@CloLanglais - could we clarify/check what you discussing? I suspect that the information is already provided - and it is our intention to ensure that equivalent fields are provided as part of CM3/ESM3 - but I may have misunderstood.
Both ACCESS-ESM1.5 and ACCESS-CM2 provide gridded runoff information via CMOR variables mrro (total = surface+deep drainage) and mrros (surface runoff). These are available at different time resolutions (monthly or daily averages) for most (all?) simulations under directories Lmon or day under the respective sections on project fs38 on the NCI.
River outflow into the ocean is also available from both models as CMOR variable friver at monthly time resolutions (directory Omon). However note that friver is provided aligned to the river model grid (1x1 degree) and this differs to the UM/land grid.
Additionally - notes of caution - i) in order to better satisfy the technical requirements of global conservation of water, river outflow in CM2 is scaled globally to balance net water added/lost elsewhere in the terrestrial system (evaporation from lakes, inland water basins). This is typically a small effect - but conceptually it does imply that runoff integrated over a catchment may not equal river outflow. ii) water conservation/runoff from the ice regions is complicated by the iceberg schemes - these differ between ESM1.5 and CM2.