diff --git a/CHANGELOG.md b/CHANGELOG.md
index 047b2012..490c0dfa 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,10 +9,29 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 
 ### Added
 
+- Command `fiboa improve` with helpers to
+  - change the CRS
+  - change the GeoParquet version and compression
+  - fill missing perimeter/area values
+  - fix invalid geometries
+  - rename columns
 - Converter for Lithuania (EuroCrops)
-- Converter for Switzerland
 - Converter for Slovenia
 - Converter for Slovakia
+- Converter for Switzerland
+- `fiboa convert`: New parameter `--original-geometries` / `-og` to keep the original geometries
+
+### Changed
+
+- `fiboa convert`:
+  - Writes custom schemas to collection metadata
+  - Geometries are made valid using GeoPanda's `make_valid` method by default
+  - MultiPolygons are converted to Polygons by default
+- `fiboa validate` uses custom schemas for validation
+- `fiboa merge` keeps custom schemas when needed
+
+### Removed
+- `fiboa convert`: Removed the explicit parameter `explode_multipolygon` from the converter
 
 ### Fixed
 
diff --git a/README.md b/README.md
index cc9f6cff..31699f62 100644
--- a/README.md
+++ b/README.md
@@ -8,12 +8,12 @@ A command-line interface (CLI) for working with fiboa.
 
 ## Getting Started
 
-In order to make working with fiboa easier we have developed command-line interface (CLI) tools such as 
+In order to make working with fiboa easier we have developed command-line interface (CLI) tools such as
 inspection, validation and file format conversions.
 
 ### Installation
 
-You will need to have **Python 3.9** or any later version installed. 
+You will need to have **Python 3.9** or any later version installed.
 
 Run `pip install fiboa-cli` in the CLI to install the validator.
 
@@ -21,6 +21,7 @@ Run `pip install fiboa-cli` in the CLI to install the validator.
 you can for example run: `pip install fiboa-cli[xyz]` with xyz being the converter name.
 
 **Note on versions:**
+
 - fiboa CLI >= 0.3.0 works with fiboa version > 0.2.0
 - fiboa CLI < 0.3.0 works with fiboa version = 0.1.0
 
@@ -44,6 +45,7 @@ fiboa CLI supports various commands to work with the files:
     - [Merge fiboa GeoParquet files](#merge-fiboa-geoparquet-files)
     - [Create JSON Schema from fiboa Schema](#create-json-schema-from-fiboa-schema)
     - [Validate a fiboa Schema](#validate-a-fiboa-schema)
+    - [Improve a fiboa Parquet file](#improve-a-fiboa-parquet-file)
     - [Update an extension template with new names](#update-an-extension-template-with-new-names)
     - [Converter for existing datasets](#converter-for-existing-datasets)
   - [Development](#development)
@@ -121,19 +123,38 @@ To validate a fiboa Schema YAML file, you can for example run:
 
 Check `fiboa validate-schema --help` for more details.
 
+### Improve a fiboa Parquet file
+
+Various "improvements" can be applied to a fiboa GeoParquet file.
+The commands allows to
+
+- change the CRS (`--crs`)
+- change the GeoParquet version (`-gp1`) and compression (`-pc`)
+- add/fill missing perimeter/area values (`-sz`)
+- fix invalid geometries (`-g`)
+- rename columns (`-r`)
+
+Example:
+
+- `fiboa improve file.parquet -o file2.parquet -g -sz -r old=new -pc zstd`
+
+Check `fiboa improve --help` for more details.
+
 ### Update an extension template with new names
 
 Once you've created and git cloned a new extension, you can use the CLI
 to update all template placeholders with proper names.
 
 For example, if your extension is meant to have
-- the title "Timestamps Extension", 
+
+- the title "Timestamps Extension",
 - the prefix `ts` (e.g. field `ts:created` or `ts:updated`),
 - is hosted at `https://github.io/fiboa/timestamps-extension`
   (organization: `fiboa`, repository `timestamps-extension`),
 - and you run fiboa in the folder of the extension.
 
