Search Documents

By R. J. Durrheim, Y. Yabe, M. S. D. Manzi, DSeis team, N. Z. Nkosi, S. B. Mngadi, Jr. H. Ogasawara

On 5 August 2014, the Orkney M5.5 earthquake, the largest in South African gold mining districts, occurred with an unusual strike-slip mechanism at great depth of 4.78 km below the surface. In a rare case, the ML 5.5 earthquake occurred within the area covered by the legacy 2D and 3D reflection seismic data acquired in the 1990s and 2000s for gold exploration and mine development. In addition, the earthquake ML 5.5 rupture plane was recorded and accurately delineated by the underground in-mine seismic network near the source region. The integration of the legacy 2D reflection seismic data and mine seismicity data allowed us to identify a near-vertical structure, striking northnorthwest-south southeast (NNW-SSE). The ICDP-DSeis team drilled three holes (Hole A, Hole B, and Hole C), and two holes (Hole B and Hole C), intersected the upper edges of the ML 5.5 rupture. These holes recovered metasediments, metabasalts, intrusive rocks including dolerite sills and lamprophyre dykes adjacent to the fault zone, and the fault gouge. Late Prof. Tullis Onstott and geomicrobiologists installed a packer in Hole A and successfully recovered saline water and detected gas (~10 MPa). Slip weakening and rupture propagation are significantly influenced by the existence of fault gouges and the production of wear material between two sliding rock surfaces. Using the fault gouge material recovered from Hole C, we conducted friction experiments under dry and room temperature conditions at high slip velocity of ~100 mm/s and normal stress of 2 MPa. The resulting steady-state frictional strength was ~0.66 over a slip weakening distance of ~9.1 m. The steady-state frictional strength was high, which may be caused by large gouge thickness, and the rate of wear generation. As a result, it is proposed that the Moab Khotsong ML 5.5 seismogenic zone is complex and could be controlled by three main processes: a) the complex structural architecture of the seismogenic zone (e.g., intersection of fault, lamprophyre dykes, and dolerite sills); b) the mechanical process induced by tectonic and/or mining related stresses; and c) the mechanical and chemical processes caused by the water and rock interaction.

Jan 12, 2025

By I. James, R. J. Durrheim, M. S. D. Manzi, O. Valeshin, J. Lepine, L. V. Socco, C. Colombero, A. Malehmir, M. K. Rapetsoa, M. Sihoyiya

The FUTURE project aims to enhance seismic surveying in South African mining by integrating advanced acquisition technologies. In this paper, we detail surface and underground seismic surveys at the South Deep gold mine as part of the project. Surface seismic acquisition used co-located broadband micro-electromechanical sensors, one and three component recorders connected to geophones, distributed acoustic sensing using a straight fibre optic cable, along with a 6 ton broadband seismic vibrator operating with 4–180 Hz linear sweeps. In addition, underground surveys employed a combination of straight, and helically wound fibre optic cables, cabled-based and nodal sensors, along with a 500 kg drophammer mounted on a skid steer bobcat. The surface and underground seismic arrays were time synchronised using a GPS-time transmitter to allow for accurate seismic data recording in a GPS-denied environment. Results show the efficacy of these techniques in acquiring high-resolution seismic data, revealing detailed subsurface geological structures and mineralisation down to 3500 m below ground surface. Surface data revealed clear seismic reflections corresponding to the Black Reef and Ventersdorp Contact Reef. An underground P-wave velocity model exhibits low velocity anomalies that are associated with fracturing, while a P-wave reflection stack section shows significant reflections and a degree of faulting below the tunnel floor that was not observed on the surface reflection seismic data. We highlight the potential of these technologies to enhance safety, efficiency, and sustainability in mining. This study sets a benchmark for future reflection seismic surveys, paving the way for safer, more efficient, deep mineral exploration and extraction practices.

