HydPy-Dam-L-Res (reservoir model adopted from LARSIM)

dam_lreservoir is a relatively simple reservoir model, similar to the “TALS” model of LARSIM. It combines the features of dam_llake (“controlled lake”) and dam_lretention (“retention basin”). Additionally, it allows controlling the stored water volume via defining target values that can vary seasonally.

Like dam_lretention, dam_lreservoir allows for combining controlled, “harmless outflow” (via parameter AllowedRelease) and uncontrolled, “spillway outflow” (via parameter WaterLevel2FloodDischarge), and like dam_llake, it allows restricting the speed of the water level decrease during periods with little inflow via parameter AllowedWaterLevelDrop (only through reducing the controlled outflow, of course). Before continuing, please first read the documentation on these two application models.

The additional feature of dam_lreservoir is its ability to track seasonal target volumes. We define these target volumes via parameter TargetVolume. The parameters VolumeTolerance, TargetRangeAbsolute, and TargetRangeRelative serve to yield more smooth and realistic reservoir responses for slight deviations from the given target values. Setting TargetRangeRelative to 0.2 and both other parameters to zero corresponds to selecting the “TALSPERRE SOLLRANGE” option in LARSIM. Please see the following examples and the documentation on method Calc_ActualRelease_V3 for more information on setting and combining the individual parameter values for different use cases.

Integration tests

Note

When new to HydPy, consider reading section Integration Tests first.

We prepare a test set similar to the ones for application models dam_llake and dam_lretention, including an identical inflow series and an identical relationship between stage and volume:

>>> from hydpy import IntegrationTest, Element, pub
>>> pub.timegrids = "01.01.2000", "21.01.2000", "1d"
>>> from hydpy.models.dam_lreservoir import *
>>> parameterstep("1d")
>>> element = Element("element", inlets="input_", outlets="output")
>>> element.model = model
>>> test = IntegrationTest(element)
>>> test.dateformat = "%d.%m."
>>> test.plotting_options.axis1 = fluxes.inflow, fluxes.outflow
>>> test.plotting_options.axis2 = states.watervolume
>>> test.inits = [
...     (states.watervolume, 0.0),
...     (logs.loggedadjustedevaporation, 0.0)]
>>> test.reset_inits()
>>> conditions = model.conditions
>>> watervolume2waterlevel(PPoly.from_data(xs=[0.0, 1.0], ys=[0.0, 1.0]))
>>> surfacearea(1.44)
>>> catchmentarea(86.4)
>>> correctionprecipitation(1.2)
>>> correctionevaporation(1.2)
>>> weightevaporation(0.8)
>>> thresholdevaporation(0.0)
>>> toleranceevaporation(0.001)
>>> with model.add_precipmodel_v2("meteo_precip_io") as precipmodel:
...     precipitationfactor(1.0)
>>> precipmodel.prepare_inputseries()
>>> precipmodel.sequences.inputs.precipitation.series = [
...     0.0, 50.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0,
...     0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
>>> element.inlets.input_.sequences.sim.series = [
...     0.0, 0.0, 6.0, 12.0, 10.0, 6.0, 3.0, 2.0, 1.0, 0.0,
...     0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]

base scenario

First, we again use the linear relation between discharge and stage used throughout the integration tests of dam_llake and in the base example of dam_lretention:

>>> waterlevel2flooddischarge(PPoly.from_data(xs=[0.0, 1.0], ys=[0.0, 10.0]))

Additionally, we set some of the remaining parameter values extremely high or low to ensure the reservoir stores all water except the one activating the spillway, which becomes “flood discharge”:

>>> targetvolume(100.0)
>>> neardischargeminimumthreshold.shape = 1
>>> neardischargeminimumthreshold.values = -100.0
>>> targetrangeabsolute(0.1)
>>> targetrangerelative(0.2)
>>> watervolumeminimumthreshold(0.0)
>>> volumetolerance(0.1)
>>> dischargetolerance(0.1)
>>> allowedrelease(100.0)
>>> allowedwaterleveldrop(100.0)

