Channels and Routing

[ktsuruta]

I have a large basin to model that I have partitioned into ten sub-basins. To reduce computational time, I would like to simulate each sub-basin separately and in parallel. In accordance with the suggestions regarding BrianHydro’s issue, I then intend to run a final simulation where the sub-basin to sub-basin connectivity is described and each sub-basin’s “rainfall” is based on its previous simulation’s hydrograph.

As a toy case, I tried to drive a single sub-basin comprised of one hru with precipitation equal to its observed hydrograph (converted into mm/day). Using ROUTE_DUMP as the catchment routing method, I expected the simulated hydrograph to exactly match observations. However, I have slightly mismatched hydrographs. It appears at each time step, half of the precipitation arrives at the outlet while the other half goes to rivulet storage and arrives at the outlet in the next time step.

Is there a way for me to fix this without pre-processing the “rainfall” hydrograph to account for rivulet storage? Is there a better way to simulate sub-basins in parallel?

Many Thanks,
Kai

[jrcraig]

good question - the short answer is to switch to :SnapshotHydrograph for output - I believe this is an output issue only in that, by default, Raven reports integrated flow over the timestep, but the state variable is instantaneous flow Q at the start/end of each timestep.

A pulse of 1 cm of runoff over a single time step, for instance, will appear as a single corresponding peak in flow e.g., 0 m3/s to x m3/s back to 0 m3/s in the internal instantaneous flow time series when ROUTE_DUMP is used - the peak of this triangular hydrograph (of length 2 Delta t) will occur at the end of the timestep during which the runoff is applied. Here, x is the scaled peak flow corresponding to the amount of runoff and basin size. However, this triangular hydrograph will be redistributed to the two time steps over which it occurs when integrated flows are reported, leading to the 'symptom' you speak of.

By default, Raven reports integrated flows over the time step to be consistent with most stream gauges which report (you guessed it!) integrated measured flows over the time step.

[Ameer]

Hello,
I am using HBV_EC model setup. I do not have channel profile data. Can i just delete the channel profile section from the .rvp file? If i do, what impact it may have on the hydrology (in general) of the basin (routing)?
Thanks,
Ameer

[rchlumsk]

Hi Ameer

The official HBV_EC setup has the Routing argument as ROUTE_NONE, implying that there is no in-channel routing performed (which is done between basins). For a single catchment with multiple HRUs, this makes sense and the channel profile is not needed.

For more complex setups with multiple basins the in-channel routing would impact the peak and timing of flows, as the hydrograph would be attenuated (stretched out, reduced peak) based on the routing method selected. The impact of this really depends on how many catchments are modelled and how large the area is. As mentioned, the HBV_EC model does not perform this in-channel routing, only catchment routing (based on the HBV_EC template file in the Raven manual, page 169).

Cheers,
Rob