As awareness of the meteotsunami (MT) (tsunami-like wave of meteorological origin) threat to human life and coastal infrastructure is growing (e.g., >100/year MT events just on the Great Lakes), so is the need for understanding and forecasting it. However, knowledge acquired in the study of seismic tsunami is not readily transferable to MT research, because MT evolution is in important aspects qualitatively different: it is much more nonlinear and more strongly affected by bottom friction. Moreover, a realistic reconstruction of MT evolution is almost impossible because of the current poor spatial and temporal resolution MT observations, overwhelmingly confined to the shoreline. Therefore, the picture of the nonlinear transformation of a MT from generation to its shoreline manifestation is substantially incomplete. Since the MTs tend to disintegrate into very short (down to ~10s) pulses, even modern tidal gauges (1 min resolution) fail to capture essential features of its evolution. This fundamental knowledge gap is evident in our recently-published high-resolution MT observations off the Louisiana coast, which show a highly-complex, multi-scale non-linear disintegration of the MT wave. Together with other gaps identified in the literature, the observations strongly suggest revising of the current MT paradigm. Field experiment: A 3-year field experiment will be conducted to collect unprecedented high-resolution (up to 4 Hz) MT observations off the Louisiana coast, to capture fully the details of MT evolution. The site is a unique "natural laboratory": (i) it experiences 20-30 frontal passages per cold season (high MT probability); (ii) allows for the study of bottom-induced dissipation over a range of sedimentary types (coarse sands to mud); (iii) crucially, it is already monitored by WAVCIS (http://www.wavcis.lsu.edu/, Coastal Studies Institute, LSU), an ocean-observing system network sponsored by NOAA and BOEM. WAVCIS will be enhanced with high-resolution capabilities and five new stations located both on the coast and offshore. The observations will be organized into a public database. Modelling: The proposed work is based on the hypothesis that in a generic Proudman-resonated MT event trapped waves are also excited. Although usually less dangerous, they might represent an important part of the process, and behave as precursors to the main wave. The work focuses on the propagation stage of MT: a theoretical description of the nonlinear evolution of both free and trapped components of MT will be developed, including solitary waves, and will be validated for the real topography, and accounting for realistic bottom friction. MT warning: Data analysis and model development will be integrated into an evaluate-learn-correct cycle that will improve our representation of the process and our ability to anticipate MT events. Effective numerical models and early-warning strategies will be developed and tested. Presently, MT early-warning systems mostly rely on interpreting atmospheric data. The new high-resolution observations will be used to incorporate in this cycle elements of real-time ocean response.