 Then the following command could be used:
+
 - `fiboa rename-extension . -t Timestamps -p ts -s timestamps-extension -o fiboa`
 
 Check `fiboa rename-extension --help` for more details.
@@ -143,13 +164,14 @@ Check `fiboa rename-extension --help` for more details.
 The CLI ships various converters for existing datasets.
 
 To get a list of available converters/datasets with title, license, etc. run:
+
 - `fiboa converters`
 
 Use any of the IDs from the list to convert an existing dataset to fiboa:
 
 - `fiboa convert de_nrw`
 
-See [Implement a converter](#implement-a-converter) for details about how to 
+See [Implement a converter](#implement-a-converter) for details about how to
 
 ## Development
 
diff --git a/fiboa_cli/__init__.py b/fiboa_cli/__init__.py
index 45731ce4..75ad59a8 100644
--- a/fiboa_cli/__init__.py
+++ b/fiboa_cli/__init__.py
@@ -6,16 +6,18 @@
 import click
 import pandas as pd
 
+from .const import COMPRESSION_METHODS, CORE_COLUMNS
 from .convert import convert as convert_
 from .convert import list_all_converter_ids, list_all_converters
 from .create_geojson import create_geojson as create_geojson_
 from .create_geoparquet import create_geoparquet as create_geoparquet_
 from .describe import describe as describe_
-from .merge import merge as merge_, DEFAULT_COLUMNS, DEFAULT_CRS
+from .improve import improve as improve_
+from .merge import merge as merge_, DEFAULT_CRS
 from .jsonschema import jsonschema as jsonschema_
 from .rename_extension import rename_extension as rename_extension_
 from .util import (check_ext_schema_for_cli, log, parse_converter_input_files,
-                   valid_file_for_cli, valid_file_for_cli_with_ext,
+                   parse_map, valid_file_for_cli, valid_file_for_cli_with_ext,
                    valid_files_folders_for_cli, valid_folder_for_cli)
 from .validate import validate as validate_
 from .validate_schema import validate_schema as validate_schema_
@@ -376,7 +378,7 @@ def jsonschema(schema, out, fiboa_version, id_):
 )
 @click.option(
     '--compression', '-pc',
-    type=click.Choice(["brotli", "gzip", "lz4", "snappy", "zstd", "none"]),
+    type=click.Choice(COMPRESSION_METHODS),
     help='Compression method for the Parquet file.',
     show_default=True,
     default="brotli"
@@ -385,7 +387,7 @@ def jsonschema(schema, out, fiboa_version, id_):
     '--geoparquet1', '-gp1',
     is_flag=True,
     type=click.BOOL,
-    help='Enforces generating a GeoParquet 1.0 file bounding box. Defaults to GeoParquet 1.1 with bounding box.',
+    help='Enforces generating a GeoParquet 1.0 file. Defaults to GeoParquet 1.1 with bounding box.',
     default=False
 )
 @click.option(
@@ -394,13 +396,20 @@ def jsonschema(schema, out, fiboa_version, id_):
     help='Url of mapping file. Some converters use additional sources with mapping data.',
     default=None
 )
-def convert(dataset, out, input, cache, source_coop, collection, compression, geoparquet1, mapping_file):
+@click.option(
+    '--original-geometries', '-og',
+    is_flag=True,
+    type=click.BOOL,
+    help='Keep the source geometries as provided, i.e. this option disables that geomtries are made valid and converted to Polygons.',
+    default=False
+)
+def convert(dataset, out, input, cache, source_coop, collection, compression, geoparquet1, mapping_file, original_geometries):
     """
     Converts existing field boundary datasets to fiboa.
     """
     log(f"fiboa CLI {__version__} - Convert '{dataset}'\n", "success")
     try:
-        convert_(dataset, out, input, cache, source_coop, collection, compression, geoparquet1, mapping_file)
+        convert_(dataset, out, input, cache, source_coop, collection, compression, geoparquet1, mapping_file, original_geometries)
     except Exception as e:
         log(e, "error")
         sys.exit(1)
@@ -518,7 +527,7 @@ def rename_extension(folder, title, slug, org = "fiboa", prefix = None):
     multiple=True,
     help='Additional column names to include.',
     show_default=True,
-    default=DEFAULT_COLUMNS,
+    default=CORE_COLUMNS,
 )
 @click.option(
     '--exclude', '-e',
@@ -536,7 +545,7 @@ def rename_extension(folder, title, slug, org = "fiboa", prefix = None):
 )
 @click.option(
     '--compression', '-pc',
-    type=click.Choice(["brotli", "gzip", "lz4", "snappy", "zstd", "none"]),
+    type=click.Choice(COMPRESSION_METHODS),
     help='Compression method for the Parquet file.',
     show_default=True,
     default="brotli"
@@ -545,7 +554,7 @@ def rename_extension(folder, title, slug, org = "fiboa", prefix = None):
     '--geoparquet1', '-gp1',
     is_flag=True,
     type=click.BOOL,
-    help='Enforces generating a GeoParquet 1.0 file bounding box. Defaults to GeoParquet 1.1 with bounding box.',
+    help='Enforces generating a GeoParquet 1.0 file. Defaults to GeoParquet 1.1 with bounding box.',
     default=False
 )
 def merge(datasets, out, crs, include, exclude, extension, compression, geoparquet1):
@@ -564,6 +573,76 @@ def merge(datasets, out, crs, include, exclude, extension, compression, geoparqu
         sys.exit(1)
 
