1.2 Types of hydrological models

Hydrological models come in various types, each suited for different purposes. From lumped models that treat catchments as single units to distributed models that capture spatial variability, the choice depends on data, resources, and study goals.

Understanding model types is crucial for effective hydrological modeling. Deterministic models use physical laws, while stochastic models incorporate uncertainties. Each type has its strengths and limitations, impacting their application in water resource management and planning.

Hydrological Model Classification

Spatial Scale Classification

Top images from around the web for Spatial Scale Classification

• 

  HESS - On the representation of water reservoir storage and operations in large-scale ... View original

  Is this image relevant?

• 

  HESS - Understanding dominant controls on streamflow spatial variability to set up a semi ... View original

  Is this image relevant?

• 

  HESS - Understanding dominant controls on streamflow spatial variability to set up a semi ... View original

  Is this image relevant?

• 

  HESS - On the representation of water reservoir storage and operations in large-scale ... View original

  Is this image relevant?

• 

  HESS - Understanding dominant controls on streamflow spatial variability to set up a semi ... View original

  Is this image relevant?

1 of 3

Top images from around the web for Spatial Scale Classification

• 

  HESS - On the representation of water reservoir storage and operations in large-scale ... View original

  Is this image relevant?

• 

  HESS - Understanding dominant controls on streamflow spatial variability to set up a semi ... View original

  Is this image relevant?

• 

  HESS - Understanding dominant controls on streamflow spatial variability to set up a semi ... View original

  Is this image relevant?

• 

  HESS - On the representation of water reservoir storage and operations in large-scale ... View original

  Is this image relevant?

• 

  HESS - Understanding dominant controls on streamflow spatial variability to set up a semi ... View original

  Is this image relevant?

1 of 3

• Hydrological models can be classified based on their spatial scale, which refers to the level of detail and granularity at which the model represents the hydrological processes and the catchment area
  • Lumped models treat the entire catchment as a single unit, with averaged or representative values for hydrological variables and parameters (e.g., HBV, Sacramento)
  • Semi-distributed models divide the catchment into smaller sub-units or hydrological response units (HRUs) based on similar hydrological characteristics, such as land use, soil type, or elevation (e.g., SWAT, TOPMODEL)
  • Distributed models discretize the catchment into a grid or mesh of small elements, allowing for a more detailed representation of spatial variability in hydrological processes and catchment characteristics (e.g., MIKE SHE, ParFlow)
  • The choice of spatial scale depends on the available data, computational resources, and the purpose of the modeling study

Temporal Scale Classification

• Hydrological models can also be classified based on their temporal scale, which refers to the time step or resolution at which the model simulates the hydrological processes
  • Event-based models simulate individual storm events or short-term periods, typically focusing on surface runoff generation and routing (e.g., HEC-HMS, KINEROS)
  • Continuous models simulate the hydrological processes over extended periods, such as months, years, or decades, considering the long-term water balance and storage dynamics (e.g., VIC, PRMS)
  • The temporal scale selection depends on the modeling objectives, data availability, and the dominant hydrological processes in the catchment
  • Models with finer temporal resolution can capture short-term dynamics and extreme events, while models with coarser resolution are suitable for long-term water resources planning and management

Deterministic vs Stochastic Models

Deterministic Models

• Deterministic hydrological models are based on physical laws and equations that describe the hydrological processes, assuming that the model outputs are uniquely determined by the model inputs and parameters
  • Deterministic models do not explicitly account for uncertainties in the input data, model structure, or parameters
  • These models provide a single set of outputs for a given set of inputs and parameters
  • Examples of deterministic models include physically-based models like SWAT, HBV, and TOPMODEL
  • Deterministic models are suitable for understanding the underlying physical processes and for scenario analysis and impact assessment

Stochastic Models

• Stochastic hydrological models incorporate probabilistic or random components to represent the inherent uncertainties and variability in hydrological processes and data
  • Stochastic models use statistical techniques, such as probability distributions, random variables, or stochastic differential equations, to describe the hydrological processes and their uncertainties
  • These models can provide probabilistic predictions or ensemble forecasts, quantifying the uncertainty in the model outputs
  • Examples of stochastic models include time series models (e.g., ARMA, ARIMA), probabilistic models (e.g., Markov chains), and stochastic differential equation models (e.g., stochastic rainfall-runoff models)
  • Stochastic models are useful for risk assessment, decision-making under uncertainty, and for quantifying the reliability of hydrological predictions

