Combining separate Y and UV cudaArray_t planes into one cudaArray_t of type NV12

I am using NV12 block linear memory through NvBuf API on a Jetson NX device. To get a cudaArray_t handle I am using the EGL interop API. This gives me two separate planes: Y and UV planes.
However, I would like to use the mipmap functionality on the combined NV12 and I would like to know if it is possible to combine these cudaArray_t somehow without copying.