Jan 12, 2025

By R. J. Durrheim, B. Manzunzu, V. Midzi

Selecting ground motion prediction models (GMPM) is crucial in assessing seismic hazard levels in a specific region. This study investigates the influence of ground motion model (GMM) selection on seismic hazard assessment in Johannesburg, South Africa. Most seismic events affecting Johannesburg are related to mining activities and water ingress into the underground mine voids. The most seismically active regions in South Africa include areas around Klerksdorp, Welkom, and Carletonville. These regions experience a significant number of earthquakes, with mining activities being a major factor contributing to seismic events. Ten ground motion prediction equations were selected from regions of similar tectonics setting and were assessed to see if they correspond to the available local data. Six of these were also compared to the ground motions produced by the earthquake of 14 August 2014, which occurred in the Orkney area, one of the mining areas in South Africa. Sensitivity analyses were also carried out to assess the influence of ground motion prediction equations (GMPE) on the seismic hazard of Johannesburg. The results indicate that the choice of GMPM significantly affects the estimation of seismic hazard levels in Johannesburg. Different GMM can lead to varying ground shaking intensity predictions, impacting risk assessments and engineering design considerations. The findings highlight the importance of carefully selecting appropriate GMM tailored to Johannesburg’s specific geological and tectonic characteristics. Understanding how different models perform in this region is essential for accurate seismic risk mitigation strategies.

Jan 12, 2025

By P. Lenegan, R. Masethe, R. I. L. Ferreira

Underground mining in high rock-stress environments can induce unanticipated dynamic rockmass deformation, posing a serious risk to the workforce. Data collected by seismic networks are imperative for the design-as-you-mine process. Often an inappropriate assessment of the seismic hazard is done when unnecessary data are included in analytical techniques, particularly short-term time frames. Depending on the methodology or application of data analysis and interpretation, some of the information may not be relevant and even detract from the objective. By using the spatial event clusters associated with active workplaces to define the shape and size of the polygons for data selection for subsequent analysis, a problematic subjective component can be eliminated. The mining of tabular reefs in South Africa, be they gold- or platinum group metals, is very extensive, and underground workplaces are often widespread. Production stoppages are very costly, especially for marginal mines, so there is just cause not to evacuate an entire operation’s workforce following the occurrence of a large mining-induced seismic event. A rationale based on the damaging peak particle velocities (PPV) of the past seismicity can exclude the unaffected underground personnel and allow a quick orderly withdrawal where required. Micro-deformation may precede rockbursting, which can occur on-shift close to the workface. A mine-wide array of geophones suitable for the accurate location and analysis of seismicity may not be able to detect these higher-frequency emissions. Inexpensive but fit-for-purpose accelerometers, providing real-time, continuous measurement of acoustic rockmass deformation near the workplace, integrated through underground and surface communications networks, are shown to record ‘rock-talk’, clearly distinguishable from the instrument background noise, which can be in the order of hours leading up to an imminent damaging seismic event.

Jan 12, 2025

By T. Matlou, S. Myendeki, V. Midzi, T. Zulu, H. Mabotja, T. Mulabisana, B. S. Zulu, L. Hlatshwayo, T. Makhateng

The Council for Geoscience (CGS) performed a macroseismic investigation following the occurrence of an earthquake in the Boksburg area on 11 June 2023. The event had a magnitude of ML 4.4 and a focal mechanism that showed the event as having strike-slip motion along a northwest-southeast oriented fault. The event resulted in aftershocks, some of which occurred away from the causative fault in the east in an area without a mapped fault. However, the cluster of past events in that area suggests that there is an active fault. Due to the size of the main earthquake, a macroseismic survey was conducted to determine its impact on the community in the region. Analysis of observed data obtained from questionnaires resulted in the creation of 153 intensity data points. Their distribution showed high intensity values close to the earthquake source area. However, the investigation highlighted a number of anomalies, such as the higher than expected intensity values observed far from the event epicentre. It could be that these anomalies are caused by site conditions, for example, amplification of ground motion due to thick soil layers.

