Flow Estimation

This tool represents the second step of the hydrological workflow, where the actual flow values are estimated. The flow estimation is based on a machine-learning method known as Random Forest (source).

This step is performed all at once, but three different processes are happening in the tool:

• Contributing Area of Gauging Station

• Calculate Geofactors

• Flow Estimation via Random Forest

Contributing Area of Gauging Station

A gauging station measures the flow of a river. However, to understand the flow dynamics, we need to determine the basin area contributing to each gauging station. This function allows us to calculate the area upstream each gauging station, which it is used as input in the next process (Fig. 4).

Fig. 4 On the left: a group of subcatchments with a gauging station. On the right: the contributing area of the gauging station (in green) calculated by the tool.

The gauging station shapefile should be a point shapefile representing the gauging stations within the catchment. Beside the basic information (like ID, coordinates, etc.), it should contain two columns related to Mean Flow and Mean Low Flow. The two average values should be calculated over a certain time series (e.g., 1991 - 2020). In Table 1 , you can see an example of the attribute table of gauging_stations.shp.

Table 1 Example of attribute table of gauging stations.

gml_id	Name	Mean Flow	Mean Low Flow
ID_1974116	Ahrenshagen	1.15200	0.2857
ID_1974265	Bützow Gesamt	7.89810	1.72140
ID_1974124	Güstrow	3.18780	0.6122
ID_1974292	Rostock-Geinitzbrücke	16.00440	2.86670
ID_1974127	Wolken	4.64300	0.8783

Calculate Geofactors

The flow estimation model uses a machine-learning (ML) approach to predict water flow in each subcatchment. At the core of this method is a regionalization process that builds a predictive relationship between model parameters and the physical and hydrological characteristics -referred as geofactors- of the subcatchments. These geofactors include properties such as area, slope, land use and other attributes known to influence hydrological behavior.

Once this relationship is established, it can be applied to ungauged subcatchments, allowing the model to estimate their parameters and simulate water flow even in the absence of direct measurements. The tool automatically derives the necessary geofactors from the provided input datasets, guaranteeing consistent and data-driven parameter prediction across all subcatchments. The input data that are used to calculate the geofactors are:

• digital elevation model

• land use land cover map data (particularly water bodies, forest and settlement area)

• precipitation data (a time serie equal to the time serie used for the flow at the gauging stations)

An example representing these input data related to the Warnow catchment (Germany) can be found in Fig. 5.

Fig. 5 From left to right: DEM, LULC and precipitation data for the Warnow catchment (Germany).

In Table 2 it is shown an example of possible sources where the necessary data can be found. For the ERA5 precipitation dataset, “Monthly averaged reanalysis” should be selected as Product type, “Total precipitation” as Variable and “NetCDF4” as Data format.

Table 2 Example of input data sources

Input data	Format	Source
DEM	Raster (.tif)	Copernicus GLO-30
Water area	Shapefile (.shp)	CORINE Land Cover 2018
Forest area	Shapefile (.shp)	CORINE Land Cover 2018
Settlement area	Shapefile (.shp)	CORINE Land Cover 2018
Precipitation data	NetCDF (.nc)	ERA 5 Total precipitation

Flow Estimation via Random Forest

In this step, the tool combines the outputs generated by Fix River Network and Calculate Geofactors. Using these inputs, the model is trained and validated on gauged subcatchments, allowing it to learn the relationship between geofactors and flow characteristics. Once trained, the model is applied to ungauged subcatchments to predict their flow behavior.

The resulting outputs include annual mean flow and annual mean low flow, calculated for each subcatchment or along the river network, depending on the selected configuration. Fig. 6 shows a simplified scheme of the model flow.

Fig. 6 Simplified scheme of flow estimation process.

