ERA-20c vs CERA-20c¶

We compare the temperature data from ERA-20c and CERA-20c.

TDLR; gobal trends are similar but ERA-20c has spurious (cold) temperatures in many parts of the world.

In [1]:

import xarray as xr
import cartopy
import cartopy.crs as ccrs
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd

In [2]:

from oggm import utils

In [3]:

t_c20c = utils.file_downloader('https://cluster.klima.uni-bremen.de/~fmaussion/climate/cera-20c/cera-20c_t2m_1901-2010.nc')
t_20c = utils.file_downloader('https://cluster.klima.uni-bremen.de/~fmaussion/climate/era-20c/era-20c_t2m_1901-2010.nc')

In [4]:

ds_c20c = xr.open_dataset(t_c20c)
ds_20c = xr.open_dataset(t_20c)

In [5]:

ds_c20c

Out[5]:

<xarray.Dataset>
Dimensions:    (latitude: 181, longitude: 360, number: 10, time: 1320)
Coordinates:
  * longitude  (longitude) float32 0.0 1.0 2.0 3.0 ... 356.0 357.0 358.0 359.0
  * latitude   (latitude) float32 90.0 89.0 88.0 87.0 ... -88.0 -89.0 -90.0
  * number     (number) int32 0 1 2 3 4 5 6 7 8 9
  * time       (time) datetime64[ns] 1901-01-01 1901-02-01 ... 2010-12-01
Data variables:
    t2m        (time, number, latitude, longitude) float32 ...
Attributes:
    Conventions:  CF-1.6
    history:      2019-05-26 18:33:08 GMT by grib_to_netcdf-2.12.0: grib_to_n...

In [6]:

ds_20c

Out[6]:

<xarray.Dataset>
Dimensions:    (latitude: 181, longitude: 360, time: 1320)
Coordinates:
  * longitude  (longitude) float32 0.0 1.0 2.0 3.0 ... 356.0 357.0 358.0 359.0
  * latitude   (latitude) float32 90.0 89.0 88.0 87.0 ... -88.0 -89.0 -90.0
  * time       (time) datetime64[ns] 1901-01-01 1901-02-01 ... 2010-12-01
Data variables:
    d2m        (time, latitude, longitude) float32 ...
Attributes:
    Conventions:  CF-1.6
    history:      2019-05-26 15:36:48 GMT by grib_to_netcdf-2.12.0: grib_to_n...

In [7]:

ds_c20c = ds_c20c.chunk(chunks={'number':1, 'longitude':180})

In [8]:

fig = plt.figure(figsize=(12, 5))
ax = plt.axes(projection=ccrs.Robinson())
(ds_20c.d2m.mean(dim='time') - 273.15).plot(ax=ax, transform=ccrs.PlateCarree()) 
ax.coastlines(); ax.gridlines(); plt.title('ERA-20C average temperature');

In [9]:

fig = plt.figure(figsize=(12, 5))
ax = plt.axes(projection=ccrs.Robinson())
(ds_c20c.t2m.mean(dim=['time', 'number']) - 273.15).plot(ax=ax, transform=ccrs.PlateCarree()) 
ax.coastlines(); ax.gridlines(); plt.title('CERA-20C average temperature');

In [10]:

fig = plt.figure(figsize=(12, 5))
ax = plt.axes(projection=ccrs.Robinson())
(ds_20c.d2m.mean(dim='time') - ds_c20c.t2m.mean(dim=['time', 'number'])).plot(ax=ax, transform=ccrs.PlateCarree()) 
ax.coastlines(); ax.gridlines(); plt.title('Diff ERA-20C - CERA-20C');

In [11]:

# Weight 
weight = np.cos(np.deg2rad(ds_c20c.latitude))
weight =  ds_20c.d2m.mean(dim='time') * 0. + weight
weight = weight / weight.sum()

In [12]:

ts_c20c = (ds_c20c.t2m.mean(dim=['number']) * weight).sum(dim=['latitude','longitude']) - 273.15
ts_20c = (ds_20c.d2m * weight).sum(dim=['latitude','longitude']) - 273.15

In [13]:

ts_c20c_a = ts_c20c.resample(time='AS').mean(dim='time')
ts_20c_a = ts_20c.resample(time='AS').mean(dim='time')

In [14]:

ts_c20c_a.plot(label='CERA-20c');
ts_20c_a.plot(label='ERA-20c');
plt.legend();

In [15]:

(ts_c20c_a - ts_c20c_a.sel(time=slice('1961', '1990')).mean(dim='time')).plot(label='CERA-20c');
(ts_20c_a - ts_20c_a.sel(time=slice('1961', '1990')).mean(dim='time')).plot(label='ERA-20c');
plt.legend();

In [16]:

ts_c20c_members = (ds_c20c.t2m * weight).sum(dim=['latitude','longitude']) - 273.15

In [17]:

fig = plt.figure(figsize=(12, 8))
ts_c20c_members.resample(time='AS').mean(dim='time').plot.line(x='time');

In [ ]: