Optical Flow for Verification

Modern Numerical Weather Prediction (NWP) models produce forecasts which are effectively gridded spatial fields. Digital images can also be viewed as gridded spatial fields, and as such, techniques from image analysis can be employed to address the problem of verification of NWP forecasts. One relatively successful technique for modeling how images change temporally is called optical flow, where it is assumed that temporal changes in images (e.g, in a video) can be represented as a fluid flowing in a conserved manner. Multiple realizations of the general idea have already been employed in verification problems as well as in data assimilation. Here, a specific formulation of optical flow, called Lucas-Kanade, is reviewed and examined as a tool for estimating forecast distortion error (i.e., size, displacement and intensity error combined). The Lucas-Kanade formulation is not the most general, but it does have the virtue of simplicity and transparency. The result of mapping an observed field to a forecast field in this manner is a field of vectors, visually displaying the map itself. Although this field has great diagnostic power, it is shown that the joint distribution of the magnitude and angle of the vectors provides a useful summary of the optical flow field. The utility of the method is assessed for performing automatic, objective, and spatially scale-sensitive verification.