Our ability to predict air quality and to understand chemistry-climate interactions depends on a comprehensive understanding of atmospheric composition, developed through the comparison of observations and models. The MELODIES framework will help determine short-comings and uncertainties in models, assess new model developments, and identify where and what type of new observations are needed to improve our understanding of atmospheric composition and processes. This will be accomplished through the design of a modular framework that integrates diverse atmospheric chemistry observational datasets with numerical model results for the evaluation of air quality predictions. Additionally, by making observational datasets and corresponding 3D chemistry model results more accessible, a larger community - including students - will be engaged in atmospheric composition research.

​

This project uses a modular framework to integrate diverse atmospheric chemistry observational datasets with chemistry model results for the evaluation of air quality and atmospheric composition (see Figure).  This framework, MELODIES (Model EvaLuation using Observations, DIagnostics and Experiments Software), is part of the Multi-Scale Infrastructure for Chemistry and Aerosols (MUSICA) project.  It is being developed as an extension of the Model and ObservatioN Evaluation Toolkit (MONET) in collaboration with scientists in the Air Resources Laboratory, Chemical Sciences Laboratory, and Global Systems Laboratory at NOAA. 

​

As opposed to existing model evaluation tools, we are developing generic, portable, and model-agnostic software that can evaluate against a variety of observations including surface, aircraft, and satellite data. During the first year of the project, we polled the community about existing atmospheric model evaluation software tools, and received feedback from 30 researchers.  This group covered many areas of research, and had a wide range of experiences in the models and observations they use.  Many researchers said that they write a lot of their own code to compare models and observations, which is exactly what MELODIES hopes to alleviate.  We received a long list of specific desired features, but three general characteristics that we will be incorporating in MELODIES are flexibility, easy-to-use, and developer-friendly.  To address these requests, we are developing comprehensive documentation.  We will also hold training sessions and hackathons to engage the community in the use and extension of MELODIES.  We will also have a workshop particularly for graduate students and early career researchers, using MELODIES to introduce atmospheric chemistry observations and models to those new to the field.

​

MELODIES-MONET is being developed in Python (implementing Xarray, NumPy, SciPy, Pandas, and other packages) and will be made available as a development version in December, with the first public release to be made in Summer 2022.