From 5ac3238c2750d9bb0f481a6f6146294818507f6b Mon Sep 17 00:00:00 2001
From: Jens <jens@localhost>
Date: Thu, 30 Jul 2026 03:35:15 +0200
Subject: [PATCH] Reject same-year post-mosaic building labels

---
 .../test_building_label_normalization.py      | 32 +++++++++++++++++++
 scripts/normalize_belgium_building_labels.py  | 15 +++++++--
 2 files changed, 44 insertions(+), 3 deletions(-)

diff --git a/backend/tests/test_building_label_normalization.py b/backend/tests/test_building_label_normalization.py
index c0530296..1f22799b 100644
--- a/backend/tests/test_building_label_normalization.py
+++ b/backend/tests/test_building_label_normalization.py
@@ -134,6 +134,38 @@ def test_normalizer_rejects_features_created_after_dated_imagery(tmp_path: Path)
     assert audit["decision_counts"] == {"accepted": 1, "created_after_imagery_period": 1}
 
 
+def test_normalizer_rejects_same_year_feature_after_annual_mosaic_start(tmp_path: Path) -> None:
+    raster_path = tmp_path / "image.tif"
+    with rasterio.open(
+        raster_path, "w", driver="GTiff", width=100, height=100, count=3,
+        dtype="uint8", crs="EPSG:4326", transform=from_origin(4.0, 51.0, 0.001, 0.001),
+    ) as dataset:
+        dataset.write(np.zeros((3, 100, 100), dtype="uint8"))
+    geometry = {
+        "type": "Polygon",
+        "coordinates": [[[4.01, 50.99], [4.02, 50.99], [4.02, 50.98], [4.01, 50.98], [4.01, 50.99]]],
+    }
+    reference_path = tmp_path / "reference.geojson"
+    reference_path.write_text(json.dumps({
+        "type": "FeatureCollection",
+        "features": [{"type": "Feature", "properties": {"BEGINDATUM": "2025-08-18"}, "geometry": geometry}],
+    }), encoding="utf-8")
+
+    normalized, audit = module.normalize(
+        reference_path=reference_path,
+        raster_path=raster_path,
+        source_name="grb",
+        min_label_px=3,
+        imagery_observed_at="2025-01-01T00:00:00Z",
+        imagery_valid_to="2025-12-31T23:59:59Z",
+        reference_observed_at="2026-07-01T00:00:00Z",
+    )
+
+    assert normalized["features"] == []
+    assert audit["decision_counts"] == {"created_after_imagery_period": 1}
+    assert audit["imagery_feature_creation_cutoff"] == "2025-01-01T00:00:00+00:00"
+
+
 def test_normalizer_merges_only_touching_visible_roof_instances(tmp_path: Path) -> None:
     raster_path = tmp_path / "image.tif"
     with rasterio.open(
diff --git a/scripts/normalize_belgium_building_labels.py b/scripts/normalize_belgium_building_labels.py
index e96a2225..3a4e983e 100644
--- a/scripts/normalize_belgium_building_labels.py
+++ b/scripts/normalize_belgium_building_labels.py
@@ -159,9 +159,16 @@ def normalize(
     decisions: list[dict[str, Any]] = []
     seen: set[str] = set()
     counts: Counter[str] = Counter()
-    imagery_cutoff = (
-        datetime.fromisoformat(imagery_valid_to.replace("Z", "+00:00")) if imagery_valid_to else None
-    )
+    imagery_dates = [
+        datetime.fromisoformat(value.replace("Z", "+00:00"))
+        for value in (imagery_observed_at, imagery_valid_to)
+        if value
+    ]
+    # Annual mosaics expose a validity interval but not the flight date for
+    # each pixel. Using the interval end admits buildings created later in the
+    # same year that are visibly absent from the mosaic. The earliest governed
+    # date is therefore the only conservative training-label cutoff.
+    imagery_cutoff = min(imagery_dates) if imagery_dates else None
     with rasterio.open(raster_path) as raster:
         if raster.crs is None:
             raise SystemExit("Raster CRS is required")
@@ -260,6 +267,8 @@ def normalize(
         "min_label_px": min_label_px,
         "imagery_observed_at": imagery_observed_at,
         "imagery_valid_to": imagery_valid_to,
+        "imagery_feature_creation_cutoff": imagery_cutoff.isoformat() if imagery_cutoff else None,
+        "imagery_feature_creation_cutoff_policy": "earliest_governed_imagery_date",
         "reference_observed_at": reference_observed_at,
         "temporal_mismatch_days": temporal_mismatch_days,
         "temporal_alignment_status": temporal_alignment_status,