 
+## IMPROVE (add area, perimeter, and fix geometries)
+@click.command()
+@click.argument('input', nargs=1, type=click.Path(exists=True))
+@click.option(
+    '--out', '-o',
+    type=click.Path(exists=False),
+    help='Path to write the GeoParquet file to. If not given, overwrites the input file.',
+    default=None
+)
+@click.option(
+    '--rename-column', '-r',
+    type=click.STRING,
+    callback=lambda ctx, param, value: parse_map(value),
+    multiple=True,
+    help='Renaming of columns. Provide the old name and the new name separated by an equal sign. Can be used multiple times.'
+)
+@click.option(
+    '--add-sizes', '-sz',
+    is_flag=True,
+    type=click.BOOL,
+    help='Computes missing sizes (area, perimeter)',
+    default=False
+)
+@click.option(
+    '--fix-geometries', '-g',
+    is_flag=True,
+    type=click.BOOL,
+    help='Tries to fix invalid geometries that are repored by the validator (uses GeoPanda\'s make_valid method internally)',
+    default=False
+)
+@click.option(
+    '--explode-geometries', '-e',
+    is_flag=True,
+    type=click.BOOL,
+    help='Converts MultiPolygons to Polygons',
+    default=False
+)
+@click.option(
+    '--crs',
+    type=click.STRING,
+    help='Coordinate Reference System (CRS) to use for the GeoParquet file.',
+    show_default=True,
+    default=None,
+)
+@click.option(
+    '--compression', '-pc',
+    type=click.Choice(COMPRESSION_METHODS),
+    help='Compression method for the Parquet file.',
+    show_default=True,
+    default="brotli"
+)
+@click.option(
+    '--geoparquet1', '-gp1',
+    is_flag=True,
+    type=click.BOOL,
+    help='Enforces generating a GeoParquet 1.0 file. Defaults to GeoParquet 1.1 with bounding box.',
+    default=False
+)
+def improve(input, out, rename_column, add_sizes, fix_geometries, explode_geometries, crs, compression, geoparquet1):
+    """
+    "Improves" a fiboa GeoParquet file according to the given parameters.
+    """
+    log(f"fiboa CLI {__version__} - Improve datasets\n", "success")
+    try:
+        improve_(input, out, rename_column, add_sizes, fix_geometries, explode_geometries, crs, compression, geoparquet1)
+    except Exception as e:
+        log(e, "error")
+        sys.exit(1)
+
+
 cli.add_command(describe)
 cli.add_command(validate)
 cli.add_command(validate_schema)
@@ -574,6 +653,7 @@ def merge(datasets, out, crs, include, exclude, extension, compression, geoparqu
 cli.add_command(converters)
 cli.add_command(rename_extension)
 cli.add_command(merge)
+cli.add_command(improve)
 