Jan 12, 2025

By S. Durapraj, R. T. Masethe, U. Maqina, A. C. Adoko

This study investigates seismicity and rockburst control within the Merensky Reef and UG2 orebody of an intermediate-depth platinum mine located in South Africa's Bushveld Complex. Utilising a mine-wide seismic monitoring system, over 1,900 seismic events were recorded between April 2023 and May 2024, revealing complex seismic patterns driven by mining-induced stress changes. The largest event, with a magnitude of ML 2.0, underscores the seismic risks associated with mining at increasing depths. Key findings include the spatial and temporal variability of seismic events, with high seismic activity observed near pillars and geological structures. A significant portion of events correlated with pillar and foundation failures. Moment tensor analysis identified dominant failure modes, such as stress-driven fracture closures and dynamic pillar failures. To mitigate risks, the study emphasises advanced support system redesign and rockburst risk assessment using the rockburst damage potential (RDP) method. The research highlights the importance of monitoring, geotechnical compliance, and adaptive designs in enhancing safety and reducing seismic hazards in underground mining operations. The study applied the RDP framework, classifying damage across R1 to R5 scales, where most instances fell into low-risk categories (R1 and R3), indicating minor to moderate damage. However, data limitations and the tabular stoping method's complexity led to inconclusive results, highlighting the need for more robust data and customised risk assessment tools. Despite these challenges, the RDP approach provided valuable insights into rockburst mechanisms and the effectiveness of existing support systems.

Jan 12, 2025

By Tim Lundquist

Disa Technologies Inc. provides innovative, sustainable solutions for mineral processing and remediation through its patented High-Pressure Slurry Ablation (HPSA) technology. HPSA processes solid feed material in slurry form, propelled by high pressure pumps through opposing nozzles that generate impinging jets within a collision housing. This process induces particle-to-particle collisions in a high energy zone, creating intergranular fractures along grain boundaries. In our batch scale and continuous operating testing, we have found that controlling key processing variables such as nozzle velocity, slurry solids percentage, and the number of passes can allow for controlled size reduction and maximum mineral liberation. Since 2018, Disa Technologies has demonstrated the versatility of HPSA by testing it on diverse materials, including phosphate, copper, uranium, vanadium, zinc, rare earth elements, lithium, nickel, mineral sands, gold tailings, and more. The intent of this paper is to present an overview of our test results describing in detail impacts on size reduction, recovery, and grade with a focus on HPSA's adaptability across ore types. Looking forward, it explores the potential to process materials such as DR-grade iron, clay, PGMs, bauxite, and various tailings. Additionally, it examines planned advancements in HPSA technology, including optimized nozzle spacing, higher capacity models, improved tonnage-to-footprint ratios, enhanced maintenance features, and expanded nozzle configurations.

Jan 1, 2025

By Clayton Vernon Deutsch, Marcel Antônio Arcari Bassani, João Felipe Coimbra Leite Costa, Vanessa Cerqueira Koppe, Pedro Henrique Alves Campos

Traditional linear approaches oversimplify the complexities of ore blending. Adopting nonlinear blending models in mining and mineral processing is critical to enhancing prediction accuracy and allowing optimization regarding nonadditive variables, such as metallurgical recovery. This study implements a simulated annealing algorithm in short-term mine planning that seeks to optimize the metal recovery by considering how to blend the ore in the mine better. Two nonlinear metallurgical recovery models are used as inputs to the algorithm to represent synergistic and antagonistic blending. The results of the case study demonstrate that the optimized schedule exhibits improvements in both blending behaviors when contrasted with conventional linear-based scheduling plans.

Jan 1, 2025

By Michael Forbes, Rick Jeuken

The challenge of managing geological uncertainty in mine planning is central to optimizing resource extraction while minimizing risks and costs. This paper introduces a novel approach to coal mine planning by integrating chance-constrained programming to address the uncertainty inherent in mining operations. The main takeaway from this work is the demonstration that focusing on reducing the variance of high-quality coal blocks, rather than low-quality ones, can lead to better blending outcomes, higher profitability and more efficient resource utilization. The study presents a case study of a coal mining operation, using chance-constrained programming to manage the uncertainty surrounding coal property predictions, specifically the predicted coke strength after reaction (CSR).

Jan 1, 2025

By John Thella, Edward McNew, Christina Thella, Mends Emmanuel, Pengbo Chu

Ore sorting is an emerging technology in mineral processing that can produce a significant impact on the total mining and processing economics. The technology is capable of preconcentrating an ore based on differences in the mineralogical characteristics, which can include color, brightness, transparency, size, shape, density or near-infrared response. The current practice of conducting an ore sorting test typically requires large amounts of materials to be sent to a commercial testing facility, which is expensive and time-consuming. Here, we present a novel method to simulate commercial ore sorting on a laboratory scale using a rare earth ore.