Lumped, Semi-distributed, and Distributed Models

Lumped Models

• Lumped models represent the entire catchment as a single unit, with averaged or representative values for hydrological variables and parameters
  • Lumped models are computationally efficient and require fewer input data compared to semi-distributed and distributed models
  • They treat the catchment as a homogeneous entity, ignoring the spatial variability of hydrological processes and catchment characteristics
  • Examples of lumped models include the Stanford Watershed Model, the Sacramento Soil Moisture Accounting model, and the HBV model
  • Lumped models are suitable for catchments with homogeneous characteristics or when limited data availability prevents the use of more complex models

Semi-distributed Models

• Semi-distributed models divide the catchment into smaller sub-units or hydrological response units (HRUs) based on similar hydrological characteristics
  • Semi-distributed models strike a balance between the simplicity of lumped models and the complexity of distributed models
  • They can capture some level of spatial variability while maintaining computational efficiency
  • Examples of semi-distributed models include the Soil and Water Assessment Tool (SWAT), the TOPography-based hydrological MODEL (TOPMODEL), and the Hydrologic Engineering Center's Hydrologic Modeling System (HEC-HMS)
  • Semi-distributed models require more detailed input data compared to lumped models but less than fully distributed models

Distributed Models

• Distributed models discretize the catchment into a grid or mesh of small elements, allowing for a detailed representation of spatial variability in hydrological processes and catchment characteristics
  • Distributed models can capture the heterogeneity of the catchment, including variations in topography, land use, soil properties, and meteorological conditions
  • They require extensive input data, including high-resolution spatial data (e.g., digital elevation models, land use maps) and detailed meteorological data
  • Examples of distributed models include the MIKE SHE model, the ParFlow model, and the Variable Infiltration Capacity (VIC) model
  • Distributed models are computationally intensive and may face challenges related to parameter estimation and calibration due to their high dimensionality

Advantages and Limitations of Models

Lumped Model Advantages and Limitations

• Lumped models:
  • Advantages: Simplicity, computational efficiency, fewer data requirements, suitable for catchments with homogeneous characteristics or limited data availability
  • Limitations: Inability to capture spatial variability, limited applicability in heterogeneous catchments, oversimplification of complex hydrological processes
  • Lumped models are useful for rapid assessments, long-term water balance studies, and for providing a general understanding of catchment behavior
  • However, they may not adequately represent the spatial variability of hydrological processes, particularly in large or heterogeneous catchments

Semi-distributed Model Advantages and Limitations

• Semi-distributed models:
  • Advantages: Balance between simplicity and complexity, ability to capture some level of spatial variability, improved representation of hydrological processes compared to lumped models
  • Limitations: Requires more detailed input data compared to lumped models, may not fully capture the spatial heterogeneity of the catchment
  • Semi-distributed models are suitable for catchments with moderate spatial variability and when computational efficiency is important
  • They can provide a reasonable compromise between model complexity and data requirements

Distributed Model Advantages and Limitations

• Distributed models:
  • Advantages: Detailed representation of spatial variability, ability to capture the heterogeneity of the catchment, improved accuracy in simulating hydrological processes
  • Limitations: High computational requirements, extensive data needs, challenges in parameter estimation and calibration, potential issues with over-parameterization and equifinality
  • Distributed models are suitable for catchments with significant spatial variability and when a detailed understanding of hydrological processes is required
  • However, the high data and computational requirements may limit their applicability in data-scarce regions or for real-time forecasting

Deterministic Model Advantages and Limitations

• Deterministic models:
  • Advantages: Based on physical laws and equations, provide a mechanistic understanding of hydrological processes, suitable for scenario analysis and impact assessment
  • Limitations: Do not explicitly account for uncertainties, may require extensive input data and parameter estimation, limited ability to quantify prediction uncertainty
  • Deterministic models are useful for understanding the cause-effect relationships in hydrological systems and for evaluating the impacts of land use or climate change
  • However, they may not adequately capture the inherent uncertainties in hydrological processes and predictions

Stochastic Model Advantages and Limitations

• Stochastic models:
  • Advantages: Incorporate uncertainties and variability, provide probabilistic predictions or ensemble forecasts, suitable for risk assessment and decision-making under uncertainty
  • Limitations: Require assumptions about the statistical properties of hydrological processes, may not fully capture the physical basis of the system, limited ability to extrapolate beyond the range of observed data
  • Stochastic models are useful for quantifying the uncertainty in hydrological predictions, for generating probabilistic flood or drought forecasts, and for risk-based water resources planning and management
  • However, the assumptions about the statistical properties of hydrological processes may not always hold, and stochastic models may have limited predictive power for extreme events or under changing environmental conditions