SINDBAD Types: what comes from where

This page serves as an overview of the type definitions used across the SINDBAD ecosystem.

• Sindbad (this repository): types defined under src/Types/

• Related packages: types that are not defined in Sindbad/src/Types, but are used/re-exported by SINDBAD (e.g. SindbadTEM, TimeSamplers, ErrorMetrics, OmniTools)

Types defined in Sindbad (this repository)

Types.jl

Main module file that defines the core type system and includes type definitions for SINDBAD.

This file establishes the purpose function, which provides descriptive information about types in the SINDBAD framework. It defines the base SindbadTypes abstract type from which all other SINDBAD types inherit. The file includes all other type definition files and provides documentation for the type system.

The sections below list only the type-definition files that are currently included by src/Types/Types.jl (i.e. what the Sindbad.Types module loads).

LandTypes.jl

Defines types for land model data structures and helpers.

This file defines the LandTypes abstract type and its subtypes, which provides types for handling SINDBAD land and how the model output in every time step is organized:

• PreAlloc: Types for preallocating memory for model outputs

• LandWrapper: A wrapper for nested data structures with dot notation access

• GroupView: For accessing groups of data within a LandWrapper

• ArrayView: For accessing specific arrays within groups

Includes methods for pretty-printing and accessing data in these structures.

ArrayTypes.jl

Defines array types used throughout the SINDBAD framework.

This file defines ArrayTypes and contains two main categories of array types:

• ModelArrayType: For internal model variables (standard arrays, static arrays, views)

• OutputArrayType: For model outputs (standard arrays, MArrays, SizedArrays, YAXArrays)

These types control how data is stored and manipulated within the model and output processes.

InputTypes.jl

Defines types for handling input data and processing.

This file defines the InputTypes abstract type and its subtypes, which includes types for:

• Data backends: BackendNetcdf, BackendZarr

• Input array types: InputArray, InputKeyedArray, InputNamedDimsArray, InputYaxArray

• Forcing variable types: ForcingWithTime, ForcingWithoutTime

• Spatial subsetting methods: Spaceid, Spacelatitude, Spacelon, etc.

These types control how input data is loaded and processed.

SimulationTypes.jl

Defines types for experiment configuration and execution.

This file defines the SimulationTypes abstract type and its subtypes, which contains types for controlling model runs:

• RunFlag: Boolean flags for various model behaviors (e.g., DoSpinupTEM, DoSaveInfo)

• ParallelizationPackage: Options for parallelization (ThreadsParallelization, QbmapParallelization)

• OutputStrategy: Controls for model output behavior (DoOutputAll, DoSaveSingleFile)

These types configure how model experiments are executed.

ParameterOptimizationTypes.jl

Defines types for model optimization and parameter estimation.

This file defines the ParameterOptimizationTypes abstract type and its subtypes, which includes:

• ParameterOptimizationMethod: Various optimization algorithms (BFGS, CMA-ES, Nelder-Mead, etc.)

• GSAMethod: Global sensitivity analysis methods (Morris, Sobol)

• CostMethod: Methods for calculating cost between model and observations

• ParameterScaling: Methods for scaling parameters during optimization

These types control how model parameters are optimized.

MachineLearningTypes.jl

Defines types for machine learning and gradient calculations.

This file defines MachineLearningTypes abstract type and its subtypes, which focuses on gradient calculation methods:

• MachineLearningGradType: Abstract type for gradient calculation methods

• Various automatic differentiation methods: ForwardDiffGrad, ZygoteGrad, EnzymeGrad

• Finite difference methods: FiniteDiffGrad, FiniteDifferencesGrad

These types control how gradients are calculated for optimization.

Types from related packages (used/re-exported by SINDBAD)

SindbadTEM (package)

Defines the core TEM model type hierarchy for SINDBAD (e.g. LandEcosystem) and associated TEM-level tooling.

TimeSamplers.jl (package)

Defines time-related types for temporal subsetting and aggregation:

• TimeSampler and a comprehensive set of time scales/aggregations (e.g. TimeDay, TimeMonth, TimeYear, MSC/anomaly aggregations, etc.)