The program post-processes an AROME or ECMWF NetCDF file by using various downscaling and calibration methods. Post-processed forecasts are placed in a second Netcdf file, which has the desired output grid.

To convert between the two grids, a downscaling method is used. Currently implemented methods are:

• nearest neighbour
• elevation gradient (interpolation to new elevations using gradients)
• smart neighbours (nearest grid points at the same elevation)
• pressure (interpolation to new elevations using a standard atmosphere)

Calibrators include:

• spatial smoothing (average within a neighbourhood)
• ensemble calibration of precipitation using zero-adjusted Gamma distribution
• polynomial regression of any order
• calculation of precipitation phase, using wetbulb temperature
• calculation of QNH from surface pressure

1. Ensure the following libraries are installed:
   • Boost libraries
   • Netcdf c++ library
   • (Optional) Google test library (if developing new code)
2. Edit CC, CFLAGS_O, IFLAGS, and LFLAGS in makefile
3. Run 'make'

To see program options, run:

./gridpp

To test the program on a fake dataset, run:

./gridpp testing/files/10x10.nc testing/files/10x10_copy.nc -v T -d gradient -v Precip -d smart numSmart=3 searchRadius=3
ncview testing/files/10x10_copy.nc

To test the program on a real operational AROME file, follow these steps:

1. Make a copy of a file to post-process:

cp /starc/DNMI_AROME_METCOOP/2015/01/01/AROME_MetCoOp_00_DEF.nc_20150101 localcopy.nc

2. Create a smoothed precipitation field in localcopy.nc:

./gridpp /starc/DNMI_AROME_METCOOP/2015/01/01/AROME_MetCoOp_00_DEF.nc_20150101 localcopy.nc -v Precip -c smooth smoothRadius=10

To run using 8 threads:

export OMP_NUM_THREADS=8
./gridpp ...

Run the program in sequence for each variable:

./gridpp input output -v T ...
./gridpp input output -v Precip ...
./gridpp input output -v RH ...

Copyright (C) 2015 MET Norway. Gridded post-processor is licensed under GPL version 2 or (at your option) any later version.