 if __name__ == '__main__':
     cli()
diff --git a/fiboa_cli/const.py b/fiboa_cli/const.py
index 4df8d47e..8f54bd14 100644
--- a/fiboa_cli/const.py
+++ b/fiboa_cli/const.py
@@ -10,3 +10,14 @@
 STAC_COLLECTION_SCHEMA = "http://schemas.stacspec.org/v{version}/collection-spec/json-schema/collection.json"
 GEOPARQUET_SCHEMA = "https://geoparquet.org/releases/v{version}/schema.json"
 STAC_TABLE_EXTENSION = "https://stac-extensions.github.io/table/v1.2.0/schema.json"
+
+COMPRESSION_METHODS = ["brotli", "gzip", "lz4", "snappy", "zstd", "none"]
+
+CORE_COLUMNS = [
+    "id",
+    "geometry",
+    "area",
+    "perimeter",
+    "determination_datetime",
+    "determination_method",
+]
diff --git a/fiboa_cli/convert.py b/fiboa_cli/convert.py
index 84030b75..8f006ff5 100644
--- a/fiboa_cli/convert.py
+++ b/fiboa_cli/convert.py
@@ -13,7 +13,8 @@ def convert(
         collection = False,
         compression = None,
         geoparquet1 = False,
-        mapping_file=None,
+        mapping_file = None,
+        original_geometries = False,
     ):
     if dataset in IGNORED_DATASET_FILES:
         raise Exception(f"'{dataset}' is not a converter")
@@ -37,6 +38,7 @@ def convert(
         compression = compression,
         geoparquet1 = geoparquet1,
         mapping_file = mapping_file,
+        original_geometries = original_geometries,
     )
 
 def list_all_converter_ids():
diff --git a/fiboa_cli/convert_utils.py b/fiboa_cli/convert_utils.py
index cec0c1fb..74865220 100644
--- a/fiboa_cli/convert_utils.py
+++ b/fiboa_cli/convert_utils.py
@@ -40,7 +40,7 @@ def convert(
         license = None,
         compression = None,
         geoparquet1 = False,
-        explode_multipolygon = False,
+        original_geometries = False,
         index_as_id = False,
         **kwargs):
     """
@@ -160,11 +160,12 @@ def convert(
             else:
                 log(f"Column '{key}' not found in dataset, skipping migration", "warning")
 
-    # 4b. For geometry column, convert multipolygon type to polygon
-    if explode_multipolygon:
+    # 4b. For geometry column, fix geometries
+    if not original_geometries:
+        gdf.geometry = gdf.geometry.make_valid()
         gdf = gdf.explode()
 
-    if has_migration or has_col_migrations or has_col_filters or has_col_additions or explode_multipolygon:
+    if has_migration or has_col_migrations or has_col_filters or has_col_additions:
         log("GeoDataFrame after migrations and filters:")
         print(gdf.head())
 
diff --git a/fiboa_cli/datasets/be_wa.py b/fiboa_cli/datasets/be_wa.py
index f4013fdc..5e32635c 100644
--- a/fiboa_cli/datasets/be_wa.py
+++ b/fiboa_cli/datasets/be_wa.py
@@ -91,7 +91,6 @@ def file_migration(data, path, uri, layer):
         license=LICENSE,
         layer_filter=lambda layer, uri: layer == LAYER,
         file_migration=file_migration,
-        explode_multipolygon=True,
         index_as_id=True,
         **kwargs
     )
diff --git a/fiboa_cli/datasets/ch.py b/fiboa_cli/datasets/ch.py
index 86634e68..e265fbe0 100644
--- a/fiboa_cli/datasets/ch.py
+++ b/fiboa_cli/datasets/ch.py
@@ -69,7 +69,6 @@ def convert(output_file, cache = None, **kwargs):
         column_migrations=COLUMN_MIGRATIONS,
         column_filters=COLUMN_FILTERS,
         providers=PROVIDERS,
-        explode_multipolygon=True,
         index_as_id=True,
         fid_as_index=True,
         **kwargs
diff --git a/fiboa_cli/datasets/ec_fr.py b/fiboa_cli/datasets/ec_fr.py
index 392c5d34..de0ab2e8 100644
--- a/fiboa_cli/datasets/ec_fr.py
+++ b/fiboa_cli/datasets/ec_fr.py
@@ -25,6 +25,5 @@ def convert(output_file, cache = None, **kwargs):
         column_filters=base.COLUMN_FILTERS,
         attribution=base.ATTRIBUTION,
         license=LICENSE,
-        explode_multipolygon=True,
         **kwargs
     )
diff --git a/fiboa_cli/datasets/es_cat.py b/fiboa_cli/datasets/es_cat.py
index 4e1f0fca..327d9199 100644
--- a/fiboa_cli/datasets/es_cat.py
+++ b/fiboa_cli/datasets/es_cat.py
@@ -35,7 +35,6 @@
 
