First pass review and alignment of templates

src/mdio/builder/template_registry.py

@@ -76,10 +76,11 @@ def _register_default_templates(self) -> None:
         self.register(Seismic3DPostStackTemplate("depth"))
 
         # CDP/CMP Ordered Data
-        self.register(Seismic2DPreStackCDPTemplate("time"))
-        self.register(Seismic2DPreStackCDPTemplate("depth"))
-        self.register(Seismic3DPreStackCDPTemplate("time"))
-        self.register(Seismic3DPreStackCDPTemplate("depth"))
+        for data_domain in ("time", "depth"):
+            for gather_domain in ("offset", "angle"):
+                self.register(Seismic3DPreStackCDPTemplate(data_domain, gather_domain))
+                self.register(Seismic2DPreStackCDPTemplate(data_domain, gather_domain))
+
         self.register(Seismic3DPreStackCocaTemplate("time"))
         self.register(Seismic3DPreStackCocaTemplate("depth"))
 

src/mdio/builder/templates/abstract_dataset_template.py

@@ -15,6 +15,7 @@
 from mdio.builder.schemas.v1.units import LengthUnitModel
 from mdio.builder.schemas.v1.variable import CoordinateMetadata
 from mdio.builder.schemas.v1.variable import VariableMetadata
+from mdio.builder.templates.types import SeismicDataDomain
 
 
 class AbstractDatasetTemplate(ABC):
@@ -24,8 +25,13 @@ class AbstractDatasetTemplate(ABC):
     to override specific steps.
     """
 
-    def __init__(self, domain: str = "") -> None:
-        self._trace_domain = domain.lower()
+    def __init__(self, data_domain: SeismicDataDomain) -> None:
+        self._data_domain = data_domain.lower()
+
+        if self._data_domain not in ["depth", "time"]:
+            msg = "domain must be 'depth' or 'time'"
+            raise ValueError(msg)
+
         self._coord_dim_names = ()
         self._dim_names = ()
         self._coord_names = ()
@@ -80,7 +86,7 @@ def default_variable_name(self) -> str:
     @property
     def trace_domain(self) -> str:
         """Returns the name of the trace domain."""
-        return self._trace_domain
+        return self._data_domain
 
     @property
     def dimension_names(self) -> tuple[str, ...]:

src/mdio/builder/templates/seismic_2d_poststack.py

@@ -3,22 +3,23 @@
 from typing import Any
 
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
+from mdio.builder.templates.abstract_dataset_template import SeismicDataDomain
 
 
 class Seismic2DPostStackTemplate(AbstractDatasetTemplate):
     """Seismic post-stack 2D time or depth Dataset template."""
 
-    def __init__(self, domain: str):
-        super().__init__(domain=domain)
+    def __init__(self, data_domain: SeismicDataDomain):
+        super().__init__(data_domain=data_domain)
 
         self._coord_dim_names = ("cdp",)
-        self._dim_names = (*self._coord_dim_names, self._trace_domain)
+        self._dim_names = (*self._coord_dim_names, self._data_domain)
         self._coord_names = ("cdp_x", "cdp_y")
         self._var_chunk_shape = (1024, 1024)
 
     @property
     def _name(self) -> str:
-        return f"PostStack2D{self._trace_domain.capitalize()}"
+        return f"PostStack2D{self._data_domain.capitalize()}"
 
     def _load_dataset_attributes(self) -> dict[str, Any]:
-        return {"surveyDimensionality": "2D", "ensembleType": "line", "processingStage": "post-stack"}
+        return {"surveyType": "2D", "gatherType": "stacked"}

src/mdio/builder/templates/seismic_2d_prestack_cdp.py

@@ -3,22 +3,31 @@
 from typing import Any
 
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
+from mdio.builder.templates.abstract_dataset_template import SeismicDataDomain
+from mdio.builder.templates.types import CdpGatherDomain
 
 
 class Seismic2DPreStackCDPTemplate(AbstractDatasetTemplate):
     """Seismic CDP pre-stack 2D time or depth Dataset template."""
 
