Enhance metadata handling for dimension coordinates in dataset templates

- Introduced a new method `_dim_coord_metadata` in `AbstractDatasetTemplate` to streamline the retrieval of metadata for dimension coordinates, ensuring unitless coordinates do not receive placeholder values.
- Updated coordinate metadata handling in `Seismic2DCdpGathersTemplate`, `Seismic3DCdpGathersTemplate`, `Seismic3DCocaGathersTemplate`, `Seismic3DReceiverGathersTemplate`, and others to utilize the new method, improving code clarity and consistency.
- Refactored coordinate addition logic across multiple templates to reduce redundancy and enhance maintainability.

Author: BrianMichell

src/mdio/builder/templates/base.py

@@ -269,6 +269,26 @@ def get_unit_by_key(self, key: str) -> AllUnitModel | None:
         """Get units by variable/dimension/coordinate name. Returns None if not found."""
         return self._units.get(key, None)
 
+    def _dim_coord_metadata(self, name: str) -> VariableMetadata | None:
+        """Build metadata for a dimension coordinate.
+
+        Dimension coordinates such as ``inline`` or ``crossline`` are integer survey
+        indices and are inherently unitless. Returning ``None`` here keeps the
+        ``metadata`` field unset on those coordinates rather than wrapping a
+        ``units_v1=None`` placeholder.
+
+        Args:
+            name: Coordinate name to look up units for.
+
+        Returns:
+            A ``VariableMetadata`` carrying the registered unit, or ``None`` if no
+            unit is registered for ``name``.
+        """
+        unit = self.get_unit_by_key(name)
+        if unit is None:
+            return None
+        return VariableMetadata(units_v1=unit)
+
     def _add_dimensions(self) -> None:
         """Add custom dimensions.
 
@@ -290,20 +310,14 @@ def _add_coordinates(self) -> None:
                 name,
                 dimensions=(name,),
                 data_type=ScalarType.INT32,
-                metadata=VariableMetadata(units_v1=self.get_unit_by_key(name)),
+                metadata=self._dim_coord_metadata(name),
             )
 
-        # Add non-dimension coordinates with computed chunk sizes
-        for name in self.coordinate_names:
-            # Compute optimal chunk size for coordinates (spatial dimensions only)
-            spatial_shape = self._dim_sizes[:-1]  # Exclude vertical dimension
-            coord_chunk_shape = get_constrained_chunksize(
-                spatial_shape,
-                ScalarType.FLOAT64,
-                MAX_COORDINATES_BYTES,
-            )
-            chunk_grid = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape))
+        spatial_shape = self._dim_sizes[:-1]
+        coord_chunk_shape = get_constrained_chunksize(spatial_shape, ScalarType.FLOAT64, MAX_COORDINATES_BYTES)
+        chunk_grid = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape))
 
+        for name in self.coordinate_names:
             self._builder.add_coordinate(
                 name=name,
                 dimensions=self.spatial_dimension_names,
@@ -313,14 +327,9 @@ def _add_coordinates(self) -> None:
             )
 
     def _add_trace_mask(self) -> None:
-        """Add trace mask variables with computed chunk sizes."""
-        # Compute optimal chunk size for trace mask (spatial dimensions only)
-        spatial_shape = self._dim_sizes[:-1]  # Exclude vertical dimension
-        mask_chunk_shape = get_constrained_chunksize(
-            spatial_shape,
-            ScalarType.BOOL,
-            MAX_SIZE_LIVE_MASK,
-        )
+        """Add trace mask variable."""
+        spatial_shape = self._dim_sizes[:-1]
+        mask_chunk_shape = get_constrained_chunksize(spatial_shape, ScalarType.BOOL, MAX_SIZE_LIVE_MASK)
         chunk_grid = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=mask_chunk_shape))
 
         self._builder.add_variable(
@@ -345,15 +354,7 @@ def _add_trace_headers(self, header_dtype: StructuredType) -> None:
         )
 