 COLUMN_MIGRATIONS = {
     "campanya": lambda col: pd.to_datetime(col, format='%Y'),
-    "geometry": lambda col: col.make_valid(),
 }
 
 MISSING_SCHEMAS = {
@@ -62,6 +61,5 @@ def convert(output_file, cache = None, **kwargs):
         license=LICENSE,
         layer="CULTIUS_DUN2023",
         index_as_id=True,
-        explode_multipolygon=True,
         **kwargs
     )
diff --git a/fiboa_cli/datasets/fi.py b/fiboa_cli/datasets/fi.py
index 743468b7..edec46c1 100644
--- a/fiboa_cli/datasets/fi.py
+++ b/fiboa_cli/datasets/fi.py
@@ -33,8 +33,6 @@
 COLUMN_MIGRATIONS = {
     # Make year (1st january) from column "VUOSI"
     "VUOSI": lambda col: pd.to_datetime(col, format='%Y'),
-    # Todo: generate a generic solution for making geometries valid
-    "geometry": lambda col: col.make_valid()
 }
 
 def migrate(gdf):
diff --git a/fiboa_cli/datasets/fr.py b/fiboa_cli/datasets/fr.py
index 36c53b82..f448363f 100644
--- a/fiboa_cli/datasets/fr.py
+++ b/fiboa_cli/datasets/fr.py
@@ -62,6 +62,5 @@ def convert(output_file, cache = None, **kwargs):
         column_filters=COLUMN_FILTERS,
         attribution=ATTRIBUTION,
         license=LICENSE,
-        explode_multipolygon=True,
         **kwargs
     )
diff --git a/fiboa_cli/datasets/ie.py b/fiboa_cli/datasets/ie.py
index b903128f..79e2cffd 100644
--- a/fiboa_cli/datasets/ie.py
+++ b/fiboa_cli/datasets/ie.py
@@ -67,6 +67,5 @@ def file_migration(data, path, uri, layer):
         license=LICENSE,
         layer_filter=lambda layer, uri: layer == LAYER,
         file_migration=file_migration,
-        explode_multipolygon=True,
         **kwargs
     )
diff --git a/fiboa_cli/datasets/nl.py b/fiboa_cli/datasets/nl.py
index f63bbefd..a70a62dd 100644
--- a/fiboa_cli/datasets/nl.py
+++ b/fiboa_cli/datasets/nl.py
@@ -80,7 +80,6 @@ def convert(output_file, cache = None, **kwargs):
         migration=migrate,
         attribution=ATTRIBUTION,
         license=LICENSE,
-        explode_multipolygon=True,
         index_as_id=True,
         **kwargs
     )
diff --git a/fiboa_cli/datasets/template.py b/fiboa_cli/datasets/template.py
index b9a5ae27..22af22a2 100644
--- a/fiboa_cli/datasets/template.py
+++ b/fiboa_cli/datasets/template.py
@@ -168,7 +168,6 @@ def convert(output_file, cache = None, **kwargs):
         attribution=ATTRIBUTION,
         license=LICENSE,
         # Other options:
-        # explode_multipolygon=True/False, # Converts MultiPolygons to Polygons
         # index_as_id=True/False, # Adds a column "id" with the index of the GeoDataFrame
         **kwargs
     )
diff --git a/fiboa_cli/describe.py b/fiboa_cli/describe.py
index c698f8c8..32649a6e 100644
--- a/fiboa_cli/describe.py
+++ b/fiboa_cli/describe.py
@@ -1,7 +1,7 @@
 import json
 import pandas as pd
 
