Snow-cover models are used in many applications in today's snow and ice research. Descriptions of changes in size and shape are a major problem in modelling the snow cover. Empirical models for kinetic growth under temperature gradients have been developed, as well as more complicated models based upon microstructure. In this work a simple, physically based model is derived which depends on one adjustable geometric factor only. Snow texture is described as a body-centred cubic lattice containing source and sink grains. The latter grow as plates due to water-vapour transport in the layer as well as between the layers. The model was implemented in a research version of the one-dimensional snow-cover model SNOWPACK. Model outputs are compared to experiments done in the cold laboratory where sieved snow is subjected to temperature gradients. Disaggregated snow samples are analyzed by digital image processing, by sieving and by visual characterization. In order to determine grain-size as objectively as possible, these various methods are evaluated for compatibility. The new model simulates very well kinetic grain growth for densities of 100-200 kg m-3 and temperature gradients up to -200 K m-1. The model will be incorporated in the operational version of SNOWPACK.