Harnessing the power of forward models: past, present and future

Variational retrievals, data assimilation, and model evaluation in observation space, all rely on accurate instrument simulators, or forward models. The scientific challenge is to find innovative approximations to the radiative transfer that make them fast enough to use iteratively, while retaining accuracy. In this presentation I will summarize how the development of various forward models, particularly in the context of synergistic retrievals from EarthCARE and the A-Train, has the capability to reveal important cloud and precipitation properties that would otherwise remain hidden, and potentially even to develop new satellite concepts. For example, our radar and lidar “Multiscatter” model enables the extinction profile to be retrieved in ice and liquid clouds even in the presence of lidar multiple scattering.  Our “FLOTSAM” solar radiance model can work with profiles containing arbitrary combinations of particles, and surprisingly can help improve rain-rate retrievals by better providing the additional information needed to partition the radar path-integrated attenuation into the contributions from liquid clouds and rain. The Two-Stream Source Function (TSSF) infrared and microwave radiance model enables us to interpret 94-GHz brightness temperature, which provides important additional information on precipitating ice and liquid clouds.  I will end by presenting a new radiance model for cloud/storm-resolving models that can efficiently represent horizontal radiation transport between columns; this could enable future retrievals and assimilation to take full account of 3D radiative effects.