feat: add source-grounded evolution and Ollama assistant

scripts/provision_official_landuse_timeseries.py

@@ -87,6 +87,8 @@ class ThemeDefinition:
     key: str
     label: str
     class_ids: tuple[int, ...]
+    reference_layer_name: str
+    metric_label: str
 
 
 @dataclass(frozen=True)
@@ -101,7 +103,18 @@ class PreparedSnapshot:
     vector_sha256: str
 
 
-THEMES = (ThemeDefinition("forest", "Bos", (12,)),)
+THEMES = (
+    ThemeDefinition("forest", "Bos", (12,), "forest", "Bosoppervlakte"),
+    ThemeDefinition("water", "Water", (17,), "water", "Wateroppervlakte"),
+    ThemeDefinition(
+        "built",
+        "Bebouwde functies",
+        (1, 2, 3, 4, 6, 7, 8),
+        "buildings",
+        "Oppervlakte bebouwde functies",
+    ),
+    ThemeDefinition("transport", "Transportinfrastructuur", (5,), "roads", "Oppervlakte transportinfrastructuur"),
+)
 
 
 def parse_args() -> argparse.Namespace:
@@ -113,7 +126,7 @@ def parse_args() -> argparse.Namespace:
     parser.add_argument("--nis-code", default=DEFAULT_NIS_CODE)
     parser.add_argument("--scope-key", default=DEFAULT_SCOPE_KEY)
     parser.add_argument("--years", default=",".join(str(year) for year in SUPPORTED_YEARS))
-    parser.add_argument("--themes", default="forest")
+    parser.add_argument("--themes", default="forest,water,built,transport")
     parser.add_argument(
         "--boundary-path",
         type=Path,
@@ -528,8 +541,8 @@ def polygonize_snapshot(
                     "properties": {
                         "source_name": "department_omgeving_land_use",
                         "source_feature_id": feature_id,
-                        "reference_layer_name": theme.key,
-                        "layer_type": theme.key,
+                        "reference_layer_name": theme.reference_layer_name,
+                        "layer_type": theme.reference_layer_name,
                         "authority_level": "authoritative",
                         "coverage_scope": scope_key,
                         **identity,
@@ -619,9 +632,12 @@ def prepare_snapshot(
     partition_boundaries: list[tuple[str, Any]],
     year: int,
     theme: ThemeDefinition,
+    refresh_source: bool = False,
 ) -> PreparedSnapshot:
     stem = f"{args.scope_key}_land_use_{theme.key}_{year}"
-    raster_path = args.output_dir / f"{stem}.tif"
+    legacy_forest_raster = args.output_dir / f"{args.scope_key}_land_use_forest_{year}.tif"
+    shared_raster = args.output_dir / f"{args.scope_key}_land_use_source_{year}.tif"
+    raster_path = legacy_forest_raster if not args.force and legacy_forest_raster.exists() else shared_raster
     vector_path = args.output_dir / f"{stem}.geojson"
     manifest_path = args.output_dir / f"{stem}.manifest.json"
     if not args.force:
@@ -636,7 +652,7 @@ def prepare_snapshot(
             return prepared
 
     raster_profile: dict[str, Any]
-    if args.force or not raster_path.exists():
+    if refresh_source or not raster_path.exists():
         if partition_boundaries:
             raster_profile = download_partitioned_raster(
                 session,
@@ -797,13 +813,18 @@ def build_source_metadata(args: argparse.Namespace, snapshot: PreparedSnapshot)
         "polygon_crs": OUTPUT_CRS,
         "land_use_class_ids": list(snapshot.theme.class_ids),
         "land_use_class_names": [LAND_USE_CLASSES[class_id] for class_id in snapshot.theme.class_ids],
-        "temporal_series_label": SERIES_LABEL,
+        "theme": snapshot.theme.reference_layer_name,
+        "temporal_series_label": (
+            SERIES_LABEL if snapshot.theme.key == "forest" else f"{SERIES_LABEL} - {snapshot.theme.label}"
+        ),
         "observation_date_precision": "year",
         "identity_stable": False,
+        "semantic_metrics": False,
         "identity_limitation": "Raster-derived polygons can split or merge between source editions; object lineage is not inferred.",
         "selection_aggregation": {
             "method": "intersection_area",
-            "label": "Oppervlakte",
+            "metric_key": f"{snapshot.theme.key}_area",
+            "label": snapshot.theme.metric_label,
             "unit": "ha",
             "is_estimate": False,
             "warning": "Oppervlakte is exact binnen de officiele 10 m rasterrepresentatie en is niet perceelsnauwkeurig.",
@@ -856,7 +877,7 @@ def upload_snapshot(
                 "source": "operator_official_import",
                 "dataset_role": "reference",
                 "source_name": "department_omgeving_land_use",
-                "reference_layer_name": snapshot.theme.key,
+                "reference_layer_name": snapshot.theme.reference_layer_name,
                 "source_metadata_json": json.dumps(build_source_metadata(args, snapshot), ensure_ascii=False),
                 "provenance_metadata_json": json.dumps(build_provenance_metadata(args, snapshot), ensure_ascii=False),
                 "area_id": area_id,
@@ -915,6 +936,7 @@ def main() -> int:
                             partition_boundaries=partition_boundaries,
                             year=year,
                             theme=theme,
+                            refresh_source=args.force and theme is definitions[0],
                         )
                     )
 

scripts/provision_regional_timeseries.py

@@ -138,7 +138,7 @@ def build_operator_commands(
                     "--years",
                     args.landuse_years,
                     "--themes",
-                    "forest",
+                    "forest,water,built,transport",
                     "--boundary-path",
                     str(boundary_path),
                     *partition_flags,