Jan 1, 2025

By B. Genc, V. J. Mahadeo, S. Ramparsad4, U. Govender

The study was intent on assessing the socio-economic impact of a zinc mine in South Africa through the application of people-centric digital tools using a case study research methodology. The global mining industry is undergoing significant technological shifts to modernize, improve operational sustainability, and align with the environmental, social, and governance (ESG) agenda. While the case study mine is a more modernized mine, continuous improvement around the ‘social’ pillar becomes critical, and to this end application of people-centered digital tools presented as the vehicle to understand community needs and impacts. Despite considerable annual corporate social investment by mines annually, the overall impact or social return on investment is marginally understood. Towards contributing to addressing this gap, digital ESG ‘social-related’ data gathering tools for mines decision-making were developed, including the training-needs assessment tool, community-social needs assessment, small, micro, and medium enterprise (SMME) communications and engagement tool, digital leadership competencies gap assessment tool, and change management blueprint, which further allowed for customization to the mining context. Hence, the novelty of this paper relates to applying digital tools to provide real-time information of the needs of a community, both socially and economically. The findings suggested that these digital tools enabled gathering of information for the establishment of regional and operational baselines towards enabling of social investment returns in relation to socio-economic transformation development interventions. These tools have been applied to a zinc mine context, with lessons and experiences obtained through this study, invaluable to the application of these digital tool in other mines in South Africa. The study was conducted in September 2023 across five host communities, surrounding the zinc mine, in the Northern Cape region, with 120 participants involved in the application of three relevant community-related digital tools, which included the training needs, community-social needs, and SMME communications and engagement tools. It was evident that the digital ESG tools could be used for data gathering for the mine’s informed decision-making and allowed for securing of insights on community needs in near real-time, and further proved practical for community use.

Dec 11, 2024

By F. Ntuli, T. V. Nyangadzayi

Chromium green is an essential metal with various applications in the paints, construction, electronics, and aerospace industries. The main source of chromium green is the carbo-thermic reduction of chromite ores, however, chromite ores are being heavily depleted on a global scale, hence the need to find alternative or secondary sources for the chromium. In order to meet the global demand for chromium there is a need to implement principles of the sustainable circular economy to keep chromium within the material loop. Ferrochrome slag and stainless steel slags are secondary resources that are potential sources of chromium. In this work, recovery of chromium from ferrochrome slag was investigated. The ferrochrome slag was alkaline leached using sodium hydroxide- sodium hypochlorite solution to reduce the use of potent reagents that have a potential negative environmental impact. The leachate produced by oxidative alkaline leaching of ferrochrome slag contained hexavalent chromium, which was used to recover the extracted chromium through a two-stage process involving reduction and precipitation. Reduction of hexavalent Cr to trivalent Cr was investigated using anhydrous FeSO4 at concentrations of 5 g/L, 10 g/L, and 15 g/L. Precipitation was conducted by dosing the reduced leach solution with 1 M sodium hydroxide at dosages of 5 ml/L, 10 ml/L, and 15 ml/L. The experiments were designed using the Taguchi (L9) orthogonal array and analysed according to the ANOVA, general linear model. The optimized conditions for the reduction step were ferrous sulfate dosage of 5 g/L with a pH of 12, and for the precipitation step, 1 M sodium hydroxide at a dosage of 15 ml/L with a setting time of 90 minutes and pH of 7. Physical changes were observed during the two-step chromium recovery, with solution changing from a yellowish colour to dark green colour indicating completion of the reaction. The chromium recovery was found to be 93.3%.

Dec 11, 2024

By J. Petersen, T. Moyo, D. Maharaj

This study investigates the effect of ammonium thiosulfate on the leaching of gold from the contacts of unpopulated cut electronic printed circuit boards (PCBs). Studies in the literature show ammonium thiosulfate to be a viable option in comparison to the commonly used but hazardous cyanide leaching route. However, the simultaneous dissolution of Cu interferes with the Au leaching reaction, often limiting the extraction to moderate levels. To optimize the system, an ammoniumthiosulfate system containing 0.5M ammonia, 0.5M ammonium thiosulfate, and Cu(II) sulfate at varying concentrations was studied as an alternative hydrometallurgical option for the recovery of gold from PCB. Tests were done with unpopulated standardized model PCBs (cut to 15 x 30mm pieces). Gold extraction of 47.3% was achieved after a 24-hour leaching experiment with a 0.02M copper background concentration. Further experiments focused using different initial background Cu concentrations (0M – 0.1M) to determine its effect on both gold and copper extractions. An initial Cu concentration of 0.045M resulted in the highest gold extraction, with 93.7% extraction after 6 hours of leaching. Further analysis of the data shows that the kinetic rate of gold leaching is correlated in a first-order dependence with the actual copper concentration in the solution, which increases with the concomitant dissolution of copper and gold from the PCBs. A lower threshold of 0.017M and an upper threshold of 0.15M copper in solution were determined, in between which gold extraction shows this first-order rate increase. The upper threshold is potentially linked to the onset of copper precipitation, which blinds the metal surfaces, slowing further leaching. The findings of this study suggest that in an industrial recovery operation of gold from waste PCB using the thiosulfate method, copper concentrations need to be carefully monitored, and the process should preferably operate with uncut or coarsely cut untreated boards to minimize exposure to copper circuits.

