As a kind of compositional data, the spatial interpolation of soil mechanical composition needs to meet four constraints of constant sum, nonnegativity, linear unbiased estimation and error minimization. Therefore, it usually needs to be transformed before spatial interpolation. A novel transformation method-spherical coordinate transform method was proposed and the study area was located in Yinbei Irrigation District, Ningxia Hui Autonomous Region. Firstly, the sampled soil mechanical composition data was transformed through three methods, including symmetry logratio transform method, asymmetry logratio transform method and the proposed spherical coordinate transform method, respectively. Then the soil mechanical composition spatial interpolation was realized with the application of ordinary Kriging. Secondly, the soil mechanical composition interpolation maps of different transform methods were compared to analyze the scientificity and feasibility of the proposed transform approach. The precision and fitting effect were assessed by utilizing the mean absolute error (MAE), root mean square error (RMSE), index of agreement (d) and Pearson’s correlation (R). The results indicated that spherical coordinate transform technology not only can satisfy the four constraint conditions but also had superior interpolation accuracy for sand and silt content than the other two logratio transform methods. However, the logratio transform methods had more accurate interpolation results than spherical coordinate transform method for clay content prediction. To summarize, a novel transform technology with the advantages of easy understanding, small calculation amount, and good interpolation precision and without consideration of zero values for spatial interpolation of soil mechanical composition was presented.