The Meteoroligal Research Division of Environment Canada and the Meteorological Service of Canada have developed jointly an analysis of 6h precipitation accumulation for use in land-surface and hydrological modelling over Canada. The approach taken consists in combining a numerical weather prediction of precipitation at that scale with ground observations of precipitation accumulation from synoptic and coop stations accross Canada. The density of the ground precipitation network being relatively low in most of Canada has motivated us to improve upon this technique by adding ground observations of precipitation type from synoptic stations, observations of clear sky from GOES imagery, lightning strikes observations as well as ground radar observations where available. In this paper, we describe the data fusion process that we used, and show how the product can be used and what impact it can have on problems ranging from hydrological prediction (both in hindcast mode and for ensemble streamflow forecasting) to land-data assimilation.