Implement HBVSimulator orchestration class

[MAfarrag]

Description

• ntroduce a new HBVSimulator class that manages simulation over time by orchestrating calls to step_run() on HBV model variants. This centralizes time control, improves reusability, and supports calibration and testing workflows.

• Create a wrapper class called HBVSimulator that encapsulates the orchestration logic needed to:

• Run hydrological simulations over time (multi-step execution),

• Handle parameter objects, state tracking, and forcing data,

• Allow model variant injection (e.g. lumped, lake, distributed),

• Provide a consistent interface for:

  • Running a simulation loop
  • Resetting states
  • Logging or exporting results.

current status

Currently, each HBV model defines a step_run() function, but:

• There is no consistent runner for stepping through time.
• Each calibration or simulation script reimplements the time loop manually.
• There is no clean separation between hydrologic model logic and time management.

By creating HBVSimulator, we encapsulate the "controller" or "simulation engine" that calls step_run() repeatedly.

class module

Create new file:

src/Hapi/core/simulator.py

New class

from typing import Type, List
import numpy as np
from Hapi.models.schemas import HBVParameters, HBVState, SimulationConstants, Forcing
from Hapi.rrm.base_model import BaseHBVModel

class HBVSimulator:
    def __init__(
        self,
        model: BaseHBVModel,
        parameters: HBVParameters,
        constants: SimulationConstants,
        initial_state: HBVState,
    ):
        self.model = model
        self.parameters = parameters
        self.constants = constants
        self.state = initial_state
        self.q_sim: List[float] = []
        self.states: List[HBVState] = []

    def run(self, forcings: List[Forcing]):
        for forcing in forcings:
            q, new_state = self.model.step_run(
                self.parameters, self.constants, forcing, self.state
            )
            self.q_sim.append(q)
            self.states.append(new_state)
            self.state = new_state

    def reset(self, state: HBVState):
        self.q_sim = []
        self.states = []
        self.state = state

methods.

Feature	Description
run()	Executes the model over a time series of forcings (e.g. precipitation, temperature)
reset()	Resets the simulator to a clean state
get_results()	(Optional) Returns discharge and state arrays
set_parameters()	(Optional) Allows replacing parameters (for calibration)
support variant injection	Accept any BaseHBVModel subclass (lumped, lake, etc.)

[MAfarrag]

graph LR
    main[run.py / calibration.py]
    simulator[core/simulator.py]
    model[rrm/hbv.py or hbv_lake.py]
    base[rrm/base_model.py]
    mixins[core/mixins.py]
    schemas[models/schemas.py]

    main --> simulator
    simulator --> model
    model --> base
    base --> mixins
    simulator --> schemas
    model --> schemas

Loading

[MAfarrag]

• All HBV models operate on raw vectors (e.g., p, p2, v, St), with no named structure or object orientation for inputs and states.

classDiagram
    class HBVModel {
        +step_run(p, p2, v, St)
    }

    class HBVLakeModel {
        +step_run(p, p2, v, St)
    }

    class HBVLumpedModel {
        +step_run(p, p2, v, St)
    }

    class HBVOldModel {
        +step_run(p, p2, v, St)
    }

    class HBVBergestromModel {
        +step_run(p, p2, v, St)
    }

Loading

• All models now consume clearly typed objects using dataclasses (or Pydantic). The simulation is now explicit and readable.

classDiagram
    class HBVParameters {
        - ltt: float
        - utt: float
        - cfmax: float
        - ...
    }

    class SimulationConstants {
        - tfac: float
        - area: float
    }

    class Forcing {
        - temp: float
        - precip: float
        - evap: float
        - ll_temp: float
    }

    class HBVState {
        - sp: float
        - sm: float
        - uz: float
        - lz: float
        - wc: float
    }

    class HBVModel {
        +step_run(HBVParameters, SimulationConstants, Forcing, HBVState)
    }

    class HBVLakeModel {
        +step_run(HBVParameters, SimulationConstants, Forcing, HBVState)
    }

    class HBVLumpedModel {
        +step_run(HBVParameters, SimulationConstants, Forcing, HBVState)
    }

    class HBVBergestromModel {
        +step_run(HBVParameters, SimulationConstants, Forcing, HBVState)
    }

    class HBVOldModel {
        +step_run(HBVParameters, SimulationConstants, Forcing, HBVState)
    }

Loading