Dec 11, 2024

By K. C. Sole, M. Makonese

Bushveld Belco is commissioning a plant for production of electrolyte used in vanadium redox flow batteries. Owing to the source material, the electrolyte is contaminated with Fe, at concentrations up to 140 mg/L, which negatively affects battery performance. This study examined the use of ion exchange to reduce the Fe concentration to the target value of < 100 mg/L from a feed solution containing 100 g/L V in 4 M H2SO4. Four resins with possible application were identified: Puromet MTS9570, Puromet MTX7010, Puromet MTS9500, and Puromet MTC1600H. Puromet MTS9570 outperformed the other three resins in all batch experiments with respect to loading capacity and selectivity over V, and was evaluated for Fe removal in fixed-bed column loading tests. Based on these data, preliminary sizing and design of a full-scale column was undertaken. Owing to the difficulty of eluting this resin, a philosophy of discarding the resin on loading to full capacity was considered. These data revealed that, despite the high selectivity of this resin for Fe over V, it would be too expensive and impractical to use this resin for full-scale operation under such conditions. It is recommended to remove iron earlier in the flowsheet prior to generation of the 4 M H2SO4 electrolyte.

Dec 11, 2024

By F. Dinter, C. McGregor, Q. G. Reynolds, L. Hockaday

The necessity of reducing greenhouse gas emissions to limit the impact of anthropogenic climate change means that all industrial processes should investigate their options towards decarbonization. Iron, steelmaking, aluminium, and cement production produce the most industrial emissions. Although most research focuses on decarbonizing these industries, minerals processing industries must be decarbonized to achieve net zero emissions goals by 2050. Incorporating renewable energy sources and avoiding the combustion of fossil fuels are two ways to reduce greenhouse gas emissions. This paper reports on the results from a dynamic process model developed to investigate the feasibility of concentrating solar thermal pretreatment of manganese ores to pretreat carbonate-rich manganese ores for increased ferromanganese smelter productivity and reduced greenhouse gas emissions. The results indicate that pretreated ores reduce the energy requirement for smelting significantly for some ores and lowers total carbon dioxide emissions by 13 to 19% compared to the traditional smelting route. A high-level economic evaluation shows that solar thermal treatment is financially viable for low cryptomelane, high braunite ore that is common in the Kalahari Manganese Field.

Dec 11, 2024

By L. J. Su, J. Y. Zhang, P. W. Gu, J. F. Xu

Surface tension of mold flux is a key factor affecting optimum casting conditions. A thermodynamic model was developed to determine the surface tension of mold flux containing fluoride based on the ion and molecule coexistence theory of slag structure and Butler’s equation. The relationship between composition and surface tension was investigated by this model. Results indicated that the calculated values showed good agreement with literature data, and the average error was 8.19%. The surface tension value of the CaO–Al2O3-based mold flux was larger than that of the CaO– SiO2-based mold flux. The surface tension in a multi-component system decreased with increasing contents of SiO2, CaF2, and Na2O, which are considered surface-active components. This trend was more significant in the CaO–Al2O3-based mold flux.

