Model Development Tool (MDT)¶

Welcome to the documentation for the Model Development Tool (MDT). MDT is a specialized verification system designed for Earth science model evaluation, leveraging the power of the Pangeo ecosystem and the MONET (Model Observation Evaluation Toolkit) suite of tools.

What is MDT?¶

MDT is an orchestration layer built on Prefect, allowing you to define complex model-to-observation verification workflows using simple YAML configuration files. It automates the process of loading diverse datasets, pairing models with observations, computing performance statistics, and generating visualizations.

Core Philosophy¶

MDT is designed around several key principles:

• Backend-Agnostic Processing: We prioritize xarray.DataArray and xarray.Dataset objects to ensure compatibility with both Eager (NumPy) and Lazy (Dask) computation backends.
• HPC Compatibility: Built-in support for Dask-Jobqueue allows MDT to scale seamlessly from a local workstation to large HPC clusters (e.g., SLURM, PBS).

The MONET Ecosystem¶

MDT acts as the "glue" for several foundational libraries:

• monetio: Handles the complex IO tasks of reading various atmospheric model and observation formats.
• monet: Provides core utilities for spatial and temporal pairing of model data to observation points or grids.
• monet-stats: A robust library for computing standardized atmospheric verification metrics.
• monet-plots: Specialized plotting routines tailored for atmospheric science.

Getting Started¶

To explore what MDT can do, check out the following sections:

• Quick Start Guide: Get up and running with a simple example.
• Workflow Examples: Detailed examples of complex verification tasks.
• YAML Format: Reference for defining your own workflows.