Climate Model Diagnostic Analyzer (CMDA) is a collaborative platform to support the full life cycle of a data analysis process, from data discovery, to data customization, to analysis, to reanalysis, to publication/sharing, and to reproduction [1,2]. CMDA was initially developed to demonstrate the methodology to evaluate and diagnose climate models through the comprehensive use of multiple observational data, reanalysis data, and model outputs. It has evolved to support collaborative scientific activities and the full life cycle of data analysis. The CMDA project has been funded by the NASA ROSES programs.

The CMDA HTML interface provides a user with a full control on input parameter selections for a given service such as dataset selection, dataset subsetting condition, dataset analysis functionality, and output data visualization parameters.

The CMDA Jupyter Notebook Interface allows users to make direct API requests to the CMDA webservices to retrieve the datasets of their choice and perform their own analysis on a Jupyter Hub server.

CMDA hosts over 2000 datasets covering model datasets, observational datasets, and reanalysis datasets. The model datasets include CMIP5 historical runs, CMIP5 AMIP runs, CMIP5 RCP4.5 projection runs, and WRF model runs with various physical parameterizations. The observation datasets cover many satellite data (AIRS, AMSR-E, AVISO, CERES, GRACE, GPCP, GPM, ISCCP, MISR, MODIS, MLS, QuickSCAT, SMAP, TES, TRMM) and ship-floats data (ARGO). The reanalysis datasets include ECMWF and GLDAP data.

CMDA has been successfully utilized in a real-world user environment. Since 2014, the annual NASA Summer School on Satellite Observations and Climate Models has used CMDA as a collaborative computing platform for students to conduct their group projects. The summer school students were given a short introduction to CMDA and were able to use it to access datasets and to analyze the datasets to generate their group project result within one week of the summer school. CMDA leveraged Amazon Elastic Kubernetes Service clusters to scale up its computation resources according to the demands from the students.

References