-    def __init__(self, domain: str):
-        super().__init__(domain=domain)
+    def __init__(self, data_domain: SeismicDataDomain, gather_domain: CdpGatherDomain):
+        super().__init__(data_domain=data_domain)
+        self._gather_domain = gather_domain.lower()
 
-        self._coord_dim_names = ("cdp", "offset")
-        self._dim_names = (*self._coord_dim_names, self._trace_domain)
+        if self._gather_domain not in ["offset", "angle"]:
+            msg = "gather_type must be 'offset' or 'angle'"
+            raise ValueError(msg)
+
+        self._coord_dim_names = ("cdp", self._gather_domain)
+        self._dim_names = (*self._coord_dim_names, self._data_domain)
         self._coord_names = ("cdp_x", "cdp_y")
-        self._var_chunk_shape = (1, 512, 4096)
+        self._var_chunk_shape = (16, 64, 1024)
 
     @property
     def _name(self) -> str:
-        return f"PreStackCdpGathers2D{self._trace_domain.capitalize()}"
+        gather_domain_suffix = self._gather_domain.capitalize()
+        data_domain_suffix = self._data_domain.capitalize()
+        return f"PreStackCdp{gather_domain_suffix}Gathers2D{data_domain_suffix}"
 
     def _load_dataset_attributes(self) -> dict[str, Any]:
-        return {"surveyDimensionality": "2D", "ensembleType": "cdp", "processingStage": "pre-stack"}
+        return {"surveyType": "2D", "gatherType": "cdp"}

src/mdio/builder/templates/seismic_2d_prestack_shot.py

@@ -5,25 +5,26 @@
 from mdio.builder.schemas.dtype import ScalarType
 from mdio.builder.schemas.v1.variable import CoordinateMetadata
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
+from mdio.builder.templates.abstract_dataset_template import SeismicDataDomain
 
 
 class Seismic2DPreStackShotTemplate(AbstractDatasetTemplate):
     """Seismic Shot pre-stack 2D time or depth Dataset template."""
 
-    def __init__(self, domain: str):
-        super().__init__(domain=domain)
+    def __init__(self, data_domain: SeismicDataDomain):
+        super().__init__(data_domain=data_domain)
 
-        self._coord_dim_names = ("shot_point", "channel")  # Custom coordinate definition for shot gathers
-        self._dim_names = (*self._coord_dim_names, self._trace_domain)
+        self._coord_dim_names = ("shot_point", "channel")
+        self._dim_names = (*self._coord_dim_names, self._data_domain)
         self._coord_names = ("gun", "source_coord_x", "source_coord_y", "group_coord_x", "group_coord_y")
-        self._var_chunk_shape = (1, 512, 4096)
+        self._var_chunk_shape = (16, 64, 1024)
 
     @property
     def _name(self) -> str:
-        return f"PreStackShotGathers2D{self._trace_domain.capitalize()}"
+        return f"PreStackShotGathers2D{self._data_domain.capitalize()}"
 
     def _load_dataset_attributes(self) -> dict[str, Any]:
-        return {"surveyDimensionality": "2D", "ensembleType": "shot_point", "processingStage": "pre-stack"}
+        return {"surveyType": "2D", "ensembleType": "common_source"}
 
     def _add_coordinates(self) -> None:
         # Add dimension coordinates
@@ -34,18 +35,18 @@ def _add_coordinates(self) -> None:
         coordinate_metadata = CoordinateMetadata(units_v1=self._horizontal_coord_unit)
         self._builder.add_coordinate(
             "gun",
-            dimensions=("shot_point", "channel"),
+            dimensions=("shot_point",),
             data_type=ScalarType.UINT8,
         )
         self._builder.add_coordinate(
             "source_coord_x",
-            dimensions=("shot_point", "channel"),
+            dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
             metadata=coordinate_metadata,
         )
         self._builder.add_coordinate(
             "source_coord_y",
-            dimensions=("shot_point", "channel"),
+            dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
             metadata=coordinate_metadata,
         )

src/mdio/builder/templates/seismic_3d_poststack.py

@@ -3,22 +3,24 @@
 from typing import Any
 
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
+from mdio.builder.templates.abstract_dataset_template import SeismicDataDomain
 
 
 class Seismic3DPostStackTemplate(AbstractDatasetTemplate):
     """Seismic post-stack 3D time or depth Dataset template."""
 
