Summary form only given. This paper re-examines the assumptions used in the Braginskii arc expansion model when it is applied, as by Martin, to water breakdown channels. In particular the electrical conductivity of water in the channel is allowed to vary with temperature and a more realistic radiation estimate is adopted. These relations are used in a zero-dimensional lumped model of the channel. The resulting channel temperatures are higher than those typically used in gas breakdown studies, and because of the hydrogen component of water, the resulting growth of conductivity with temperature is correspondingly more pronounced. These results are compared to one-dimensional simulations using the ALEGRA-HEDP code. A simplified form of the zero dimensional model, which is easily implemented as a nonlinear circuit element, is also briefly described. Finally, experiments done in the Sandia National Laboratories Z20 test stand are discussed and compared to the model results