     def _add_raw_headers(self) -> None:
-        """Add raw binary headers variable.
-
-        This variable stores the raw binary header bytes for each trace, which can be useful
-        for preserving original SEG-Y header information. Only supported in Zarr v3 format.
-
-        The raw headers variable has:
-        - Same spatial dimensions as trace headers (all dimensions except vertical)
-        - No coordinates
-        """
+        """Add raw 240-byte trace header variable. Zarr v3 only."""
         chunk_grid = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=self.full_chunk_shape[:-1]))
         self._builder.add_variable(
             name="raw_headers",

src/mdio/builder/templates/seismic_2d_cdp.py

@@ -53,28 +53,21 @@ def _add_coordinates(self) -> None:
             "cdp",
             dimensions=("cdp",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("cdp")),
         )
         self._builder.add_coordinate(
             self._gather_domain,
             dimensions=(self._gather_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self._gather_domain)),
+            metadata=self._dim_coord_metadata(self._gather_domain),
         )
         self._builder.add_coordinate(
             self.trace_domain,
             dimensions=(self.trace_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self.trace_domain)),
+            metadata=self._dim_coord_metadata(self.trace_domain),
         )
 
-        # Add non-dimension coordinates
-        coord_spatial_shape = (self._dim_sizes[0],)  # cdp
-        coord_chunk_shape = get_constrained_chunksize(
-            coord_spatial_shape,
-            ScalarType.FLOAT64,
-            MAX_COORDINATES_BYTES,
-        )
+        coord_chunk_shape = get_constrained_chunksize(self._dim_sizes[:1], ScalarType.FLOAT64, MAX_COORDINATES_BYTES)
         chunk_grid = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape))
 
         compressor = Blosc(cname=BloscCname.zstd)

src/mdio/builder/templates/seismic_2d_streamer_shot.py

@@ -27,14 +27,20 @@ def _load_dataset_attributes(self) -> dict[str, Any]:
         return {"surveyType": "2D", "gatherType": "common_source"}
 
     def _add_coordinates(self) -> None:
-        # Add dimension coordinates
-        for name in self._dim_names:
+        # Add dimension coordinates. shot_point and channel are unitless integer
+        # indices; only the trace (time/depth) domain carries a unit.
+        for name in ("shot_point", "channel"):
             self._builder.add_coordinate(
                 name,
                 dimensions=(name,),
                 data_type=ScalarType.INT32,
-                metadata=VariableMetadata(units_v1=self.get_unit_by_key(name)),
             )
+        self._builder.add_coordinate(
+            self._data_domain,
+            dimensions=(self._data_domain,),
+            data_type=ScalarType.INT32,
+            metadata=self._dim_coord_metadata(self._data_domain),
+        )
 
         # Add non-dimension coordinates
         compressor = compressors.Blosc(cname=compressors.BloscCname.zstd)

src/mdio/builder/templates/seismic_3d_cdp.py

@@ -53,34 +53,26 @@ def _add_coordinates(self) -> None:
             "inline",
             dimensions=("inline",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("inline")),
         )
         self._builder.add_coordinate(
             "crossline",
             dimensions=("crossline",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("crossline")),
         )
         self._builder.add_coordinate(
             self._gather_domain,
             dimensions=(self._gather_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self._gather_domain)),
+            metadata=self._dim_coord_metadata(self._gather_domain),
         )
         self._builder.add_coordinate(
             self.trace_domain,
             dimensions=(self.trace_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self.trace_domain)),
+            metadata=self._dim_coord_metadata(self.trace_domain),
         )
 
-        # Add non-dimension coordinates
-        coord_spatial_shape = (self._dim_sizes[0], self._dim_sizes[1])  # inline, crossline
-        coord_chunk_shape = get_constrained_chunksize(
-            coord_spatial_shape,
-            ScalarType.FLOAT64,
-            MAX_COORDINATES_BYTES,
-        )
+        coord_chunk_shape = get_constrained_chunksize(self._dim_sizes[:2], ScalarType.FLOAT64, MAX_COORDINATES_BYTES)
         chunk_grid = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape))
 
         compressor = Blosc(cname=BloscCname.zstd)

src/mdio/builder/templates/seismic_3d_coca.py

@@ -44,40 +44,32 @@ def _add_coordinates(self) -> None:
             "inline",
             dimensions=("inline",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("inline")),
         )
         self._builder.add_coordinate(
             "crossline",
             dimensions=("crossline",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("crossline")),
         )
         self._builder.add_coordinate(
             "offset",
             dimensions=("offset",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("offset")),  # same unit as X/Y
+            metadata=self._dim_coord_metadata("offset"),
         )
         self._builder.add_coordinate(
             "azimuth",
             dimensions=("azimuth",),
             data_type=ScalarType.FLOAT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("azimuth")),
+            metadata=self._dim_coord_metadata("azimuth"),
         )
         self._builder.add_coordinate(
             self.trace_domain,
             dimensions=(self.trace_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self.trace_domain)),
+            metadata=self._dim_coord_metadata(self.trace_domain),
         )
 
-        # Add non-dimension coordinates
-        coord_spatial_shape = (self._dim_sizes[0], self._dim_sizes[1])  # inline, crossline
-        coord_chunk_shape = get_constrained_chunksize(
-            coord_spatial_shape,
-            ScalarType.FLOAT64,
-            MAX_COORDINATES_BYTES,
-        )
+        coord_chunk_shape = get_constrained_chunksize(self._dim_sizes[:2], ScalarType.FLOAT64, MAX_COORDINATES_BYTES)
         chunk_grid = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape))
 
         compressor = compressors.Blosc(cname=compressors.BloscCname.zstd)

src/mdio/builder/templates/seismic_3d_obn.py

@@ -97,31 +97,16 @@ def _add_coordinates(self) -> None:
             self._data_domain,
             dimensions=(self._data_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self._data_domain)),
+            metadata=self._dim_coord_metadata(self._data_domain),
         )
 
-        # Add non-dimension coordinates with computed chunk sizes
-        # For 1D coordinates (over receiver)
-        coord_spatial_shape_1d = (self._dim_sizes[1],)  # receiver
-        coord_chunk_shape_1d = get_constrained_chunksize(
-            coord_spatial_shape_1d,
-            ScalarType.FLOAT64,
-            MAX_COORDINATES_BYTES,
+        # Chunk grids for receiver-indexed (1D) and shot-indexed (3D) non-dim coordinates.
+        receiver_chunk_shape = get_constrained_chunksize(
+            self._dim_sizes[1:2], ScalarType.FLOAT64, MAX_COORDINATES_BYTES
         )
-        chunk_grid_1d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape_1d))
-
-        # For 3D coordinates (over shot_line, gun, shot_index)
-        coord_spatial_shape_3d = (
-            self._dim_sizes[2],
-            self._dim_sizes[3],
-            self._dim_sizes[4],
-        )  # shot_line, gun, shot_index
-        coord_chunk_shape_3d = get_constrained_chunksize(
-            coord_spatial_shape_3d,
-            ScalarType.FLOAT64,
-            MAX_COORDINATES_BYTES,
-        )
-        chunk_grid_3d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape_3d))
+        chunk_grid_1d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=receiver_chunk_shape))
+        shot_chunk_shape = get_constrained_chunksize(self._dim_sizes[2:5], ScalarType.FLOAT64, MAX_COORDINATES_BYTES)
+        chunk_grid_3d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=shot_chunk_shape))
 
         compressor = Blosc(cname=BloscCname.zstd)
         self._builder.add_coordinate(

src/mdio/builder/templates/seismic_3d_offset_tiles.py

@@ -39,37 +39,31 @@ def _add_coordinates(self) -> None:
             "inline",
             dimensions=("inline",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("inline")),
         )
         self._builder.add_coordinate(
             "crossline",
             dimensions=("crossline",),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("crossline")),
         )
         self._builder.add_coordinate(
             "inline_offset_tile",
             dimensions=("inline_offset_tile",),
             data_type=ScalarType.INT16,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("inline_offset_tile")),
         )
         self._builder.add_coordinate(
             "crossline_offset_tile",
             dimensions=("crossline_offset_tile",),
             data_type=ScalarType.INT16,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("crossline_offset_tile")),
         )
         self._builder.add_coordinate(
             self.trace_domain,
             dimensions=(self.trace_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self.trace_domain)),
+            metadata=self._dim_coord_metadata(self.trace_domain),
         )
 
