Taking advantage of the multidimensional unified and simplicity expression of movement characteristics of geometric algebra

an adaptive template matching method for convergence and divergence structure of multi-dimensional vector fields was proposed.The optimal rotor between the original vector field and the standard template is established based on SVD (Singular Value Decomposition).The data adaptive divergence-convergence template generation method is then constructed based on the structure consistency of rotor rotation

and the classification of geometric structure of the vector field based on the rotor rotation angle is proposed.Finally

the adaptive template matching method is constructed based on the geometric convolution.These methods are verified with the wind field of North America.The results suggest that our method can effectively resolve the structural features of the vector field with different dimensions and can do structure-based classification of vector fields.