Input data

• ungauged_subcatch_geofactors.shp (from Fix River Network)

• fixed_river_network.shp (from Fix River Network)

• gauging_stations.shp (see Contributing Area of Gauging Station)

• DEM.tif (see Calculate Geofactors)

• water_area.shp (see Calculate Geofactors)

• forest_area.shp (see Calculate Geofactors)

• settlement_area.shp (see Calculate Geofactors)

• precipitation data (see Calculate Geofactors)

Workflow

1. Go in the Processing Toolbox and look for the APRIORA plugin. Click on Hydro-Module and open 2 - Flow Estimation

2. Choose ungauged_subcatchments.shp as input for Ungauged subcatchments

3. Choose fixed_river_network.shp as input for Fixed river network

4. Choose gauging_stations.shp as input for Gauging stations

5. Select the Mean Flow field and Mean Low Flow field from gauging_stations.shp

6. Choose DEM.tif as input for Digital surface model

7. Choose water_area.shp as input for Water area

8. Choose forest_area.shp as input for Forest area

9. Choose settlement_area.shp as input for Settlement area

10. Select the precipitation data folder containing your NetCDF (.nc) data. The tool accepts multiple files (one .nc file per year) or single file (one aggregated .nc file containing the full time series). Then tick the box accordingly (e.g., if the precipitation file has been downloaded from ERA5, tick this box)

11. Select which is the driest month in the catchment (default value: August)

12. Select your preferred option for Model parameter estimation source. If your catchment has no or few gauging stations, a pre-trained model for different regions can be used. Alternatively, you can estimate the paramteres from the gauging stations data previously provided.

13. Click on Run

Video: Worflow of the Flow Estimation tool.

Output data:

• subcatchment_level.shp

• river_level.shp

• gauged_subcatch_geofactors.shp [optional]

• ungauged_subcatch_geofactors.shp [optional]

The tool generates two output including estimated flow: one at the subcatchment level and one at the river-section level. Both contain a similar set of attributes. The newly added flow-related fields are displayed in Table 3.

Table 3 New flow-related fields added to output.

Column ID	Description	Unit
Mean_Flow	Estimated annual mean flow for the individual river section or subcatchment	m³/s
M_Low_Flow	Estimated annual mean low flow for the individual river section or subcatchment	m³/s
acc_Mean	Accumulated mean flow, representing the total upstream contribution of mean flow up to the given river section	m³/s
acc_M_Low	Accumulated mean low flow, representing the total upstream contribution of mean low flow up to the given river section	m³/s

These outputs can be used for regional water balance assessments, hydrological model initialization, environmental flow studies and downstream analysis requiring spatially distributed flow estimates.

Additionally, two optional output can be generated. Open the attribute table of gauged_subcatch_geofactors.shp and check its features. Each feature represents the contributing area for a gauging station. If there is more than one gauging station, you will notice that the contributing areas might be overlapping. If you want to highlight a specific feature, right-click on it and select Flash Feature. Important to notice: there are two extra fields called Mean_Flow and M_Low_Flow that were directly transferred from gauging_stations.shp. The other fields are also present in ungauged_subcatch_geofactors.shp and Table 4 explains what each field represents.

Table 4 Attribute table of the output of “Calculate Geofactors”.

Column ID	Full name	Description	Unit
Mean_Flow [1]	Mean flow	Average standard flow calculated for a certain time series at the gauging station	m³/s
M_Low_Flow [1]	Mean Low Flow	Average low flow calculated for a certain time series at the gauging station	m³/s
H_mean	Average height	Average height within the subcatchment	m
H_stdev	Minimum height	Minimum height within the subcatchment	m
H_min	Standard deviation of the height	Standard deviation of the height within the subcatchment	m
AREA_SC	Area of the subcatchment	Area of the subcatchment	km²
PERIM_SC	Perimeter of the subcatchment	Perimeter of the subcatchment	km
SHAPE_SC	Shape of the subcatchment	Add formula somewhere	[-]
Slp_mean	Average slope	Average slope within the subcatchment	%
Slp_stdev	Standard deviation of the slope	Standard deviation of the slope within the subcatchment	%
RivNetDens	River network density	sum of the river network’s lenght within the subcatchment divided by the area of the subcatchment	km/km²
PropWatAr	Proportion of water area	(Area of water bodies divided by the area of the subcatchment)*100	%
Forest %	Forest share	(Area of forest divided by the area of the subcatchment)*100	%
Settl %	Settlement share	(Area of settlement divided by the area of the subcatchment)*100	%
PrecYearly	Yearly precipitation	Average yearly precipitation	mm
PrecDry	Dry month precipitation	Average precipitation during the dry month	mm

[1] (1,2)

Present only in gauged_subcatch_geofactors.shp