Mineral resources are of great significance in the development of the national economy. Prospecting and forecasting are the key to ensure the security of mineral resources supply, promote economic development, and maintain social stability. The methods for prospecting prediction have evolved from qualitative to quantitative prediction, from empirical research to mathematical analysis. In recent years, deep learning algorithms have gradually entered the attention of geologists due to their robust learning and simulation ability in the application of prospecting prediction. Deep learning algorithms can effectively analyze and predict data, which have great significance in improving the efficiency and accuracy of mineral exploration. However, there are not many specific examples of their application in mineral exploration prediction, and researchers have not yet conducted a comprehensive discussion on the advantages, disadvantages, and accuracy of deep learning algorithms in mineral prospectivity mapping applications. This paper reviews and discusses the application of deep learning in prospecting prediction, highlighting the challenges faced by deep learning in the application of prospecting prediction in data preprocessing, data enhancement, system parameter adjustment, and accuracy evaluation, and puts forward specific suggestions for research in these aspects. The purpose of this paper is to provide a reference for the application of deep learning to researchers and practitioners in the field of prospecting prediction.

The Taochong iron deposit is one of the important skarn deposits in the Middle–Lower Yangtze River metallogenic belt, Eastern China. There are two types of ores in the deposit: skarn- and quartz–calcite-type ores. The skarn-type ore, which is composed of hematite (Hm-1), garnet, pyroxene, actinolite, chlorite, quartz (Q-1), and calcite (Cal-1), is crosscut locally by a quartz–calcite-type ore vein. The quartz–calcite-type ore consists mainly of hematite (Hm-2), magnetite, quartz (Q-2 and 3), and calcite (Cal-2). The δ¹⁸O_water value (~2.67‰) of the fluids in equilibrium with Hm-1 is similar to the values of the mixtures of magmatic and meteoric fluids. However, the δ¹⁸O values of the fluids in equilibrium with Hm-2 are in the range of 7.64–8.54‰, similar to those of magmatic fluids. The δ¹⁸O values decrease systematically from the fluids in equilibrium with Hm-2 (7.64‰ to 8.54‰) to the fluids in equilibrium with magnetite, Q-3, and Cal-2 (−0.12‰ to 4.17‰) and the fluids in equilibrium with Cal-3 (−2.17‰ to 0.36‰). These features of oxygen isotopes indicate that two episodes of hydrothermal activity took place in the Taochong deposit, and both episodes began with a magmatic origin and then progressively evolved by mixing with meteoric water. The results of quantitative simulations suggest that the deposition of the skarn-type ores was most likely caused by the mixing of magmatic and meteoric fluids, whilst the deposition of the quartz–calcite-type ores was most likely caused by the boiling of magmatic fluids and the mixtures of magmatic brine and meteoric water.

The mechanisms of iron enrichment in skarn iron deposits have been widely discussed; however, key factors responsible for genesis of high-grade skarn iron deposits still remain unclear. The Litun skarn iron deposit in the Luxi district, eastern China, has an average Fe grade of 58.2 wt% and provides an excellent opportunity to study the formation of high-grade iron ores. The Litun deposit is characterized by three generations (Mag-1, -2, and -3) of magnetite, which exhibit a clear cross-cutting relationship and exhibit a well-defined zoned texture. X-ray mapping and laser ablation–inductively coupled plasma–mass spectrometry analyses show that magnetites of different generations have significant compositional differences. The Fe contents show the greatest quantitative variability among the different magnetite generations (Mag-1, average [avg.] 71.33 wt%; Mag-2a, avg. 67.37 wt%; Mag-2b, avg. 69.54 wt%; Mag-3, avg. 68.10 wt%). The Si contents show the greatest degree of variability among the different magnetite generations, varying by an order of magnitude (Mag-1, avg. 3155 ppm; Mag-2a, avg. 13,817 ppm; Mag-2b, avg. 7891 ppm; Mag-3, avg. 10,748 ppm). In this study, to constrain the controlling factors for the formation of high-grade iron ores, a chemical mass transfer model has been employed to simulate the ore-forming processes. The results of simulation modeling in a low-pressure environment (500 bar) show that the solubility of iron decreases rapidly from 20,148 ppm to 3 ppm when the temperature of a hydrothermal fluid decreases from 500 to 350 °C, but that the solubility of silicon decreases only slightly from 945 ppm to 572 ppm. In contrast, in a high-pressure environment (1000 bar), the solubility of iron (from 735 ppm to 0.7 ppm) and silicon (from 1657 ppm to 646 ppm) has a similar decreasing rate when the temperature of the fluid decreases from 500 °C to 350 °C. These data imply that a Fe-rich hydrothermal system is more likely to form magnetite without/with minor silicate minerals (the main mineral of the high-grade ores) in a low-pressure environment. It is well known that a transition from lithostatic conditions to hydrostatic conditions commonly occurs during the mineralization process in high-grade skarn iron deposits. Therefore, the rapid saturation of iron in a low-pressure environment relative to other components such as silicon may be the main reason for the formation of high-grade iron ores in the Litun deposit, which may also be one of the important mechanisms for the formation of high-grade skarn iron deposits worldwide.

The North China Craton hosts abundant skarn iron resources, yet the regional large-scale mineralization mechanism remains incompletely understood. The Litun deposit is a newly discovered skarn iron deposit in the North China Craton. Integrated field geological investigations, petrographic observations, fluid inclusion microthermometry and stable isotope geochemistry are applied to constrain evaporite contributions to metallogenic processes. Four mineralization stages are identified: skarn, oxide, sulfide, and carbonate. Early skarn-stage fluids are iron-rich magmatic hydrothermal fluids with high temperatures (498 to >550 °C), high salinities (18.6 to 59.4 wt% NaCl eqv.), and magmatic δ¹⁸O values of 8.3 to 10.8‰. Subsequent oxide to late carbonate stages record continuous infiltration of meteoric water, supported by H–O isotopic trends of rising meteoric water proportions. Pyrite from the magnetite ores has δ³⁴S_V-CDT values between 12.0 and 15.0‰, significantly higher than those of pyrite in the Litun diorite (−0.8 to 1.1‰), indicating the contributions of sulfur from evaporites (δ³⁴S_V-CDT 26.9 to 28.6‰) in the mineralization process. Moreover, vein pyrite formed in later stages displays even higher δ³⁴S values (17.3 to 20.9‰), demonstrating progressive enrichment of evaporite-derived sulfur as hydrothermal activity evolves. Synchronous rises in meteoric water fraction and evaporite sulfur proportion indicate evaporites are delivered into the ore-forming system via meteoric water mixing. The mixing of meteoric water containing dissolved evaporites and iron-rich magmatic-hydrothermal fluids may be the major mechanism of magnetite precipitation in the Litun deposit.

Figure S1: Field photographs, photomicrographs, and backscattered electron images showing textural features of the magnetite in the Litun deposit. Table S1: LA-ICP-MS trace element compositions of magnetite from the Litun skarn iron deposit. Table S2: Model results for Fe solubility in hydrothermal fluids. Table S3: Model results for Si solubility in hydrothermal fluids.