Right now, when dt_surface > dt_atmos for some surface, the atmosphere computes precipitation multiple times per timestep of the surface. However only the precipitation flux computed just before the surface's step actually gets used by the surface model in its boundary condition.
To fix this, we should accumulate precipitation like we do for turbulent fluxes when dt_surface > dt_atmos. Precipitation is computed at each stage within an atmosphere timestep, so we'll also need to include flux accumulation within the atmosphere's timestep for this to actually be accurate. Because of this extra complexity, this has been left as a separate task from the turbulent flux accumulation in #1945.
Right now, when
dt_surface > dt_atmosfor some surface, the atmosphere computes precipitation multiple times per timestep of the surface. However only the precipitation flux computed just before the surface's step actually gets used by the surface model in its boundary condition.To fix this, we should accumulate precipitation like we do for turbulent fluxes when
dt_surface > dt_atmos. Precipitation is computed at each stage within an atmosphere timestep, so we'll also need to include flux accumulation within the atmosphere's timestep for this to actually be accurate. Because of this extra complexity, this has been left as a separate task from the turbulent flux accumulation in #1945.