-    def __init__(self, domain: str):
-        super().__init__(domain=domain)
-        # Template attributes to be overridden by subclasses
+    def __init__(self, data_domain: SeismicDataDomain):
+        super().__init__(data_domain=data_domain)
+
         self._coord_dim_names = ("inline", "crossline")
-        self._dim_names = (*self._coord_dim_names, self._trace_domain)
+        self._dim_names = (*self._coord_dim_names, self._data_domain)
         self._coord_names = ("cdp_x", "cdp_y")
         self._var_chunk_shape = (128, 128, 128)
 
     @property
     def _name(self) -> str:
-        return f"PostStack3D{self._trace_domain.capitalize()}"
+        domain_suffix = self._data_domain.capitalize()
+        return f"PostStack3D{domain_suffix}"
 
     def _load_dataset_attributes(self) -> dict[str, Any]:
-        return {"surveyDimensionality": "3D", "ensembleType": "line", "processingStage": "post-stack"}
+        return {"surveyType": "3D", "gatherType": "stacked"}

src/mdio/builder/templates/seismic_3d_prestack_cdp.py

@@ -3,22 +3,31 @@
 from typing import Any
 
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
+from mdio.builder.templates.abstract_dataset_template import SeismicDataDomain
+from mdio.builder.templates.types import CdpGatherDomain
 
 
 class Seismic3DPreStackCDPTemplate(AbstractDatasetTemplate):
-    """Seismic CDP pre-stack 3D time or depth Dataset template."""
+    """Seismic CDP pre-stack 3D gathers Dataset template."""
 
-    def __init__(self, domain: str):
-        super().__init__(domain=domain)
+    def __init__(self, data_domain: SeismicDataDomain, gather_domain: CdpGatherDomain):
+        super().__init__(data_domain=data_domain)
+        self._gather_domain = gather_domain.lower()
 
-        self._coord_dim_names = ("inline", "crossline", "offset")
-        self._dim_names = (*self._coord_dim_names, self._trace_domain)
+        if self._gather_domain not in ["offset", "angle"]:
+            msg = "gather_type must be 'offset' or 'angle'"
+            raise ValueError(msg)
+
+        self._coord_dim_names = ("inline", "crossline", self._gather_domain)
+        self._dim_names = (*self._coord_dim_names, self._data_domain)
         self._coord_names = ("cdp_x", "cdp_y")
-        self._var_chunk_shape = (1, 1, 512, 4096)
+        self._var_chunk_shape = (8, 8, 32, 512)
 
     @property
     def _name(self) -> str:
-        return f"PreStackCdpGathers3D{self._trace_domain.capitalize()}"
+        gather_domain_suffix = self._gather_domain.capitalize()
+        data_domain_suffix = self._data_domain.capitalize()
+        return f"PreStackCdp{gather_domain_suffix}Gathers3D{data_domain_suffix}"
 
     def _load_dataset_attributes(self) -> dict[str, Any]:
-        return {"surveyDimensionality": "3D", "ensembleType": "cdp", "processingStage": "pre-stack"}
+        return {"surveyType": "3D", "gatherType": "cdp"}

src/mdio/builder/templates/seismic_3d_prestack_coca.py

@@ -6,25 +6,26 @@
 from mdio.builder.schemas.v1.units import AngleUnitModel
 from mdio.builder.schemas.v1.variable import CoordinateMetadata
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
+from mdio.builder.templates.abstract_dataset_template import SeismicDataDomain
 
 
 class Seismic3DPreStackCocaTemplate(AbstractDatasetTemplate):
     """Seismic Shot pre-stack 3D time or depth Dataset template."""
 
-    def __init__(self, domain: str):
-        super().__init__(domain=domain)
+    def __init__(self, data_domain: SeismicDataDomain):
+        super().__init__(data_domain=data_domain)
 
         self._coord_dim_names = ("inline", "crossline", "offset", "azimuth")
-        self._dim_names = (*self._coord_dim_names, self._trace_domain)
+        self._dim_names = (*self._coord_dim_names, self._data_domain)
         self._coord_names = ("cdp_x", "cdp_y")
         self._var_chunk_shape = (8, 8, 32, 1, 1024)
 
     @property
     def _name(self) -> str:
-        return f"PreStackCocaGathers3D{self._trace_domain.capitalize()}"
+        return f"PreStackCocaGathers3D{self._data_domain.capitalize()}"
 
     def _load_dataset_attributes(self) -> dict[str, Any]:
-        return {"surveyDimensionality": "3D", "ensembleType": "cdp_coca", "processingStage": "pre-stack"}
+        return {"surveyType": "3D", "gatherType": "common_offset_common_azimuth"}
 
     def _add_coordinates(self) -> None:
         # Add dimension coordinates

src/mdio/builder/templates/seismic_3d_prestack_shot.py

@@ -5,25 +5,26 @@
 from mdio.builder.schemas.dtype import ScalarType
 from mdio.builder.schemas.v1.variable import CoordinateMetadata
 from mdio.builder.templates.abstract_dataset_template import AbstractDatasetTemplate
+from mdio.builder.templates.abstract_dataset_template import SeismicDataDomain
 
 
 class Seismic3DPreStackShotTemplate(AbstractDatasetTemplate):
     """Seismic Shot pre-stack 3D time or depth Dataset template."""
 
