Skip to main content
 首页 » 编程设计

python xarray 按纬度/经度选择并将点数据提取到数据帧

2025年04月02日12Free-Thinker

我想选择纬度/经度范围内的所有网格单元格,对于每个网格单元格,将其导出为日期框,然后导出到 csv 文件(即 df.to_csv)。我的数据集如下。我可以使用 xr.where(...) 屏蔽掉我输入之外的网格单元,但不确定如何遍历未屏蔽掉的剩余网格。或者,我尝试使用 xr.sel 函数,但它们似乎不接受像 ds.sel(gridlat_0>45) 这样的运算符。 xr.sel_points(...) 也可能有效,但我无法找出在我的案例中使用的索引器的正确语法。提前感谢您的帮助。

<xarray.Dataset> 
Dimensions:    (time: 48, xgrid_0: 685, ygrid_0: 485) 
Coordinates: 
    gridlat_0  (ygrid_0, xgrid_0) float32 44.6896 44.6956 44.7015 44.7075 ... 
  * ygrid_0    (ygrid_0) int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 ... 
  * xgrid_0    (xgrid_0) int64 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 ... 
  * time       (time) datetime64[ns] 2016-07-28T01:00:00 2016-07-28T02:00:00 ... 
    gridlon_0  (ygrid_0, xgrid_0) float32 -129.906 -129.879 -129.851 ... 
Data variables: 
    u          (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 
    gridrot_0  (time, ygrid_0, xgrid_0) float32 nan nan nan nan nan nan nan ... 
    Qli        (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 
    Qsi        (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 
    p          (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 
    rh         (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 
    press      (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 
    t          (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 
    vw_dir     (time, ygrid_0, xgrid_0) float64 nan nan nan nan nan nan nan ... 

请您参考如下方法:

执行此操作的最简单方法可能是循环遍历每个网格点,如下所示:

# (optionally) create a grid dataset so we don't need to pull out all 
# the data from the main dataset before looking at each point 
grid = ds[['gridlat_0', 'gridlon_0']] 
 
for i in range(ds.coords['xgrid_0'].size): 
    for j in range(ds.coords['ygrid_0'].size): 
        sub_grid = grid.isel(xgrid_0=i, ygrid_0=j) 
        if is_valid(sub_grid.gridlat_0, sub_grid.gridlon_0): 
            sub_ds = ds.isel(xgrid_0=i, ygrid_0=j) 
            sub_ds.to_dataframe().to_csv(...) 

即使是 685x485,循环遍历每个点也应该只需要几秒钟。

使用 ds = ds.where(..., drop=True) 进行预过滤(在下一个 xarray 版本中可用,将于本周晚些时候发布)可以显着加快速度,但您仍然会遇到可能无法在正交轴上表示所选网格的问题。

最后一个选项(可能是最简洁的选项)是使用 stack 将数据集转换为 2D。然后,您可以沿新的 'space' 维度使用标准选择和分组操作:

ds_stacked = ds.stack(space=['xgrid_0', 'ygrid_0']) 
ds_filtered = ds_stacked.sel(space=(ds_stacked.gridlat_0 > 45)) 
for _, ds_one_place in ds_filtered.groupby('space'): 
    ds_one_place.to_dataframe().to_csv(...)