Abnormal analysis of geochemical elements is a significant link in the mineral resources prediction. Element anomaly often directly or indirectly reflects the existence of mineralization anomaly. Due to the polyphyleticism of the metallogenic material and ore-forming process is characterized by multiple stages. there are two type of the metallogenic elements ,i.e., main ore-forming elements and associated elements. Therefore, it presents different assemblage characteristics in the spatial distribution, which also exist in the abnormal distribution. The spatial distribution mechanism of the combination elements are relate to geological conditions and the property of the elements. Based on the geochemical and transport theory, we explore the space quantitative anomaly seeking model. Then we proposed a new method - source tracing analysis for the inference of ore-caused anomaly from the combination anomaly, and the criteria of the proposed method is given. It can realize the intelligent search of the middle-large scale mineral resources prediction target effectively.