Reading and Transforming Forecast and Observations Data

Read Gridded forecasts

• Reading and writing is done with harpIO package
• read_grid() to read from a named file
• read_forecast() to read from multiple files
• Can read grib, FA and NetCDF
• NetCDF requires the user to provide information about dimensions
• All gridded data can be interpolated to points, regridded or cross sections extracted

read_grid()

• At minimum takes a file name and a parameter
• Can also specify lead time, ensembe member, vertical level
• Output can be a list of geofields or a data frame
• Must specify the vertical level type if upper air data
• Gridded data are of geofield class.
• If more than 1 2d field a geolist is returned

Examples

read_forecast()

• A wrapper to call read_grid() for multiple files
• Needs date-times and a file template
• Sequences of date-times can be made using seq_dates()
• Can deal with lagged and multimodel ensembles
• By default no data are returned to the calling environment
• Can output to new files (when interpolate is selected)

Anatomy of a file template

• Everything in `{ }´ is replaced by dynamic data

"{fcst_model}/{YYYY}/{MM}/{DD}/{HH}/mbr{MBR3}/fc{YYYY}{MM}{DD}{HH}+{LDT3}grib"

Examples

Combining parameters

• Some parameters can be calculated as they are read in
• Most obvious example is wind speed / direction from U and V
• Parameters can be shown with show_harp_parameters()
  • And for a file format with show_harp_parameters("<format>")

Reading analysis data

• Analyses can be read with the new read_analysis() function
• It will always get every time from the file
• plot_field() doesn’t yet work properly on data frames returned by read_analysis()

Examples

Reading point observations

• The read_obs() function is used
• Works in exactly the same way as read_forecast()
• Only works for vobs currently

Examples

Other data formats

• Any data format can conceivably by read in by read_forecast() and read_obs()
• All that’s required is a function that takes the appropriate arguments and returns data in the appropriate format
• read_forecast() and read_obs() will look for a function called read_<file_format>().
• Special options can be passed via `file_format_opts()

Transformations

• Transformations are done using the transformation and transformation_opts() arguments
• Available transformations are
  • "interpolate"
  • "regrid"
  • "xsection"
• Each has a function to generate options
  • interpolate_opts()
  • regrid_opts()
  • xsection_opts()

Examples

Up next - Verification workflow

• We will cover writing out data from read_forecast() and read_obs() to SQLite files at the beginning of that session

HAVE A NICE EVENING!