Data assimilation with the Canadian middle atmosphere model

Abstract A data assimilation scheme has been coupled to the Canadian Middle Atmosphere Model, providing, for the first time, the capability of assimilating data from the ground to the top of the mesosphere (about 95 km). This model is a full general circulation model with on‐line fully interactive chemistry involving 127 gas‐phase and heterogeneous reactions. Thus, feedback between dynamics, chemistry and radiation occurs in every model time step. In this work, validation of the system for tropospheric and lower stratospheric analyses is undertaken with the standard observation set used in operational weather forecasting. Results are found to agree reasonably well with radiosonde observations and with Met Office (UK) analyses. Although ozone is not assimilated, ozone fields match total column observations well in terms of synoptic patterns. However, due to model biases, total column values are too large at mid‐latitudes and too small in the tropics. Since the assimilation scheme was designed for tropospheric weather prediction, its application to a middle atmosphere model can help to identify the challenges of assimilating data from this region of the atmosphere.