The cwcomposite tool combines data variables across multiple data files by computing a value at each pixel: mean, median, minimum, maximum, or latest value. There are generally two scenarios for using cwcomposite — (i) for creating a spatial composite of data sampled over a short time period (~ 1 day):

and (ii) for creating a temporal composite of data sampled over a longer time period (days/weeks/months):

Type the following command to create a 7-day composite from daily SMAP sea surface salinity data:

cwcomposite -v --match sss example_smap_salinity_apr_*.nc example_smap_salinity_apr_7day.hdf

You should see the composite running in verbose mode as follows:

Terminal screen showing cwcomposite running on SMAP salinity files

The default for cwcomposite is to compute the mean value at each pixel — other composite methods are set using the --method option. You can view the output file in CDAT, or render it as follows:

cwrender --enhance sss --range 30/40 --palette Ocean-haline --grid white --coast black/gray40 example_smap_salinity_apr_7day.hdf example_smap_salinity_apr_7day.png

You should see the following image:

Global SMAP sea surface salinity

Bonus exercises:

Read the cwcomposite Unix man page or user's guide section.
Add an option to the cwcomposite command line to compute the minimum value at each pixel.
Find out how to make cwcomposite mark a pixel invalid if the number of valid input values at the pixel is less than 5.
Suppose a researcher asked for an uninterrupted image of the Pacific Ocean. Try using cwmaster and cwregister2 from the previous exercises to create one, for example: