def get_seaflux_data(catalog_name=catalog_name,
dest=_dest,
n_jobs=1,
verbose=False):
"""Downloads SeaFlux data from Zenodo using the default yaml file containing
the paths to the latest SeaFlux data. The data is downloaded and then
combined. You can create your own yaml file to customise the files you want
to access."""
from datetime import datetime as dt
import fetch_data as fd
import xarray as xr
from . import config
from .utils import preprocess
cat = fd.read_catalog(catalog_name)
key = list(cat.keys())[0]
entry = cat[key]
entry["dest"] = dest
flist = fd.download(**entry, n_jobs=n_jobs, verbose=verbose)
xds = xr.open_mfdataset(flist, preprocess=preprocess())
xds = xds.assign_attrs(product_name="SeaFlux",
product_version=config.version,
date_accessed=dt.now().strftime("%Y-%m-%d"),
contact=config.contact,
**entry["meta"])
return xds
def get_jena_mls(entry):
"""processes data"""
flist = download(**entry)
xds = xr.open_mfdataset(flist)
xda = xds.pCO2.resample(mtime="1MS").mean("mtime")
xda = xda.rename("jena_mls")
xda = (xda.interp(
lat=np.arange(-89.5, 90),
lon=np.arange(-179.5, 180),
method="nearest"
).roll(lon=180, roll_coords=False).interpolate_na(
"lon", limit=20
).roll(
lon=-180,
roll_coords=False
).rename(
mtime="time"
).assign_attrs(
units="uatm",
source=entry["url"],
**
entry["meta"],
history=(
"[Seaflux] resampled from daily to monthly and "
"interpolated to 1 degree using nearest neighbour interpolation"
),
))
xda = preprocess()(xda)
return xda
def get_jma_mlr(entry):
"""processes data"""
def decode_time(xds):
"""processes data"""
import pandas as pd
from seaflux.data.utils import add_history
time = xds.time
unit = time.attrs.get("units")
year = pd.to_datetime(unit.split()[-1]).year
y0, y1 = str(year), str(year + 1)
time = pd.date_range(y0, y1, freq="1MS", closed="left")
xds = xds.assign_coords(time=time)
xds = add_history(xds, "decode times manually")
return xds
flist = download(**entry, n_jobs=8)
xda = xr.open_mfdataset(flist,
decode_times=False,
preprocess=preprocess(decode_time)).pCO2s
xda = xda.assign_attrs(units="uatm")
return xda
def calc_seafrac(process_dest="../data/processed/etopo1_seafrac.nc", ):
from fetch_data import download
from numpy import arange
from xarray import open_mfdataset
fname = download(
url=("https://www.ngdc.noaa.gov/mgg/global/relief/ETOPO1/data/"
"ice_surface/cell_registered/netcdf/ETOPO1_Ice_c_gmt4.grd.gz"),
dest="../data/raw/",
verbose=True,
)
ds = open_mfdataset(fname).rename(x="lon", y="lat", z="topography")
sea = ds.topography < 0
seafrac = sea.coarsen(lat=60, lon=60).sum().compute() / 60**2
seafrac = seafrac.assign_coords(lat=arange(-89.5, 90),
lon=arange(-179.5, 180)).rename("seafrac")
seafrac.attrs = dict(
description=
"Fraction of pixel that is covered by ocean. Calculated from ETOPO1. ",
unit="frac",
)
seafrac.to_netcdf(process_dest)
return process_dest
def get_nies_fnn(entry):
"""processes data"""
from warnings import filterwarnings
from fetch_data import read_catalog
from ..fco2_pco2_conversion import fCO2_to_pCO2
from .aux_vars import download_era5_slp, download_sst_ice
from .utils import add_history
filterwarnings("ignore", category=RuntimeWarning)
def decode_time(xds):
"""processes data"""
import pandas as pd
from datetime_matcher import DatetimeMatcher
re_date = DatetimeMatcher()
fname = xds.encoding["source"]
datetime = re_date.extract_datetime("flux.%Y.ver", fname)
year = pd.Timestamp(datetime).year
y0, y1 = str(year), str(year + 1)
time = pd.date_range(y0, y1, freq="1MS", closed="left")
xds = xds.rename(month="time").assign_coords(time=time)
xds = add_history(xds, "decode times manually")
return xds
flist = download(**entry)
xda = xr.open_mfdataset(flist, preprocess=preprocess(decode_time)).fco2
aux_cat = read_catalog("../data/aux_data.yml")
t0, t1 = [str(s) for s in xda.time.values[[0, -1]]]
sst = xr.open_dataset(download_sst_ice(
aux_cat["oisst_v2"]))["sst"].sel(time=slice(t0, t1))
msl = xr.open_dataset(
download_era5_slp())["sp"].sel(time=slice(t0, t1)) / 100
pco2 = xr.DataArray(
fCO2_to_pCO2(xda, sst, msl),
coords=xda.coords,
dims=xda.dims,
attrs=dict(units="uatm", source=entry["url"], **entry["meta"]),
)
pco2 = add_history(
pco2,
"re-shaped data from [year month lat lon] to [time lat lon].")
pco2 = add_history(
pco2, "converted fCO2 to pCO2 using OISST v2.1, and ERA5 MSLP")
return pco2
def get_seamask(entry):
"""processes data"""
flist = download(**entry)
xds = xr.open_mfdataset(flist, preprocess=preprocess())
xda = xds.seamask.assign_attrs(**entry["meta"])
return xda
def get_csir_ml6(entry):
"""processes data"""
flist = download(**entry)
xds = xr.open_mfdataset(flist, preprocess=preprocess())
xds = xds["spco2"].assign_attrs(units="uatm",
source=entry["url"],
**entry["meta"])
return xds
def get_mpi_somffn(entry):
"""processes data"""
flist = download(**entry)
xda = xr.open_mfdataset(flist, drop_variables="date").spco2_raw
xda = xda.rename("mpi_somffn").assign_attrs(units="uatm",
source=entry["url"],
**entry["meta"])
xda = preprocess()(xda)
return xda
def get_mpi_ulb_somffn(entry):
"""processes data"""
flist = download(**entry)
xds = xr.open_mfdataset(flist)
xda = xds.pco2.where(xds.pco2 > 0).coarsen(lat=4, lon=4).mean()
xda = xda.rename("mpiulb_somffn").rename(time="month")
pp = preprocess(rename_coordinates=False, center_months=False)
xda = pp(xda)
return xda
def get_cmems_ffnn(entry):
"""processes data"""
flist = download(**entry, n_jobs=8)
xds = xr.open_mfdataset(flist, combine="nested", concat_dim="time")
xda = ((xds.spco2 * 9.867).assign_coords(
longitude=(xds.longitude - 180) % 360 - 180).rename(
latitude="lat", longitude="lon").resample(
time="1MS").mean().sortby("lon").assign_attrs(
units="uatm", source=entry["url"], **entry["meta"]))
xda = preprocess()(xda)
return xda
def process_item(item):
if not item.startswith('http'):
return item + '\n'
# the server will give EVERYTHING if empty
if item.endswith('='):
return '[{0}Format{2}] {0}{1}{2}\n'.format(colorama.Fore.RED, item, colorama.Style.RESET_ALL)
try:
data = fetch_data.download(item)
except:
return '[{0}Retrive{2}] {0}{1}{2}\n'.format(colorama.Fore.RED, item, colorama.Style.RESET_ALL)
result = parse_data(data)
if result == '[404]':
result = '[{0}Not found{2}] {0}{1}{2}\n'.format(colorama.Fore.RED, item, colorama.Style.RESET_ALL)
return result
def download_salinity(
catalog_entry,
verbose=True,
process_dest="../data/processed/en4_salt_temp.nc",
):
"""Downloads salinity from MetOffice for 1982 until today"""
import xarray as xr
from fetch_data import download
from .utils import preprocess
if path(process_dest).is_file():
return process_dest
flist = download(**catalog_entry, verbose=verbose)
ds = preprocess()(xr.open_mfdataset(paths=flist)[["salinity"]].sel(
depth=0, method="nearest").drop("depth"))
encode = {k: dict(zlib=True, complevel=4) for k in ds}
ds.load().to_netcdf(process_dest, encoding=encode)
return process_dest
def download_sst_ice(
catalog_entry,
process_dest="../data/processed/noaa_oisst_sst_icec.nc",
):
"""Downloads OISSTv2 data from NOAA"""
import xarray as xr
from fetch_data import download
from .utils import preprocess
if path(process_dest).is_file():
return process_dest
flist = download(**catalog_entry)
ds = (xr.open_mfdataset(paths=flist, preprocess=preprocess()).where(
lambda a: a.icec.notnull()).drop("time_bnds"))
ds.to_netcdf(process_dest,
encoding={k: dict(zlib=True, complevel=4)
for k in ds})
return process_dest
def get_jra55_wind_speed(
url="leave empty - replaced in function", # for transparency
download_dest="../data/raw/jra_55/{file_format}/{year}",
process_dest="../data/processed/jra55_wind_speed_moments.nc",
years=range(1982, 2021),
verbose=False,
n_jobs=8,
):
"""
TODO: add readme to netCDF folder.
