 surface samples), deriving representative compositions for 649 deposits integrated with the October 2025 cadastre. Each metal returns a different answer. Residual copper is ubiquitous (median 0.15%, above 0.2% in 260 deposits) and its indicative endowment of ∼24 Mt, where the impoundments are counted at their authorised capacity, or ∼20 Mt on the tonnage actually deposited, over 90% of it in twenty impoundments above 100 Mt, is the most robust target. Cobalt sits just below crustal abundance (median 14 g/t) but shows an enriched tail: 104 deposits exceed twice the crustal reference, up to 1080 g/t, concentrated in small, old deposits. A discontinuity in the source release splits the nickel determinations into two segments with medians of 70 and 7.9 g/t; chromium, vanadium and a regional soil background all place the lower segment on the rock, so nickel is reported over its 304 deposits, at a median of 7.9 g/t, one sixth of the crustal reference. Cu, Co and Ni are uncorrelated at the deposit level ($\rho \le 0.13$), so surface geochemistry gives a plant no signal on which to base joint recovery. A comparison of surface against depth over twelve paired deposits also differs by metal: copper ordering is preserved, while the same twelve pairs are too few to resolve whether surface cobalt orders the interior. Assessing the battery metals of these tailings, therefore, requires sampling at depth, and the screening names the deposits where that campaign should start.

Narrative extraction algorithms construct storylines by finding coherent paths through document collections. The Narrative Trails algorithm frames this as maximum-capacity path optimization, where path quality depends on a coherence function measuring document relationships. We introduce quantum kernels as coherence functions for narrative extraction—to the best of our knowledge, the first systematic characterisation of quantum kernel methods for storyline extraction—and compare them against classical baselines on two corpora using a multi-seed protocol. The sweep covers 93 method evaluations (54 quantum kernels across three encoder families—R_Y+CNOT-ring, IQP/ZZ-feature-map, and a projected quantum kernel—and 39 classical kernels—cosine, RBF, and the cluster-aware Narrative Trails baseline). On 11,215 human navigation paths from Wikispeedia, evaluation metrics divide into two clusters that disagree with each other: alignment-based metrics (length-normalised DTW and per-step DTW similarity) favour methods that produce long alignment-rich paths, while set-overlap metrics (Jaccard and F1) favour methods that produce shorter paths with higher article overlap. On LLM-judged coherence for Cuban news storylines, evaluated under a 12-method × 5-seed × 30-endpoint-pair × 2-judge design (Claude Sonnet 4.5 and GPT-4o, both at $T=0$ via structured tool calling), the cluster-aware classical baseline is the top method in terms of mean overall coherence; the 5-method quantum-kernel pool and the 7-method classical-kernel pool on matched projection input show no significant differences after Holm correction. Cross-task analysis reveals that LLM coherence rank correlates with alignment-cluster Wikispeedia metrics (Spearman $\rho \approx +0.70$) and anti-correlates with overlap-cluster metrics ($\rho \approx -0.62$). A closed-form theoretical analysis shows that the depth-1 R_Y+CNOT-ring kernel reduces to a classical product-of-cosines kernel order equivalent to RBF, explaining the absence of empirical separation at low depth; deeper encoders break the cancellation but exponentially concentrate kernel values, eroding inter-pair distinguishability. Our results characterise quantum coherence kernels as competitive with classical kernels on the same projected input rather than decisively superior, with the cluster-aware classical baseline retaining a modest advantage attributable to its explicit topical structure.