Dec 11, 2024

By O. S. L. Bruinsma, M. J. Matsela, D. J. van der Westhuizen

The free ligand (FL) model is an intuitive model to calculate the equilibrium concentrations in solvent extraction (SX) based on the distribution coefficient (Kd) at infinite dilution and the fraction of free ligand available for extraction. In this work, the FL model is validated to account for co-extraction of Ni and for designing the extraction and scrubbing sections of Co/Ni separation in a base metal refinery. Cyanex 272 is employed as a selective Co extractant in the process to yield high purity Ni. To enhance the loading capacity of the extractant, 50 mol% of NaOH is pre-loaded into it. Experimental and FL modelling results show that four equilibrium stages are required to purify the 99% Ni feed to a 99.97% Ni raffinate. The 53.59% co-extract with 26.76% free ligand is purified with a reflux scrub of 30 g Co/L in one equilibrium stage to 99.01%, Co in the scrubbed extract. The FL model as well as the Co/Ni/Na-analysis of the aqueous phase confirm that in the scrubbing section, Ni in the extract is mainly replaced by Co. Finally, in the one stage stripper the Co concentration in the scrubbed extract is increased from 5.6 g Co/L in the extract to 30.5 g Co/L that is partially used as reflux scrub liquor.

Dec 11, 2024

By IPMI Educational and Scientific Foundation

It is with great pleasure that we announce the availability of this fifth issue of the IPMI Journal, a publication of the International Precious Metals Educational and Scientific Foundation. This digital peer reviewed publication has been created as a vehicle for the sharing of technical information related to the science and technology of precious metals. Since 1976, the International Precious Metals Institute has championed the exchange of information of critical importance to the precious metals scientific and commercial communities through its various communications vehicles. This has included instructional videos, annual conference symposia, special topic committee meetings, regional seminars, and professional and student achievement awards. Due to the breadth of important topics related to all aspects of precious metals, it has been the vision of the IPMI to provide a publication available to the general public that would become a key source of pertinent topics containing high quality publications by experts in their fields of study. Historically, the IPMI has attracted world renowned experts in many fields (including Nobel Laureates Henry Taube, Ei-ishi Negishi, Robert Grubs, Professor Ben Feringa, and David MacMillan) in the dissemination of discoveries, inventions, and industry proven practices. Such information has a total value greater than the sum of its parts as seemingly unrelated innovations from distinct focus areas can be adapted to solve problems. Because of this important characteristic of information exchange, we have decided to create a digital journal, available to the general public, that will not only communicate state of the art discoveries and sound practices, but will also review historical communications from the IPMI’s archives that have value and use even in today’s environment.

Dec 1, 2024

By Wedam Nyaaba, Edward Wellman, Angelina Anani, Eduardo Cordova, Sefiu O, Cluber Rojas, Adewuyi

This study aims to investigate the optimal method for ore pass construction in incompetent rock masses. We evaluated the conventional and raise boring methods based on safety, efficiency, excavation control and ground support for ore pass construction. A stability analysis using analytical Q-raise (QR method) and kinematic analysis methods for ore pass construction with a raise borer before and after grout injection of the rock mass was performed. As a case study, an ore pass (diameter of 3 m, depth of 100 m) within an incompetent rock mass was considered to gain further insight. The rock mass was characterized according to the classification methods Q Barton, rock quality designation (RQD), rock mass rating (RMR) and geological strength index (GSI). The grout intensity number (GIN) method of grout injection was used. The safety factor of less than 1.075 obtained before injection was lower than the acceptance criteria. However, grout injection before raise borer excavation resulted in a safety factor greater than 1.5. Using a raise borer without pre-grouting indicated that the maximum unsupported diameter (MUSD) of the ore pass was less than the required 3 m and equal to or larger than 3 m post-grouting.

Dec 1, 2024

By Firat Atalay

Estimating the spatial distribution of ore grade is one of the most critical steps to proceed with an investment decision on a deposit. Kriging is the most widely used method to estimate the ore grade, but alternative techniques are being developed. Machine learning algorithms can be used as alternative methods to classical kriging. In this paper, the iron (Fe) grade of a deposit is estimated with the XGBoost algorithm, and the results are compared with kriging estimation results. For the estimations, samples collected from drillholes were used. Due to the different natures of the estimation methods, different steps were taken to perform the estimations. The results show that XGBoost produced higher-range estimates, which is a desired result in ore grade estimation, and the minimum and maximum of the estimates were lower and higher than the kriging estimates, respectively. However, like kriging, the estimation results were smoother than composites, and the variance of the XGBoost estimates was lower than the variance of composites. This means that even though estimation with XGBoost mitigates the smoothing effect, the estimation results suffer from smoothing like kriging.

Dec 1, 2024