The Nuocang Cu polymetallic district is located in western Gangdese, close to the south of the Luobadui–Milashan fault zone. A large number of metal deposits with the potential to be prospected, such as Chagele, Zhalong, Longgen, and Sangmola, have been found near this district. To further isolate the target prospecting area, we used a portable X-ray fluorescence spectrometer (pXRF) to conduct an in situ 1:10,000 soil pedogeochemical survey. The results show that the use of a pXRF combined with the geological connotation method (GCM) can accurately delineate the anomalies related to mineralization and highlight “weak” and “small” anomalies. It was also shown to effectively shorten the working cycle and ensure the continuity and timeliness of field work. Through sizing tests, the analysis of −10~+60 mesh soil samples achieved the best anomaly delineation effect. By studying the supra-ore, near-ore, sub-ore halo, and Th/U, the degree of denudation and the oxidation-reduction environment of the deposit were judged to be moderate. Ultimately, depending on the target area delineated by the pXRF, six Cu-Pb-Zn orebodies were uncapped by five exploratory trenches, which proved the potentiality of the Cu-Pb-Zn polymetallic epithermal deposit controlled by minor faults in Central Nuocang.

Porphyry systems may host several deposit types. These deposits commonly differ in element associations, host rocks, structures, and alteration features, which increases the difficulty of mineral prospectivity mapping (MPM) in porphyry systems. This study proposes a geological factor-enhanced variational autoencoder (VAE) method, which can be used for large-scale mineral prospectivity mapping in porphyry systems. First, element geochemical data were transformed into dimensionless anomaly coefficients. These anomaly coefficients were then used as VAE inputs. The VAE was used to calculate reconstruction errors. Geological factors, including intrusions, structures, lithological units, and alteration, were transformed into distance-decay weights. These geological weights were used as soft-gating terms to enhance the VAE reconstruction error. Model testing was conducted in the Zhunuo ore district in the western Gangdese belt. Datasets include 14,475 stream-sediment samples, geological map data, and remote sensing information. This method conducts separate MPM for porphyry, skarn, carbonate-replacement, epithermal, and hydrothermal vein-type deposits. Results show that high-prospectivity areas effectively identified known deposits and ore occurrences. Zhunuo–Beimulang is the main target area for porphyry mineralization. Northern Cimabanshuo is a key area for the next stage of exploration for porphyry-type mineralization. Tangge and western Luobuzhen are favorable for skarn and carbonate-replacement mineralization. Luobuzhen–Yalongri–Hongshan is favorable for epithermal and hydrothermal vein-type mineralization. ROC-AUC results further confirm the model performance. For porphyry-type deposits, the geological factor-enhanced VAE increased the AUC from 0.912 to 0.959. AUC values for skarn, carbonate-replacement, and epithermal deposits were 0.855, 0.966, and 0.968. These results indicate that the method improves MPM performance. This method can be regarded as a new approach for integrated anomaly identification, with the ability to incorporate geological factors.