-    def __init__(self, domain: str):
-        super().__init__(domain=domain)
+    def __init__(self, data_domain: SeismicDataDomain):
+        super().__init__(data_domain=data_domain)
 
-        self._coord_dim_names = ("shot_point", "cable", "channel")  # Custom coordinates for shot gathers
-        self._dim_names = (*self._coord_dim_names, self._trace_domain)
+        self._coord_dim_names = ("shot_point", "cable", "channel")
+        self._dim_names = (*self._coord_dim_names, self._data_domain)
         self._coord_names = ("gun", "source_coord_x", "source_coord_y", "group_coord_x", "group_coord_y")
-        self._var_chunk_shape = (1, 1, 512, 4096)
+        self._var_chunk_shape = (8, 2, 128, 1024)
 
     @property
     def _name(self) -> str:
-        return f"PreStackShotGathers3D{self._trace_domain.capitalize()}"
+        return f"PreStackShotGathers3D{self._data_domain.capitalize()}"
 
     def _load_dataset_attributes(self) -> dict[str, Any]:
-        return {"surveyDimensionality": "3D", "ensembleType": "shot_point", "processingStage": "pre-stack"}
+        return {"surveyType": "3D", "ensembleType": "common_source"}
 
     def _add_coordinates(self) -> None:
         # Add dimension coordinates
@@ -33,18 +34,18 @@ def _add_coordinates(self) -> None:
         # Add non-dimension coordinates
         self._builder.add_coordinate(
             "gun",
-            dimensions=("shot_point", "cable", "channel"),
+            dimensions=("shot_point",),
             data_type=ScalarType.UINT8,
         )
         self._builder.add_coordinate(
             "source_coord_x",
-            dimensions=("shot_point", "cable", "channel"),
+            dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )
         self._builder.add_coordinate(
             "source_coord_y",
-            dimensions=("shot_point", "cable", "channel"),
+            dimensions=("shot_point",),
             data_type=ScalarType.FLOAT64,
             metadata=CoordinateMetadata(units_v1=self._horizontal_coord_unit),
         )

src/mdio/builder/templates/types.py

@@ -0,0 +1,8 @@
+"""Module that contains type aliases for templates."""
+
+from typing import Literal
+from typing import TypeAlias
+
+SeismicDataDomain: TypeAlias = Literal["depth", "time"]
+
+CdpGatherDomain: TypeAlias = Literal["offset", "angle"]

tests/integration/test_segy_import_export.py

@@ -255,12 +255,11 @@ def test_dataset_metadata(self, zarr_tmp: Path) -> None:
 
         attributes = ds.attrs["attributes"]
         assert attributes is not None
-        assert len(attributes) == 6
+        assert len(attributes) == 5
         # Validate all attributes provided by the abstract template
         assert attributes["defaultVariableName"] == "amplitude"
-        assert attributes["surveyDimensionality"] == "3D"
-        assert attributes["ensembleType"] == "line"
-        assert attributes["processingStage"] == "post-stack"
+        assert attributes["surveyType"] == "3D"
+        assert attributes["gatherType"] == "stacked"
         assert attributes["textHeader"] == text_header_teapot_dome()
         assert attributes["binaryHeader"] == binary_header_teapot_dome()
 

tests/integration/test_segy_import_export_masked.py

@@ -280,9 +280,9 @@ def test_import(self, test_conf: MaskedExportConfig, export_masked_path: Path) -
             case "3d_stack":
                 template_name = "PostStack3D" + domain
             case "2d_gather":
-                template_name = "PreStackCdpGathers2D" + domain
+                template_name = "PreStackCdpOffsetGathers2D" + domain
             case "3d_gather":
-                template_name = "PreStackCdpGathers3D" + domain
+                template_name = "PreStackCdpOffsetGathers3D" + domain
             case "2d_streamer":
                 template_name = "PreStackShotGathers2D" + domain
             case "3d_streamer":