VIC Model Setup Clarification and Grid-Based Forcing Input Strategy

[snur24]

Dear VIC User Community,

I am a PhD student conducting comparative hydrological modeling. Previously, I developed models using SWAT and HEC-HMS for a small sub-basin. These models were constructed based on observational data from a single meteorological station and one streamflow gauge, supported by DEM data and watershed delineation tools.

Currently, I am trying to evaluate the performance of the VIC (Variable Infiltration Capacity) model over the same basin to compare with the previous results. For this purpose, I have installed VIC on a Windows system using a secondary Linux environment (Ubuntu via WSL). I have successfully compiled and executed the VIC model using the Stehekin sample data with the following steps:

bash
Kopyala
Düzenle
sudo apt-get update
sudo apt-get install build-essential
sudo apt-get install csh
git clone https://github.com/UW-Hydro/VIC.git
cd VIC/vic/drivers/classic
make
git clone https://github.com/UW-Hydro/VIC_sample_data.git
After modifying the global parameter file (Stehekin/global_param.STEHE.txt), I updated paths to point to the necessary files:

FORCING1: .../forcings/full_data_

SOIL: .../parameters/Stehekin_soil.txt

VEGPARAM: .../parameters/Stehekin_vegparam.txt

VEGLIB: .../parameters/Stehekin_veglib.txt

SNOW_BAND: .../parameters/Stehekin_snowbands.txt

RESULT_DIR: .../sample_classic/

I executed the model using:

bash
Kopyala
Düzenle
./vic_classic.exe -g path/to/global_param.txt
This ran without errors.

My first question is whether this workflow aligns with best practices for VIC setup.

My second question concerns the representation of spatial data in VIC:
Unlike SWAT and HEC-HMS, where I could define the full watershed using DEM and stream network data, the VIC model requires gridded meteorological forcings. My current setup uses data from a single meteorological station, and I am unsure how best to represent the entire basin. My advisor suggested applying the observed values only at the grid cell containing the station and assigning NoData or null values to other grid cells. Meanwhile, for static spatial inputs like soil and vegetation, it was advised to keep full coverage for all relevant grid cells.

Is this a scientifically valid approach for VIC modeling in a sub-basin with sparse meteorological data? Or would it be better to interpolate the station data across the basin (e.g., using IDW) to ensure full grid coverage?

Your guidance would be greatly appreciated.

Kind regards,

[jschap1]

Hi snur24,

Your workflow sounds fine to me.

Typically, you would run the VIC model with distributed meteorological forcing data. However, if you want to use data from a single meteorological station, one option might be to treat the entire basin as one grid cell (i.e. a "lumped" model). You can specify different land covers within the grid cell (but not their locations) via the vegetation parameter file and set up the soil parameter file to use similar inputs as you used for SWAT and HEC-HMS. I think this should be similar to what your adviser suggested.

I cannot speak to the scientific validity of one approach over another - perhaps it could be helpful to consider why you are you interested in comparing the VIC results with the outputs of SWAT and HEC-HMS?