Calculating temperature advection from ERA5 data

[nazimurrashidchy]

I have an assignment on reproducing this case study (https://unidata.github.io/python-training/workshop/MetPy_Case_Study/metpy-case-study/) with ERA5 data. So first I was trying to just calculate 850hPa temperature advection. Here's my code:

data = xr.open_dataset('/home/nazimur/Downloads/Data/Storm of the Century/nonamestorm.nc').sel(time=data.time[0])
lat = data.latitude.data
lon = data.longitude.data
dx, dy = mpcalc.lat_lon_grid_deltas(lon, lat)

t850 = gaussian_filter(data.t.sel(level=data.level[2]), sigma=1.0)*units.K
h850 = gaussian_filter(data.z.sel(level=data.level[2]), sigma=1.0)*units.meter
u850 = gaussian_filter(data.u.sel(level=data.level[2]), sigma=1.0)*units('m/s')
v850 = gaussian_filter(data.v.sel(level=data.level[2]), sigma=1.0)*units('m/s')

temp_adv_850 = mpcalc.advection(t850, [u850, v850], (dx, dy), dim_order='yx').to('degC/s')

Output is given as an error: got an unexpected keyword argument 'dim_order'

Can you please point out what is wrong here?

[dopplershift]

MetPy 1.0 changed to more natively work with xarray data and eliminate the dim_order argument. The upshot of all of that is you should be able to do:

data = xr.open_dataset('/home/nazimur/Downloads/Data/Storm of the Century/nonamestorm.nc').sel(time=data.time[0])

t850 = mpcalc.smooth_gaussian(data.t.sel(level=data.level[2]), n=6)
u850 = mpcalc.smooth_gaussian(data.u.sel(level=data.level[2]), n=6)
v850 = mpcalc.smooth_gaussian(data.v.sel(level=data.level[2]), n=6)

temp_adv_850 = mpcalc.advection(t850, u=u850, v=v850).metpy.convert_units('degC/s')

MetPy has smooth_gaussian which wraps gaussian_filter, but using an n of 6 should be close to what you had before. Going that way lets MetPy handle making sure you get all the right unit-handling and xarray metadata after smoothing.

The example you reference is unfortunately out of date. We have some updated copies, but for various reasons have not updated the published content, though doing so is very high on our todo list.

[nazimurrashidchy]

Thank you! But there's still an issue. There's an attribute error given : 'DataArray' object has no attribute 'to'. How do I solve this problem?

[dopplershift]

🐑 The bug is always in the line you think "That looks right, I don't need to run it".

I've updated my sample to fix the call to to() (which only works with Pint Quantity instances) to use MetPy's xarray accessor (.metpy.convert_units()), since the output from the function is a DataArray.

[nazimurrashidchy]

The unit of geopotential height in ERA5 data is m^2s^-2. I need to convert it to meter. If I try to change its unit by writing
h500 = h500.metpy.convert_units('meter')
it gives "DimensionalityError: Cannot convert from 'meter ** 2 / second ** 2' ([length] ** 2 / [time] ** 2) to 'meter' ([length])"

[DanielAdriaansen]

ERA5 does not provide geopotential height, but rather geopotential (hence the units you are seeing, which are m^2s^2 and not m). Try using this:
https://unidata.github.io/MetPy/latest/api/generated/metpy.calc.geopotential_to_height.html