文章/答案/技术大牛

发布

社区首页 >专栏 >如何将不同分辨率的ERA5数据插值为统一网格：CDO与xesmf方法

如何将不同分辨率的ERA5数据插值为统一网格：CDO与xesmf方法

用户11172986

发布于 2025-05-10 02:40:15

11500

代码可运行

文章被收录于专栏：气python风雨气python风雨

运行总次数：0

代码可运行

如何将不同分辨率的ERA5数据插值为统一网格：CDO与xesmf方法

前言

当处理气象数据时，经常会遇到不同分辨率的数据集需要一起使用的情况。

小编在气象家园看到如下帖子：

下了两份ERA5资料，一份的分辨率是0.250.25，另一份地面气压分辨率是0.10.1，不同分辨率没办法一起计算，求助怎样转分辨率并输出文件？

本文将介绍两种方法来统一ERA5数据的分辨率：使用CDO命令行工具和Python的xesmf库。

1. 创建样例数据

因为手头没有现成的数据，直接制造文件是比较快捷的

首先，我们使用xarray创建两个不同分辨率的样例数据集并保存为NetCDF文件。

import xarray as xr
import numpy as np

# 创建0.25°分辨率的样例数据
lat_025 = np.arange(-90, 90.25, 0.25)
lon_025 = np.arange(0, 360, 0.25)
data_025 = np.random.rand(len(lat_025), len(lon_025))

ds_025 = xr.Dataset(
    {"temperature": (["latitude", "longitude"], data_025)},
    coords={
        "latitude": lat_025,
        "longitude": lon_025,
    },
)
ds_025.to_netcdf("era5_025deg.nc")

# 创建0.1°分辨率的样例数据
lat_01 = np.arange(-90, 90.1, 0.1)
lon_01 = np.arange(0, 360, 0.1)
data_01 = np.random.rand(len(lat_01), len(lon_01))

ds_01 = xr.Dataset(
    {"surface_pressure": (["latitude", "longitude"], data_01)},
    coords={
        "latitude": lat_01,
        "longitude": lon_01,
    },
)
ds_01.to_netcdf("era5_01deg.nc")

1.1 检查一下生成的文件

1.1.1 ncdump

!ncdump -c /home/mw/project/era5_01deg.nc | head -20

netcdf era5_01deg {

dimensions:

	latitude = 1801 ;

	longitude = 3600 ;

variables:

	double surface_pressure(latitude, longitude) ;

		surface_pressure:_FillValue = NaN ;

	double latitude(latitude) ;

		latitude:_FillValue = NaN ;

	double longitude(longitude) ;

		longitude:_FillValue = NaN ;

data:



 latitude = -90, -89.9, -89.8, -89.7, -89.6, -89.5, -89.4, -89.3, -89.2, 

    -89.1000000000001, -89.0000000000001, -88.9000000000001, 

    -88.8000000000001, -88.7000000000001, -88.6000000000001, 

    -88.5000000000001, -88.4000000000001, -88.3000000000001, 

    -88.2000000000001, -88.1000000000001, -88.0000000000001, 

    -87.9000000000001, -87.8000000000001, -87.7000000000001, 

    -87.6000000000001, -87.5000000000001, -87.4000000000001,

!ncdump -c /home/mw/project/era5_025deg.nc | head -20

netcdf era5_025deg {

dimensions:

	latitude = 721 ;

	longitude = 1440 ;

variables:

	double temperature(latitude, longitude) ;

		temperature:_FillValue = NaN ;

	double latitude(latitude) ;

		latitude:_FillValue = NaN ;

	double longitude(longitude) ;

		longitude:_FillValue = NaN ;

data:



 latitude = -90, -89.75, -89.5, -89.25, -89, -88.75, -88.5, -88.25, -88, 

    -87.75, -87.5, -87.25, -87, -86.75, -86.5, -86.25, -86, -85.75, -85.5, 

    -85.25, -85, -84.75, -84.5, -84.25, -84, -83.75, -83.5, -83.25, -83, 

    -82.75, -82.5, -82.25, -82, -81.75, -81.5, -81.25, -81, -80.75, -80.5, 

    -80.25, -80, -79.75, -79.5, -79.25, -79, -78.75, -78.5, -78.25, -78, 

    -77.75, -77.5, -77.25, -77, -76.75, -76.5, -76.25, -76, -75.75, -75.5, 

    -75.25, -75, -74.75, -74.5, -74.25, -74, -73.75, -73.5, -73.25, -73,

1.1.2 cdo查看文件信息

当然使用cdo查看网格会更简洁（大概）

!cdo griddes /home/mw/project/era5_01deg.nc

#

# gridID 1

#

gridtype  = generic

gridsize  = 6483600

xsize     = 3600

ysize     = 1801

xname     = longitude

yname     = latitude

xfirst    = 0

xinc      = 0.1

yfirst    = -90

yinc      = 0.0999999999999943

[32mcdo    griddes: [0mProcessed 1 variable [0.08s 51MB]

!cdo griddes /home/mw/project/era5_025deg.nc

#

# gridID 1

#

gridtype  = generic

gridsize  = 1038240

xsize     = 1440

ysize     = 721

xname     = longitude

yname     = latitude

xfirst    = 0

xinc      = 0.25

yfirst    = -90

yinc      = 0.25

[32mcdo    griddes: [0mProcessed 1 variable [0.05s 51MB]

2. 使用CDO进行插值

CDO (Climate Data Operators) 是我们的老朋友了，专门用于处理气候和NWP模型数据。对于nc数据可谓是瑞士军刀一般的存在

2.1 获取目标网格描述文件

首先，我们需要获取目标网格的描述文件：

!cdo griddes era5_025deg.nc > grid_025.txt
!cdo griddes era5_01deg.nc > grid_01.txt

这会生成一个描述0.25°网格的文本文件。

!cat /home/mw/project/grid_025.txt

#

# gridID 1

#

gridtype  = generic

gridsize  = 1038240

xsize     = 1440

ysize     = 721

xname     = longitude

yname     = latitude

xfirst    = 0

xinc      = 0.25

yfirst    = -90

yinc      = 0.25

2.2 执行双线性插值

将0.1°分辨率的数据插值到0.25°分辨率：

!cdo remapbil,grid_025.txt era5_01deg.nc era5_01deg_regridded_to_025.nc

[31m

cdo    remapbil (Abort): [0mUnsupported target grid type (generic)!

坏了，cdo不支持这个 grid type (generic)

那么只能手动修改了

2.3 手动修改nc数据坐标属性

!sed -i "s/generic/lonlat/g" /home/mw/project/grid_025.txt
!cat /home/mw/project/grid_025.txt

#

# gridID 1

#

gridtype  = lonlat

gridsize  = 1038240

xsize     = 1440

ysize     = 721

xname     = longitude

yname     = latitude

xfirst    = 0

xinc      = 0.25

yfirst    = -90

yinc      = 0.25

!sed -i "s/generic/lonlat/g" /home/mw/project/grid_01.txt
!cat /home/mw/project/grid_01.txt

#

# gridID 1

#

gridtype  = lonlat

gridsize  = 6483600

xsize     = 3600

ysize     = 1801

xname     = longitude

yname     = latitude

xfirst    = 0

xinc      = 0.1

yfirst    = -90

yinc      = 0.0999999999999943

!cdo setgrid,grid_025.txt era5_025deg.nc era5_025deg_002.nc
!cdo griddes /home/mw/project/era5_025deg_002.nc

[32mcdo    setgrid: [0mProcessed 1038240 values from 1 variable over 1 timestep [0.13s 5096MB]
#
# gridID 1
#
gridtype  = lonlat
gridsize  = 1038240
xsize     = 1440
ysize     = 721
xname     = longitude
xlongname = "longitude"
xunits    = "degrees_east"
yname     = latitude
ylongname = "latitude"
yunits    = "degrees_north"
xfirst    = 0
xinc      = 0.25
yfirst    = -90
yinc      = 0.25
[32mcdo    griddes: [0mProcessed 1 variable [0.03s 5096MB]

!cdo setgrid,grid_01.txt era5_01deg.nc era5_01deg_002.nc
!cdo griddes /home/mw/project/era5_01deg_002.nc

[32mcdo    setgrid: [0mProcessed 6483600 values from 1 variable over 1 timestep [0.37s 5096MB]
#
# gridID 1
#
gridtype  = lonlat
gridsize  = 6483600
xsize     = 3600
ysize     = 1801
xname     = longitude
xlongname = "longitude"
xunits    = "degrees_east"
yname     = latitude
ylongname = "latitude"
yunits    = "degrees_north"
xfirst    = 0
xinc      = 0.1
yfirst    = -90
yinc      = 0.0999999999999943
[32mcdo    griddes: [0mProcessed 1 variable [0.04s 5096MB]

2.4 一句插值

!cdo remapbil,grid_025.txt era5_01deg_002.nc era5_01deg_regridded_to_025.nc

[32mcdo    remapbil: [0mBilinear weights from lonlat (3600x1801) to lonlat (1440x721) grid
cdo    remapbil:                        1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 5 5 6 6 6 6 6 6 6 6 6 6 7 7 7 7 7 7 7 7 7 7 8 8 8 8 8 8 8 8 8 8 9 9 9 9 9 9 9 9 9 91[32mcdo    remapbil: [0mProcessed 6483600 values from 1 variable over 1 timestep [0.92s 5096MB]

这个命令使用双线性插值法(remapbil)将输入文件(era5_01deg.nc)插值到grid_025.txt描述的网格上，并输出新文件。

文件生成并不符合cdo的格式，需要使用set指令调整一下。

插值速度是非常快的，按照官方说法可以达到c语言的程度，可以看到不到一秒就完成了

3. 使用Python的xesmf进行插值

xesmf是一个基于ESMF的Python库，专门用于地球科学数据的重网格化。

3.1 安装xesmf

pip install xesmf

3.2 执行重网格化

import xarray as xr
import xesmf as xe

# 加载数据
ds_target = xr.open_dataset("era5_025deg.nc")  # 目标网格
ds_source = xr.open_dataset("era5_01deg.nc")   # 源数据

# 创建重网格器
regridder = xe.Regridder(ds_source, ds_target, "bilinear")

# 执行重网格化
ds_regridded = regridder(ds_source)

# 保存结果
ds_regridded.to_netcdf("era5_01deg_regridded_to_025_xesmf.nc")

!cdo griddes /home/mw/project/era5_01deg_regridded_to_025_xesmf.nc

#

# gridID 1

#

gridtype  = generic

gridsize  = 1038240

xsize     = 1440

ysize     = 721

xname     = longitude

yname     = latitude

xfirst    = 0

xinc      = 0.25

yfirst    = -90

yinc      = 0.25

[32mcdo    griddes: [0mProcessed 1 variable [0.04s 5096MB]