Geopandas读取自制GeoJSON文件失败问题求助

问题重现 #

执行以下代码导出并读取GeoJSON时触发DataSourceError：

import geopandas as gpd

# 读取shp文件
vec_data = gpd.read_file("map.shp") 
vec_data.head()
vec_data['LPIS_name'].unique()

# 筛选永久草地并导出为GeoJSON
sel_crop = vec_data[vec_data.LPIS_name == 'Permanent Grassland']
sel_crop.to_file("Permanent_Grassland.geojson", driver='GeoJSON')

# 读取生成的GeoJSON时出错
feature = gpd.read_file("Permanent_Grassland.geojson")

完整错误信息 #

---------------------------------------------------------------------------
DataSourceError                           Traceback (most recent call last)
Cell In[8], line 1
----> 1 feature = gpd.read_file(path_feature)

File c:\\Users\\bventura\\AppData\\Local\\anaconda3\\Lib\\site-packages\\geopandas\\io\\file.py:294, in _read_file(filename, bbox, mask, columns, rows, engine, **kwargs)
    291             from_bytes = True
    293 if engine == "pyogrio":
---> 294     return _read_file_pyogrio(
    295         filename, bbox=bbox, mask=mask, columns=columns, rows=rows, **kwargs
    296     )
    298 elif engine == "fiona":
    299     if pd.api.types.is_file_like(filename):

File c:\\Users\\bventura\\AppData\\Local\\anaconda3\\Lib\\site-packages\\geopandas\\io\\file.py:547, in _read_file_pyogrio(path_or_bytes, bbox, mask, rows, **kwargs)
    538     warnings.warn(
    539         "The 'include_fields' and 'ignore_fields' keywords are deprecated, and "
    540         "will be removed in a future release. You can use the 'columns' keyword "
   (...)
    543         stacklevel=3,
    544     )
    545     kwargs["columns"] = kwargs.pop("include_fields")
---> 547 return pyogrio.read_dataframe(path_or_bytes, bbox=bbox, **kwargs)

File c:\\Users\\bventura\\AppData\\Local\\anaconda3\\Lib\\site-packages\\pyogrio\\geopandas.py:261, in read_dataframe(path_or_buffer, layer, encoding, columns, read_geometry, force_2d, skip_features, max_features, where, bbox, mask, fids, sql, sql_dialect, fid_as_index, use_arrow, on_invalid, arrow_to_pandas_kwargs, **kwargs)
    256 if not use_arrow:
    257     # For arrow, datetimes are read as is.
    258     # For numpy IO, datetimes are read as string values to preserve timezone info
    259     # as numpy does not directly support timezones.
    260     kwargs["datetime_as_string"] = True
---> 261 result = read_func(
    262     path_or_buffer,
    263     layer=layer,
    264     encoding=encoding,
    265     columns=columns,
    266     read_geometry=read_geometry,
    267     force_2d=gdal_force_2d,
    268     skip_features=skip_features,
    269     max_features=max_features,
    270     where=where,
    271     bbox=bbox,
    272     mask=mask,
    273     fids=fids,
    274     sql=sql,
    275     sql_dialect=sql_dialect,
    276     return_fids=fid_as_index,
    277     **kwargs,
    278 )
    280 if use_arrow:
    281     meta, table = result

File c:\\Users\\bventura\\AppData\\Local\\anaconda3\\Lib\\site-packages\\pyogrio\\raw.py:196, in read(path_or_buffer, layer, encoding, columns, read_geometry, force_2d, skip_features, max_features, where, bbox, mask, fids, sql, sql_dialect, return_fids, datetime_as_string, **kwargs)
     56 """Read OGR data source into numpy arrays.
     57 
     58 IMPORTANT: non-linear geometry types (e.g., MultiSurface) are converted
   (...)
    191 
    192 """
    194 dataset_kwargs = _preprocess_options_key_value(kwargs) if kwargs else {}
---> 196 return ogr_read(
    197     get_vsi_path_or_buffer(path_or_buffer),
    198     layer=layer,
    199     encoding=encoding,
    200     columns=columns,
    201     read_geometry=read_geometry,
    202     force_2d=force_2d,
    203     skip_features=skip_features,
    204     max_features=max_features or 0,
    205     where=where,
    206     bbox=bbox,
    207     mask=_mask_to_wkb(mask),
    208     fids=fids,
    209     sql=sql,
    210     sql_dialect=sql_dialect,
    211     return_fids=return_fids,
    212     dataset_kwargs=dataset_kwargs,
    213     datetime_as_string=datetime_as_string,
    214 )

File c:\\Users\\bventura\\AppData\\Local\\anaconda3\\Lib\\site-packages\\pyogrio\\_io.pyx:1239, in pyogrio._io.ogr_read()

File c:\\Users\\bventura\\AppData\\Local\\anaconda3\\Lib\\site-packages\\pyogrio\\_io.pyx:219, in pyogrio._io.ogr_open()

DataSourceError: Failed to read GeoJSON data

怀疑原因 #

问题大概率出在Polygon和MultiPolygon未遵循右手定则——OGR/GeoJSON规范要求多边形顶点按右手定则排列（外环逆时针，内环顺时针），不符合该规则的几何会导致读取失败。

解决方案 #

方案1：修复几何顶点顺序（推荐） #

导出GeoJSON前，通过shapely的orient函数统一调整顶点顺序，使其符合右手定则：

import geopandas as gpd
from shapely.geometry.polygon import orient

# 读取shp文件
vec_data = gpd.read_file("map.shp") 

# 筛选目标数据
sel_crop = vec_data[vec_data.LPIS_name == 'Permanent Grassland']

# 修复几何顶点顺序，sign=1.0表示外环逆时针
sel_crop['geometry'] = sel_crop['geometry'].apply(lambda geom: orient(geom, sign=1.0))

# 导出为GeoJSON
sel_crop.to_file("Permanent_Grassland.geojson", driver='GeoJSON')

# 验证读取
feature = gpd.read_file("Permanent_Grassland.geojson")
feature.head()

方案2：跳过无效几何（临时应急） #

若无需修复几何，可在读取时指定on_invalid='skip'参数跳过无效条目，但这会丢失部分数据，仅作为临时方案：

feature = gpd.read_file("Permanent_Grassland.geojson", on_invalid='skip')

内容的提问来源于stack exchange，提问作者sylar_80