From Radio Detection and Ranging (RADAR) to Meso-Analysis System (MAS)

Since the introduction of weather radar more than a half century ago, the processing of radar observations of the atmosphere has undergone two stages: first, we used a qualitative and subjective interpretation of visual observations on a cathode ray tube displaying in real time returns from precipitation. With the introduction of computers in the late 70s an effort was made to match the skill of subjective visual interpretation by the radar operators using an objective Radar Data Processor (RDP). The algorithms for this are based on simple ideas derived from conceptual models of storm structure and storm circulation and occasionally making use of AI. For example, to resolve the inherent ambiguity of observing only one component of the 3D air velocity, radar wind profiling is based on the assumption of linear dependence with space of the components of the horizontal air velocity; a Rankine vortex is the geometrical model for tornado or mesocyclone detection; and so on.

In the 90s data assimilation was recognized as an opportunity to resolve the limitations and ambiguities of radar surveillance of a limited number of atmospheric variables. This path, proposed some twenty years ago, was slow in producing results with operational potential. At McGill, we have initiated the development in this direction following a methodology that appeared particularly adequate for operational implementation: assimilation of radar data with numerical model used as a weak constraint. After several stages of exploration we have completed a robust assimilation system that is now implemented in real time. It incorporates three volumetric scans of radar data, the Canadian GEM-LAM forecasts at 2.5 km horizontal resolution as a background term and some preliminary estimates of the error structure of all the components of the assimilation system.

Our first objective is to produce an analysis of the state of the atmosphere so that the traditional radar data processor is replaced by a Meso-Analysis System (MAS) from which all "radar" products can be derived. MAS also has the advantage of naturally being able to utilize information from other sensors, as well as provide a physical basis and additional background information for much of the traditional radar data corrections such as extrapolation of precipitation to the ground, dealiasing, etc. In this presentation, we will outline the concept behind MAS and show some preliminary results