Due to the same the neural network configuration, the results are identical to the ones of the base example of dam_llake and the base example of dam_lretention:

>>> test("dam_lreservoir_base_scenario")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> from hydpy import round_
>>> round_(model.check_waterbalance(conditions))
0.0

spillway

When we reuse the more realistic relationship between flood discharge and stage of the spillway example on dam_lretention, we again get the same flood discharge time series:

>>> waterlevel2flooddischarge(ANN(weights_input=10.0, weights_output=50.0,
...                               intercepts_hidden=-20.0, intercepts_output=0.0))
>>> test("dam_lreservoir_spillway")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> round_(model.check_waterbalance(conditions))
0.0

target volume

During dry periods, application model dam_lretention generally releases all its water until the basin runs dry, as long as AllowedRelease is larger than zero. Application model dam_lreservoir instead allows fo defining target volumes. dam_lreservoir tries to control its outflow so that the actual volume approximately equals the (potentially seasonally varying) target volume. However, it cannot release arbitrary amounts of water to fulfil this task due to its priority to release a predefined minimum amount of water (for ecological reasons) and its second priority to not release too much water (for flood protection). In this example, we activate these mechanisms by changing some corresponding parameter values (see the documentation on method Calc_ActualRelease_V3 for more detailed examples, including the numerous corner cases):

>>> targetvolume(0.5)
>>> neardischargeminimumthreshold(0.1)
>>> allowedrelease(4.0)
>>> allowedwaterleveldrop(1.0)

Compared with the allowed release results of dam_lretention, dam_lreservoir dampens the given flood event less efficiently. dam_lretention releases all initial inflow, while dam_lreservoir stores most of it until it reaches the target volume of 0.5 million m³. After peak flow, dam_lreservoir first releases its water as fast as allowed but then tries to meet the target volume again. The slow negative trend away from the target value at the end of the simulation period results from the lack of inflow while still needing to release at least 0.1 m³/s:

>>> test("dam_lreservoir_target_volume")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> round_(model.check_waterbalance(conditions))
0.0

sharp transitions

Due to smoothing, the above results deviate from the ones one would expect from LARSIM simulations to some degree. However, if we set both “target range” parameters to zero (like one does not select the LARSIM option “TALSPERRE SOLLRANGE”) and both “tolerance” parameters to zero (to turn off any smoothing), we get more similar results:

>>> targetrangeabsolute(0.0)
>>> targetrangerelative(0.0)
>>> volumetolerance(0.0)
>>> dischargetolerance(0.0)
>>> test("dam_lreservoir_sharp_transitions")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> round_(model.check_waterbalance(conditions))
0.0

higher accuracy

The first water volume calculated in the sharp transitions example is negative, resulting from the limited numerical accuracy of the underlying integration algorithm. We can decrease such errors by defining smaller error tolerances but at the risk of relevant increases in computation times (especially in case one applies zero smoothing values):

>>> solver.abserrormax(1e-6)
>>> test("dam_lreservoir_higher_accuracy")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> round_(model.check_waterbalance(conditions))
0.0

target range

In the last example, the reservoir behaviour changes abruptly when the actual volume transcends the target volume. According to its documentation, LARSIM predicts unrealistic jumps in discharge in such cases. To solve this issue, LARSIM offers the “TALSPERRE SOLLRANGE” option, which ensures smoother transitions between 80 % and 120 % of the target volume, accomplished by linear interpolation. dam_lreservoir should never output similar jumps as it controls the correctness of its results. As a drawback, correcting these jumps (which still occur “unseeable” and possibly multiple times within each affected simulation time step) costs computation time. Hence, at least for small smoothing parameter values, dam_lreservoir can also benefit from this approach. You can define the interpolation range freely via TargetRangeAbsolute and TargetRangeRelative, depending on your specific needs. Setting the latter to 0.2 corresponds to the original “TALSPERRE SOLLRANGE”-configuration:

