Streamflow Underestimation in Coupled WRF-Hydro Simulation

[PREETI JAKHWAL]

Dear WRF-Hydro Users Group,

I am running a coupled simulation using WRF-Hydro with WRF in online mode and  observing significantly low streamflow values in the output. The model executes successfully, but the simulated discharge is on the order of 1–2 m³/s, whereas expected values for the basin are approximately 100–200 m³/s.

The current setup includes: Simulation period: August 6 to August 27, Domain: d03, 1 km land surface resolution (WRF),100 m routing resolution (DXRT = 100 m, AGGFACTRT = 10),Forcing from NCEP FNL (0.25°), Channel routing using diffusive wave option (channel_option = 3), A drainage network consisting of ~1000 channel cells, Standard Noah-MP land surface configuration, Streamflow is extracted from CHRTOUT outputs (gridded channel flow).

I would appreciate any insights into potential causes of such low streamflow estimates, particularly with respect to routing configuration, scale mismatch between atmospheric forcing and hydrological resolution, or channel network representation.

Thank you for your time and support.

Best regards,
Preeti Jakhwal

[aubrey]

Hi Preeti:

A few quick checks:

1) Does your precipitation from WRF for the event look like the right general amount and in the right general area?

2) Are you spinning up the hydrologic components? If you are not running model spinup before the event, the rainfall event may be falling on dry soils, empty groundwater buckets, or empty channels, so may be lagged before generating streamflow.

3) Is your channel network connected and are you modeling the full upstream drainage area within your hydro model domain?

4) Are you extracting streamflow at the correct mainstem location (vs. a side channel)?

5) Are you running with one of the baseflow options (pass-through or exponential bucket)? If your GW baseflow is off, that water will not be added to the channel and will result in lower flows.

You might also try running your standalone WRF-Hydro setup (uncoupled to WRF) with the idealized forcing option (FORC_TYP = 4 in the namelist.hrldas) to make sure it is showing expected runoff response to a prescribed precipitation event. That will give you a good indication that at least the routing is setup properly before you add the additional complexity of WRF.

Let us know how it goes.

Thanks!

Aubrey