-from .util import log, load_parquet_data, load_parquet_schema, load_parquet_metadata, parse_metadata, log_extensions
+from .util import log, load_parquet_data, load_parquet_schema, load_parquet_metadata, parse_metadata, log_extensions, is_schema_empty
 
 def describe(file, display_json=False, num=10, columns=None):
     metadata = load_parquet_metadata(file)
@@ -26,6 +26,10 @@ def describe(file, display_json=False, num=10, columns=None):
         if "fiboa_extensions" in collection and isinstance(collection["fiboa_extensions"], list):
             log_extensions(collection, log)
 
+        custom_schemas = collection.get("fiboa_custom_schemas", {})
+        if not is_schema_empty(custom_schemas):
+            log("Custom schemas: " + ", ".join(custom_schemas["properties"].keys()))
+
         if (display_json):
             log(json.dumps(collection, indent=2))
         elif "stac_version" in collection:
diff --git a/fiboa_cli/improve.py b/fiboa_cli/improve.py
new file mode 100644
index 00000000..b9d8efe8
--- /dev/null
+++ b/fiboa_cli/improve.py
@@ -0,0 +1,78 @@
+import os
+
+from .const import CORE_COLUMNS
+from .parquet import create_parquet
+from .util import load_parquet_data, load_parquet_schema, log, parse_metadata, pick_schemas, is_schema_empty
+
+
+def improve(input, out = None, rename_columns = {}, add_sizes = False, fix_geometries = False, explode_geometries = False, crs = None, compression = None, geoparquet1 = False):
+    # Prepare and determine location of the output file
+    if not out:
+        out = input
+    else:
+        dir = os.path.dirname(out)
+        if dir:
+            os.makedirs(dir, exist_ok=True)
+
+    # Load the dataset
+    schema = load_parquet_schema(input)
+    collection = parse_metadata(schema, b"fiboa")
+    columns = list(schema.names)
+    # Remove the bbox column to avoid conflicts when writing GeoParquet file later
+    columns.remove("bbox")
+    gdf = load_parquet_data(input, columns = columns)
+
+    # Change the CRS
+    if crs is not None:
+        gdf.to_crs(crs=crs, inplace=True)
+        log(f"Changed CRS to {crs}", "info")
+
+    # Fix geometries
+    if fix_geometries:
+        gdf.geometry = gdf.geometry.make_valid()
+        log("Fixed geometries", "info")
+
+    # Convert MultiPolygons to Polygons
+    if explode_geometries:
+        gdf = gdf.explode()
+        log("Exploded geometries", "info")
+
+    # Rename columns
+    if len(rename_columns) > 0:
+        for col in rename_columns:
+            columns[columns.index(col)] = rename_columns[col]
+            if col in CORE_COLUMNS:
+                log(f"Column {col} is a fiboa core field - do you really want to rename it?", "warning")
+            if ":" in col:
+                log(f"Column {col} may be a fiboa extension field - do you really want to rename it?", "warning")
+        gdf.rename(columns=rename_columns, inplace=True)
+        log("Renamed columns", "info")
+
+    # Add sizes
+    if add_sizes:
+        # Add the area and perimeter columns
+        for name in ["area", "perimeter"]:
+            if name not in columns:
+                # Create column if not present
+                gdf[name] = None
+                columns.append(name)
+
+        gdf_m = gdf
+        # Determine whether the given CRS is in meters
+        if gdf.crs.axis_info[0].unit_name not in ["m", "metre", "meter"]:
+            # Reproject the geometries to an equal-area projection if needed
+            gdf_m = gdf.to_crs("EPSG:6933")
+
+        # Compute the missing area and perimeter values
+        gdf["area"] = gdf_m["area"].fillna(gdf_m.geometry.area * 0.0001)
+        gdf["perimeter"] = gdf_m["perimeter"].fillna(gdf_m.geometry.length)
+
+    custom_schemas = collection.get("fiboa_custom_schemas", {})
+    custom_schemas = pick_schemas(custom_schemas, columns, rename_columns)
+    if not is_schema_empty(custom_schemas):
+        collection["fiboa_custom_schemas"] = custom_schemas
+
+
+    # Write the merged dataset to the output file
+    create_parquet(gdf, columns, collection, out, {}, compression=compression, geoparquet1=geoparquet1)
+    log(f"Wrote data to {out}", "success")
diff --git a/fiboa_cli/merge.py b/fiboa_cli/merge.py
index 4833310e..7e315277 100644
--- a/fiboa_cli/merge.py
+++ b/fiboa_cli/merge.py
@@ -2,18 +2,11 @@
 