>>> targetrangerelative(0.2)
>>> test("dam_lreservoir_target_range")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> round_(model.check_waterbalance(conditions))
0.0

minimum volume

In all the examples above, the dam would run dry entirely after a certain time to fulfil the downstream demand defined by parameter NearDischargeMinimumThreshold. Usually, this is neither desired nor technically possible. The following example shows that the parameter WaterVolumeMinimumThreshold allows setting a minimum amount of water below which no release occurs:

>>> watervolumeminimumthreshold(0.45)
>>> test("dam_lreservoir_minimum_volume")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> round_(model.check_waterbalance(conditions))
0.0

evaporation

This example takes up the evaporation example of application model dam_llake. The reservoir can no longer maintain the target water level at the end of the simulation period due to missing precipitation or inflow for compensating evaporation:

>>> with model.add_pemodel_v1("evap_ret_io") as pemodel:
...     evapotranspirationfactor(1.0)
>>> pemodel.prepare_inputseries()
>>> pemodel.sequences.inputs.referenceevapotranspiration.series = 10 * [1.0] + 10 * [5.0]
>>> test("dam_lreservoir_evaporation")
Click to see the table
Click to see the graph

There is no indication of an error in the water balance:

>>> round_(model.check_waterbalance(conditions))
0.0
class hydpy.models.dam_lreservoir.Model[source]

Bases: Main_PrecipModel_V2, Main_PEModel_V1

HydPy-Dam-L-Res (reservoir model adopted from LARSIM).

The following “inlet update methods” are called in the given sequence at the beginning of each simulation step:
The following methods define the relevant components of a system of ODE equations (e.g. direct runoff):

  • Calc_AdjustedPrecipitation_V1 Adjust the given precipitation.

  • Pic_Inflow_V1 Update the inlet sequence Inflow.

  • Calc_WaterLevel_V1 Determine the water level based on an interpolation approach approximating the relationship between water volume and water level.

  • Calc_ActualEvaporation_V1 Calculate the actual evaporation.

  • Calc_SurfaceArea_V1 Determine the surface area based on an interpolation approach approximating the relationship between the water level and the surface area.

  • Calc_AllowedDischarge_V2 Calculate the maximum discharge not leading to exceedance of the allowed water level drop.

  • Calc_ActualRelease_V3 Calculate an actual water release that tries to change the water storage into the direction of the actual target volume without violating the required minimum and the allowed maximum flow.

  • Calc_FloodDischarge_V1 Calculate the discharge during and after a flood event based on seasonally varying interpolation approaches approximating the relationship(s) between discharge and water stage.

  • Calc_Outflow_V1 Calculate the total outflow of the dam.

The following methods define the complete equations of an ODE system (e.g. change in storage of fast water due to effective precipitation and direct runoff):
The following “outlet update methods” are called in the given sequence at the end of each simulation step:

  • Calc_WaterLevel_V1 Determine the water level based on an interpolation approach approximating the relationship between water volume and water level.

  • Pass_Outflow_V1 Update the outlet link sequence Q.

Users can hook submodels into the defined main model if they satisfy one of the following interfaces:

  • PrecipModel_V2 Simple interface for determining precipitation in one step.

  • PETModel_V1 Simple interface for calculating all potential evapotranspiration values in one step.

DOCNAME: DocName = ('Dam-L-Res', 'reservoir model adopted from LARSIM')
precipmodel: modeltools.SubmodelProperty

Optional submodel that complies with the following interface: PrecipModel_V2.

pemodel: modeltools.SubmodelProperty

Optional submodel that complies with the following interface: PETModel_V1.

check_waterbalance(initial_conditions: dict[str, dict[str, dict[str, float | ndarray[Any, dtype[float64]]]]]) → float[source]

Determine the water balance error of the previous simulation run in million m³.

