Abstract: Excessive copper concentration in the soil will cause certain harm to human body and the environment, so it is of great significance to explore the inversion of heavy metal copper content. In this study, 43 soil samples in the typical iron tailings area of Tangshan were taken as examples, and the reflectance spectrum and content information of copper in the soil were measured at the same time. After a variety of spectral transformations, the correlation analysis method (CA) and the continuous projection method (SPA) were carried out. The characteristic wavebands of soil copper content were selected, and then the inversion model of copper content was established using multiple linear regression (MLR) and partial least square regression (PLSR) algorithms, and the inversion results of various spectral data were obtained. The results show that the spectrum data after the second-order differential processing has the best inversion effect. Among the two inversion models of CA-PLSR and SPA-MLR, the inversion accuracy of SPA-MLR is relatively accurate; after the second-order differential spectrum transformation te SPA-MLR model has more advantages in estimating soil copper content.