Welcome to AmesCAP’s Documentation!

AmesCAP is the name of the repository hosting the Community Analysis Pipeline (CAP), a Python-based command-line tool that performs analysis and creates plots from netCDF files output by the Mars Global Climate Model (MGCM). The MGCM and AmesCAP are developed and maintained by the Mars Climate Modeling Center (MCMC) at NASA’s Ames Research Center in Mountain View, CA.

CAP is inspired by the need for increased access to MGCM output. MGCM data products are notoriously complex because the output files are typically tens of GB in size, hold multi-dimensional arrays on model-specific computational grids, and require post-processing in order to make the data usable in scientific, engineering, and educational applications. From simple command-line calls to CAP executables, users can access functions that pressure-interpolate, regrid, perform tidal analyses and time averages (e.g., diurnal, hourly, composite days), and derive secondary variables from the data in MGCM output files.

An image illustrating the CAP plot cycle.

CAP also has a robust plotting routine that requires no coding to use. CAP’s plotting routine references a template that CAP generates and the user modifies to specify the figures CAP will create. Templates are generalizable and can be referenced repeatedly to create plots from multiple data products. A web-based version of CAP’s plotting routine is available on the NAS Data Portal: the Mars Climate Modeling Center Data Portal Web Interface is a point-and-click plotting tool that requires no coding or command-line interaction to use. The Web Interface can create plots from MGCM data hosted on the NAS Data Portal and it can even provide the user a netCDF file of the subset of the data from which the plots were created.

CAP reads output from the MCMC’s Legacy and FV3-based MGCMs directly. Both models are publicly available on GitHub, and output from simulations performed with them is provided by the MCMC on the NAS Data Portal here. With MarsFormat, CAP also converts output from the LMD Mars Planetary Climate Model (PCM), the Mars Weather Research and Forecasting Model (MarsWRF), and the OpenMARS and EMARS reanalyses to the same format.

Note

CAP is continually in development and we appreciate any feedback you have for us.

Contents:

Indices and tables