A physical retrieval system is developed for all sky single field-of-view (FOV) hyperspectral sounding. This system can be used to retrieve atmospheric temperature, moisture, and trace gas profiles, along with cloud height and cloud microphysical properties simultaneously from single FOV radiances measured by hyperspectral sounders. Single FOV retrieval results allow the users to extract horizontal gradient information with a spatial resolution defined by the size of one FOV. Moreover, the system finds solutions by directly fitting observed spectral radiances and therefore can be used to obtain radiative kernels that fulfill the `radiance closure' needed for climate applications. As a comparison, current operational retrieval algorithms find solutions by fitting the `cloud cleared' radiances generated from several adjacent FOVs. Not only the spatial resolution of those results is several times coarser than that defined by a single FOV, but the results cannot be used to build radiometric consistent radiative kernels under cloudy sky conditions. In this paper, some applications of the system are demonstrated to highlight the benefit of the full spatial resolution retrieval and the capability of building radiometric consistent radiative kernels under all sky conditions.