On the stochastic Richards equation