         # Add non-dimension coordinates
         compressor = compressors.Blosc(cname=compressors.BloscCname.zstd)
-
-        # CDP coordinates (vary by inline, crossline)
         self._builder.add_coordinate(
             "cdp_x",
             dimensions=("inline", "crossline"),

src/mdio/builder/templates/seismic_3d_receiver_gathers.py

@@ -39,19 +39,17 @@ def _add_coordinates(self) -> None:
             "receiver",
             dimensions=("receiver",),
             data_type=ScalarType.UINT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("receiver")),
         )
         self._builder.add_coordinate(
             "shot_line",
             dimensions=("shot_line",),
             data_type=ScalarType.UINT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key("shot_line")),
         )
         self._builder.add_coordinate(
             self.trace_domain,
             dimensions=(self.trace_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self.trace_domain)),
+            metadata=self._dim_coord_metadata(self.trace_domain),
         )
 
         # Add non-dimension coordinates

src/mdio/builder/templates/seismic_3d_shot_receiver_line.py

@@ -58,7 +58,7 @@ def _add_coordinates(self) -> None:
             self._data_domain,
             dimensions=(self._data_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self._data_domain)),
+            metadata=self._dim_coord_metadata(self._data_domain),
         )
 
         # Add non-dimension coordinates

src/mdio/builder/templates/seismic_3d_streamer_field.py

@@ -85,37 +85,16 @@ def _add_coordinates(self) -> None:
             self._data_domain,
             dimensions=(self._data_domain,),
             data_type=ScalarType.INT32,
-            metadata=VariableMetadata(units_v1=self.get_unit_by_key(self._data_domain)),
+            metadata=self._dim_coord_metadata(self._data_domain),
         )
 
-        # Add non-dimension coordinates with computed chunk sizes
-        # For 3D coordinates (over sail_line, gun, shot_index)
-        coord_spatial_shape_3d = (
-            self._dim_sizes[0],
-            self._dim_sizes[1],
-            self._dim_sizes[2],
-        )  # sail_line, gun, shot_index
-        coord_chunk_shape_3d = get_constrained_chunksize(
-            coord_spatial_shape_3d,
-            ScalarType.FLOAT64,
-            MAX_COORDINATES_BYTES,
+        # Chunk grids for shot-indexed (3D) and receiver-indexed (5D) non-dim coordinates.
+        shot_chunk_shape = get_constrained_chunksize(self._dim_sizes[:3], ScalarType.FLOAT64, MAX_COORDINATES_BYTES)
+        chunk_grid_3d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=shot_chunk_shape))
+        receiver_chunk_shape = get_constrained_chunksize(
+            self._dim_sizes[:5], ScalarType.FLOAT64, MAX_COORDINATES_BYTES
         )
-        chunk_grid_3d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape_3d))
-
-        # For 5D coordinates (over sail_line, gun, shot_index, cable, channel)
-        coord_spatial_shape_5d = (
-            self._dim_sizes[0],
-            self._dim_sizes[1],
-            self._dim_sizes[2],
-            self._dim_sizes[3],
-            self._dim_sizes[4],
-        )  # sail_line, gun, shot_index, cable, channel
-        coord_chunk_shape_5d = get_constrained_chunksize(
-            coord_spatial_shape_5d,
-            ScalarType.FLOAT64,
-            MAX_COORDINATES_BYTES,
-        )
-        chunk_grid_5d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=coord_chunk_shape_5d))
+        chunk_grid_5d = RegularChunkGrid(configuration=RegularChunkShape(chunk_shape=receiver_chunk_shape))
 
         compressor = Blosc(cname=BloscCname.zstd)
         self._builder.add_coordinate(