I see you use pandas to read from dbfs.

But pandas will only read from local files,

see this topic also. It is about databricks-connect but the same principles apply.

So what you should do is first read the file using spark.read.csv and then converting the spark df to a pandas df.

I had the same issue: geopandas in Databricks notebooks does not open shapefiles from an Azure Storage mount.
I managed to copy the shapefile to the Databricks workspace using 

dbutils.fs.cp(shapefile_path, f"file:{local_shapefile_copy_dest_path}") 

The 'file:' prefix proved to be crucial here.

and then: 

gdf = gpd.read_file(shapefile_path.replace('dbfs:', ''))
gdf.display()

I copy the results back to the dbfs mount using

dbutils.fs.cp(f"file:{geoparquet_path}", f"{raw_path}{geoparquet_file_basename}.parquet")

---

@-werners- wrote:

I see you use pandas to read from dbfs.

But pandas will only read from local files,

see this topic also. It is about databricks-connect but the same principles apply.

So what you should do is first read the file using spark.read.csv and then converting the spark df to a pandas df.

---

---

@-werners- wrote:

I see you use pandas to read from dbfs.

But pandas will only read from local files,

see this topic also. It is about databricks-connect but the same principles apply.

So what you should do is first read the file using spark.read.csv and then converting the spark df to a pandas df.

---

Ik convert to parquet using

gdf.to_parquet(f"/dbfs{raw_path}/{file_name}.parquet")