 import pandas as pd
 
+from .const import CORE_COLUMNS
 from .parquet import create_parquet
-from .util import load_parquet_data, load_parquet_schema, log, parse_metadata
+from .util import load_parquet_data, load_parquet_schema, log, merge_schemas, parse_metadata, pick_schemas, is_schema_empty
 from .version import fiboa_version
 
-DEFAULT_COLUMNS = [
-    "id",
-    "geometry",
-    "area",
-    "perimeter",
-    "determination_datetime",
-    "determination_method",
-]
 DEFAULT_CRS = "EPSG:4326"
 
 def merge(datasets, out, crs = DEFAULT_CRS, includes = [], excludes = [], extensions = [], compression = None, geoparquet1 = False):
@@ -21,15 +14,17 @@ def merge(datasets, out, crs = DEFAULT_CRS, includes = [], excludes = [], extens
     if dir:
         os.makedirs(dir, exist_ok=True)
 
-    columns = DEFAULT_COLUMNS.copy()
+    columns = CORE_COLUMNS.copy()
     columns.extend(includes)
     columns = list(set(columns) - set(excludes))
 
     # Load the datasets
     all_gdf = []
+    custom_schemas = {}
     for dataset in datasets:
         # Load the dataset
         schema = load_parquet_schema(dataset)
+        collection = parse_metadata(schema, b"fiboa")
         file_columns = list(set(columns) & set(schema.names))
         gdf = load_parquet_data(dataset, columns=file_columns)
 
@@ -37,12 +32,15 @@ def merge(datasets, out, crs = DEFAULT_CRS, includes = [], excludes = [], extens
         gdf.to_crs(crs=crs, inplace=True)
 
         # Add collection column to each dataset
-        collection = parse_metadata(schema, b"fiboa")
         if collection is not None and "id" in collection:
             gdf["collection"] = collection["id"]
         else:
             gdf["collection"] = os.path.splitext(os.path.basename(dataset))[0]
 
+        # Merge custom schemas
+        custom_schema = collection.get("fiboa_custom_schemas", {})
+        custom_schemas = merge_schemas(custom_schemas, custom_schema)
+
         all_gdf.append(gdf)
 
     merged = pd.concat(all_gdf, ignore_index=True)
@@ -58,6 +56,11 @@ def merge(datasets, out, crs = DEFAULT_CRS, includes = [], excludes = [], extens
         "fiboa_extensions": extensions,
     }
 
+    # Add custom schemas
+    custom_schemas = pick_schemas(custom_schemas, columns)
+    if not is_schema_empty(custom_schemas):
+        collection["fiboa_custom_schemas"] = custom_schemas
+
     # Write the merged dataset to the output file
     create_parquet(merged, columns, collection, out, {}, compression=compression, geoparquet1=geoparquet1)
     log(f"Merged data to {out}", "success")
diff --git a/fiboa_cli/parquet.py b/fiboa_cli/parquet.py
index 43e03e19..11cc159a 100644
--- a/fiboa_cli/parquet.py
+++ b/fiboa_cli/parquet.py
@@ -5,7 +5,7 @@
 from shapely.geometry import shape
 
 from .types import get_geopandas_dtype, get_pyarrow_type_for_geopandas, get_pyarrow_field
-from .util import log, load_fiboa_schema, load_file, merge_schemas
+from .util import log, load_fiboa_schema, load_file, merge_schemas, is_schema_empty
 from .geopandas import to_parquet
 
 ROW_GROUP_SIZE = 25000
@@ -15,6 +15,11 @@ def create_parquet(data, columns, collection, output_file, config, missing_schem
     fiboa_schema = load_fiboa_schema(config)
     schemas = merge_schemas(missing_schemas, fiboa_schema)
 