TODO: add readme to
"""
from pathlib import Path as path
from dask.diagnostics import ProgressBar
from fetch_data import download
from fetch_data.core import create_download_readme
from fetch_data.utils import commong_substring
from pandas import Timestamp
from xarray import concat
years = list(years)
process_dest = p = path(process_dest)
process_dest = p.parent / f"{p.stem}_{years[0]}-{years[-1]}{p.suffix}"
if path(process_dest).is_file():
return process_dest
else:
print(f"File does not exist: {process_dest}")
cookies = RDAMScookies().get_cookies()
grib_names = []
for y in years:
t0 = Timestamp(f"{y}")
t1 = Timestamp(f"{y+1}")
grib_names += download(
# JRA URLs switch from annual to monthly in 2014
url=make_jra_6hrly_urls(t0=t0, t1=t1),
dest=download_dest.format(
year=y, file_format="grib"), # store the data per year
login=dict(cookies=cookies),
verbose=verbose,
n_jobs=n_jobs,
log_name="../downloading.log",
readme_fname="../README.txt",
meta=jra_meta,
)
# replace the path '/grib/' with netcdf for he conversion
netcdf_names = [
f.replace("/grib/", "/netcdf/") + ".nc" for f in grib_names
]
# the function grib_to_netcdf has been made to run in parallel with the decorator
flist = grib_to_netcdf(grib_names, netcdf_names, n_jobs=n_jobs)
jra_meta["processing"] = (
"Data has been converted from grib file format to netCDF4 using the cfgrib "
"package. Variables without dimensions have been dropped. ")
jra_meta["variables"] = "u10, v10"
jra_meta["grib_source"] = download_dest.format(year="YYYY",
file_format="grib")
create_download_readme(
"README.md",
url=commong_substring(grib_names) + "...",
dest=str(
path(download_dest.format(year="YYYY",
file_format="netcdf")).parent),
meta=jra_meta,
)
# we get the folders for each year
folders = sorted(list(set([path(f).parent for f in flist])))
xds = []
for folder in folders:
# list nc files - assumes u10 and v10 in the folder
ylist = list(folder.glob("*.nc"))
xds += (calculate_wind_speed(ylist), )
with ProgressBar():
wind_speed = concat(xds, "time").load()
process_dest = path(process_dest)
process_dest.parent.mkdir(exist_ok=True, parents=True)
jra_meta["processing"] = (
"Data has been converted from grib file format to netCDF4)."
"u10 and v10 data was loaded and the wind_speed was calculated with "
"(u10^2 + v10^2)^0.5. The first, second, and third moments "
"(wind_speed^n) were calculated from wind_speed. Note that these "
"variables were calculated at the model resolution and then scaled "
"up to monthly by 1 degree to preserve the variability that would "
"otherwise be lost in the squared function. ")
jra_meta["variables"] = "wind_speed, wind_speed^2, wind_speed^3"
jra_meta["netcdf_source"] = download_dest.format(year="YYYY",
file_format="netcdf")
wind_speed.attrs = jra_meta
wind_speed.to_netcdf(
str(process_dest),
encoding={k: {
"complevel": 4,
"zlib": True
}
for k in wind_speed},
)
return str(process_dest)
