The extraction of remote sensing information from HF radar sea clutter has been practiced for over forty years, but the nature of radar products delivered to users has changed relatively little over the years, especially for skywave radar. One might argue that this is a consequence of four factors: (i) low priority attached to phenomena which occur only sporadically or are restricted to particular locations, (ii) a belief that the coarse spatial resolution of HF skywave radar restricts its application to large-scale, slowly-varying phenomena, (iii) recognition that disturbed propagation conditions at HF frequently obscure even the most readily extracted information, so aiming for more ambitious measurement objectives may be futile, and (iv) lack of suitable geophysical models of localized or transient phenomena with which to inform radar echo interpretation. In this paper we refute the proposition that HF skywave radars are intrinsically incapable of detecting and characterizing such phenomena. We argue that by using geophysical models of spatio-temporal forcing of ocean surface geometry and dynamics by stimuli which are localized in space or time (or both), it is possible to derive corresponding signature models. Adopting mesoscale convective cells as an example, we illustrate the procedure and, within limits, validate the predictions by comparisons with radar measurements.