+    # Add the custom schemas to the collection for future use
+    if not is_schema_empty(missing_schemas):
+        collection = collection.copy()
+        collection["fiboa_custom_schemas"] = missing_schemas
+
     # Load all extension schemas
     extensions = {}
     if "fiboa_extensions" in collection and isinstance(collection["fiboa_extensions"], list):
diff --git a/fiboa_cli/util.py b/fiboa_cli/util.py
index 1e9d46a6..fffb990a 100644
--- a/fiboa_cli/util.py
+++ b/fiboa_cli/util.py
@@ -224,6 +224,22 @@ def parse_converter_input_files(ctx, param, value):
     return sources
 
 
+def parse_map(value, separator = "="):
+    if value is None:
+        return {}
+    elif not isinstance(value, tuple):
+        raise click.BadParameter('Input files must be a tuple')
+    elif len(value) == 0:
+         return {}
+
+    mapping = {}
+    for v in value:
+        key, value = v.split(separator, 2)
+        mapping[key] = value
+
+    return mapping
+
+
 def name_from_uri(url):
     if "://" in url:
         try:
@@ -246,6 +262,10 @@ def check_ext_schema_for_cli(value, allow_none = False):
     return map_
 
 
+def is_schema_empty(schema):
+    return len(schema.get("properties", {})) == 0 and len(schema.get("required", {})) == 0
+
+
 def merge_schemas(*schemas):
     """Merge multiple schemas into one"""
     result = {
@@ -260,6 +280,24 @@ def merge_schemas(*schemas):
     return result
 
 
+def pick_schemas(schema, property_names, rename = {}):
+    """Pick and rename schemas for specific properties"""
+    result = {
+        "required": [],
+        "properties": {}
+    }
+    required = schema.get("required", [])
+    properties = schema.get("properties", {})
+    for prop in property_names:
+        prop2 = rename[prop] if prop in rename else prop
+        if prop in required:
+            result["required"].append(prop2)
+        if prop in properties:
+            result["properties"][prop2] = properties[prop]
+
+    return result
+
+
 def migrate_schema(schema):
     """Migrate schema to a new version"""
     return schema.copy()
diff --git a/fiboa_cli/validate.py b/fiboa_cli/validate.py
index 14c1bb42..08c0ac31 100644
--- a/fiboa_cli/validate.py
+++ b/fiboa_cli/validate.py
@@ -207,6 +207,10 @@ def validate_parquet(file, config):
     for ext in extensions.values():
         schemas = merge_schemas(schemas, ext)
 
+    # Add custom schemas
+    custom_schemas = collection.get("fiboa_custom_schemas", {})
+    schemas = merge_schemas(schemas, custom_schemas)
+
     # Check that all required fields are present
     for key in schemas.get("required", []):
         if key not in parquet_schema.names:
diff --git a/tests/test_improve.py b/tests/test_improve.py
new file mode 100644
index 00000000..b5e0634d
--- /dev/null
+++ b/tests/test_improve.py
@@ -0,0 +1,16 @@
+from fiboa_cli import improve, validate
+from click.testing import CliRunner
+from glob import glob
+
+
+def test_improve(tmp_file):
+    # merge files in directory
+    files = glob("tests/data-files/merge/at.parquet")
+    runner = CliRunner()
+    args = files + ['-o', tmp_file.name, '-sz', '-g', '-e']
+    result = runner.invoke(improve, args)
+    assert result.exit_code == 0, result.output
+
+    # Merged parquet file should be valid
+    result = runner.invoke(validate, [tmp_file.name, '--data'])
+    assert result.exit_code == 0, result.output
diff --git a/tests/test_merge.py b/tests/test_merge.py
index 085d5e8b..0f3c75d5 100644
--- a/tests/test_merge.py
+++ b/tests/test_merge.py
@@ -1,5 +1,3 @@
-from geopandas._compat import import_optional_dependency
-
 from fiboa_cli import merge, validate
 from click.testing import CliRunner
 from glob import glob
@@ -12,7 +10,7 @@
 """
 
 
-def test_converter(tmp_file):
+def test_merge(tmp_file):
     # merge files in directory
     files = glob("tests/data-files/merge/*.parquet")
     runner = CliRunner()