Method check_waterbalance() calculates the balance error as follows:

\(Seconds \cdot 10^{-6} \cdot \sum_{t=t0}^{t1} \big( AdjustedPrecipitation_t - ActualEvaporation_t + Inflow_t - Outflow_t \big) + \big( WaterVolume_{t0}^k - WaterVolume_{t1}^k \big)\)

The returned error should always be in scale with numerical precision so that it does not affect the simulation results in any relevant manner.

Pick the required initial conditions before starting the simulation run via property conditions. See the integration tests of the application model dam_lreservoir for some examples.

REUSABLE_METHODS: ClassVar[tuple[type[ReusableMethod], ...]] = ()
class hydpy.models.dam_lreservoir.AideSequences(master: Sequences, cls_fastaccess: type[TypeFastAccess_co] | None = None, cymodel: CyModelProtocol | None = None)

Bases: AideSequences

Aide sequences of model dam_lreservoir.

The following classes are selected:
class hydpy.models.dam_lreservoir.ControlParameters(master: Parameters, cls_fastaccess: type[FastAccessParameter] | None = None, cymodel: CyModelProtocol | None = None)

Bases: SubParameters

Control parameters of model dam_lreservoir.

The following classes are selected:
class hydpy.models.dam_lreservoir.DerivedParameters(master: Parameters, cls_fastaccess: type[FastAccessParameter] | None = None, cymodel: CyModelProtocol | None = None)

Bases: SubParameters

Derived parameters of model dam_lreservoir.

The following classes are selected:
class hydpy.models.dam_lreservoir.FactorSequences(master: Sequences, cls_fastaccess: type[TypeFastAccess_co] | None = None, cymodel: CyModelProtocol | None = None)

Bases: FactorSequences

Factor sequences of model dam_lreservoir.

The following classes are selected:
class hydpy.models.dam_lreservoir.FluxSequences(master: Sequences, cls_fastaccess: type[TypeFastAccess_co] | None = None, cymodel: CyModelProtocol | None = None)

Bases: FluxSequences

Flux sequences of model dam_lreservoir.

The following classes are selected:
class hydpy.models.dam_lreservoir.InletSequences(master: Sequences, cls_fastaccess: type[TypeFastAccess_co] | None = None, cymodel: CyModelProtocol | None = None)

Bases: InletSequences

Inlet sequences of model dam_lreservoir.

The following classes are selected:

  • Q() Inflow [m³/s].

class hydpy.models.dam_lreservoir.LogSequences(master: Sequences, cls_fastaccess: type[TypeFastAccess_co] | None = None, cymodel: CyModelProtocol | None = None)

Bases: LogSequences

Log sequences of model dam_lreservoir.

The following classes are selected:
class hydpy.models.dam_lreservoir.OutletSequences(master: Sequences, cls_fastaccess: type[TypeFastAccess_co] | None = None, cymodel: CyModelProtocol | None = None)

Bases: OutletSequences

Outlet sequences of model dam_lreservoir.

The following classes are selected:

  • Q() Outflow [m³/s].

class hydpy.models.dam_lreservoir.SolverParameters(master: Parameters, cls_fastaccess: type[FastAccessParameter] | None = None, cymodel: CyModelProtocol | None = None)

Bases: SubParameters

Solver parameters of model dam_lreservoir.

The following classes are selected:

  • AbsErrorMax() Absolute numerical error tolerance [m³/s].

  • RelErrorMax() Relative numerical error tolerance [1/T].

  • RelDTMin() Smallest relative integration time step size allowed [-].

  • RelDTMax() Largest relative integration time step size allowed [-].

class hydpy.models.dam_lreservoir.StateSequences(master: Sequences, cls_fastaccess: type[TypeFastAccess_co] | None = None, cymodel: CyModelProtocol | None = None)

Bases: StateSequences

State sequences of model dam_lreservoir.

The following classes are selected: