Objective The effectiveness of geological prospecting depends on the accuracy of prospecting target prediction. This paper introduces the application of big data concepts and methodologies to deeply explore the correlation between regional geochemical information and known gold deposits, aiming to achieve precise prediction of regional gold prospecting targets.

Methods Taking the quantitative prediction of Au prospecting targets using big data methods in the West Qinling region of Gansu Province as an example, this study emphasizes the importance of deep processing of raw data to eliminate various errors. It highlights the modeling approach of classifying geochemical information at different scales to construct a series of prediction models and integrating the prediction results.

Results Through self−developed geochemical data processing software (including modules for determining anomaly recognition thresholds and sub−region adjustment), systematic errors in the raw data were effectively eliminated, addressing the issue of missing low−contrast anomalies. This laid a high−quality data foundation for prospecting prediction. The innovative "classified modeling−integrated prediction results" process was proposed, wherein target prediction models were constructed separately for geophysical, geochemical, and geological information at different scales, and the results were integrated. This approach mitigated the impact of differences in information accuracy and representativeness, thereby significantly enhancing the necessity and feasibility of improving the accuracy of prospecting target prediction.

Conclusions This method significantly reduced the prospecting target area to only 3.38% of the total study area, with 50% of these targets being mineralized. The probability of discovering mineralization in the predicted target areas reached 36.8%, greatly improving prediction efficiency and reliability. The research methodology exhibits strong universality and can be extended to regional geophysical and geological information prediction fields, providing a new paradigm for intelligent prospecting target prediction and effectively enhancing prediction accuracy.