how to use EC-EARTH data in HYSPLIT?

[thomas]

Hello all,

I want to compute backward trajectories using HYSPLIT4 (I have the windows PC version -may 2012 release- installed). The data I want to use are (historical) EC-EARTH (CMIP5) data, downloaded from the ESGF website. I have (in netcdf format):

* u and v windcomponents (horizontal wind)
* temperature
* pressure at surface
* geopotential heigth

The dataset is gridded and the timespan is 3 hours. The number of levels however, is rather low (3 for the wind and temperature and 7 for the geopotential heigth). According to the description of the HYSPLIT_4 modelling system (Draxler; Hess; 2004) I have the minimum variables needed.
With cdo I made grib1 files (but I also have the netcdf files). How can I convert this to the ARL format which I later on can use to make the trajectories?

Many thanks in advance,
Thomas.

[chron]

What is the vertical coordinate? If the winds are on isobaric surfaces, you could at least compute isobaric trajectories. I fear that 3 levels would be too few to perform reasonable interpolation of vertical motion, though.

You might be able to modify arw2arl.f to do the necessary conversion (it works with netcdf, and doesn't require too many changes for different models). I assume that the data are on a regular horizontal grid, since the output is from CMIP5. This aspect should make the conversion much easier (raw output is often on gaussian grids, which are not supported).

[thomas]

Thank you for your quick reply. The vertical coordinate are pressure levels, so indeed I can only make isobaric trajectories. And as for my problem of too few vertical levels, I thought that I maybe can perform a vertical interpolation of the different fields first (although I know that this does not improve the accuracy)?

I shall investigate the possibility of using the arw2arl tool, but in my first try I got an error already (couldn't execute ".../exec/arw2arl..."). When I started the tool, I also got a message "NetCDF I/O not available under Windows".

[chron]

I imagine that the levels are something like 850 hPa, 700 hPa, 500 hPa? You can certainly interpolate this vertically, but again it will not be very accurate because there should be lots of directional shear between those heights. I would recommend using isobaric trajectories and initializing them as close as possible to the data levels. Hysplit will actually do the interpolation for you, but I wouldn't try to stray far from the data.

It sounds like you're using the windows version, and that arw2arl wasn't compiled for this. I don't have any experience setting up netcdf libraries for hysplit on Windows, but it isn't too hard on linux/unix/mac. Perhaps someone else can help you with that, unless you have access to a unix system as well.