wflow
A short description and selected case studies
Hydrological model wflow
• Fully distributed
• Can run with limited data input
• Two source maps needed:
• DEM
• Land-use
• Preparation script that performs the landscape analysis (catchment
delineation etc.)
• All model parameters linked to land-use
• Can perform state updating in real-time applications
• Based upon work for cqflow VU in 2005 and 2006, now adapted for
phd project Hylke Beck
• Written in python and pcraster
• Flexible and open
Running fine but not a polished software product
Doc in being worked upon
Why python/pcraster?
• Can use (parts of) exiting pcraster modules
• Can use python for generic programming (command-line options,
reading XML, updating, debugger, IDE etc)
• Free! (pcraster not open soruce but can be downloaded free of
charge)
• Fast! If you can avoid loops and perform all operation on
vectors/matrices
• Other operations can also be done in python:
• Retrieving ftp data, link to openDAP
• Data copying and cleaning
• Logging
• Plotting and analysis (similar to Matlab)
• etc.
Why another model?
• A balance between conceptual and physical representation of the
catchment is needed in many applications
• New data sets (DEM, RS data etc) cannot be used by many existing
models
• This model maximizes the use of available (spatial) data
• Can be used in data rich and data sparse environments
Ok What can it do?
• Simulations of water level and
discharge (for simulations or
operational purposes)
• Investigate the effect of a changing
environment (climate, land used, e.g.
urbanisation)
• Can work on different catchment
sizes
• All variables are distributed in
space
• Can start simple and expand
later on
Bandung
Rhine
1: Terrain analysis
1. Optional cutout part of DEM
2. Set outlet at lowest gauge and
extra points (for later output) at
other gauges using gauge
coordinates
3. Determine river network (can use
existing to burn-in if needed)
4. Determine LDD and sanitized
DEM
5. Resample land-use map to DEM
1: Terrain analysis
wflow_catchment.map
wflow_dem.map
wflow_gauges.map
wflow_landuse.map
wflow_ldd.map
wflow_river.map
wflow_streamorder.map
wflow_subcatch.map
2: Model parameters
• All parameters are linked to landuse/soil types via so called lookup
tables
• Links parameters to land-use map
and/or soil map
• Calibration/Verification step
The processes: Interception
• Rainfall interception via Gash
model → daily timesteps
The processes: The soil
• Soil accounting scheme
based on TOPOG_SBM
(Vertessy and Elsenbeer
1999)
• Schematic representation of
the hydrologic processes
modeled by Topog_SBM.
Symbol definitions:
• rf, rainfall; in, infiltration;
st, transfer between
unsaturated and saturated
zone; ie,
• infiltration excess; se,
saturation excess; ex,
exfiltration; of, overland
flow; and sf, subsurface
flow.
The processes: The soil
•Inputs to the model:
• Et + Es from the canopymodel
•Total throughfall + stemflow from the
canopy model
•Determines:
•In- exfiltration
•Lateral saturated flow
•Transfer between saturated and
unsaturated store
•Reduces Et + Es to an 'actual
evaporation' if water stress occurs
(takes rooting depth into account).
•Surface runoff via kinematic wave
The processes: The soil
•
•
•
Ksat decreases exponentially in
depth (M parameter)
Transfer between unsaturated and
saturated store based on K at that
depth
Infiltration can include
• sub-cell parameters for % of
compacted soil.
Effect of the M parameter
Saguling pilot in java
Atmospheric
forcing
the hydrological model simulates
reservoir inflows
hydrological
model
Forecasted
Atmospheric
forcing
climate
Reservoir
model
Storage
Outflow
the reservoir
model simulates
reservoir levels
and outflows
Inflow in a reservoir
All output distributed
Linking to Delft-FEWS


Add initial conditions to Delft-FEWS → ColdStates
Create General Adapter config

Export states

Export mapstacks

Run model

Run postadapter

Import mapstacks and/or timeseries
States

State variables:

Water Level

Unsaturated staore

Surface runoff (Q)

Saturated store
(FirstZoneDepth)

Snow depth (snow)
→ optional

Snow Water →
optional
Exporting mapstacks
Export:

P.xml → precipitation

PET.xml → potential
evapotranspiration

TEMP.xml → temperature
(optional, only needed for
snow simulation)
Running the model and adapter
wflow [-h][-v level][-F runinfofile][-L logfile][-C casename][-R runId]
[-c configfile][-T timesteps][-s seconds][-W][-E][-N][-U discharge]
[-P parameter multiplication]
<executeActivity>
<description>Run wflow</description>
<command>
<executable>bin-wflow\wflow.exe</executable>
</command>
Manual gives explanation
<arguments>
<argument>-C</argument>
<argument>rhine</argument>
<argument>-F</argument>
<argument>rhine/inmaps/runinfo.xml</argument>
</arguments>
Importing the results
<importTimeSeriesActivity>
<importFile>%ROOT_DIR%\rhine\run_default\run.tss.xml</importFile>
<timeSeriesSets>
<timeSeriesSet>
<moduleInstanceId>WFLOW_Rhine_GA_Historical</moduleInstanceId>
<valueType>scalar</valueType>
<parameterId>Q.simulated</parameterId>
<locationSetId>Rhine_Distmodel_Gauges</locationSetId>
<timeSeriesType>simulated historical</timeSeriesType>
<timeStep unit="day" multiplier="1"/>
<readWriteMode>add originals</readWriteMode>
</timeSeriesSet>
</timeSeriesSets>
</importTimeSeriesActivity>
<importMapStacksActivity>
<importFile>%ROOT_DIR%\rhine\run_default\outmaps\run.xml</importFile
Results for Rhine @Cochem
Deltares is working on expanding and better documenting the model at
the moment
Mail for more info: jaap.schellekens@deltares.nl
Thanks!