diff --git a/CHANGELOG.md b/CHANGELOG.md
index c41fa0b27..85f6c87c3 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,7 @@
 
 ### Fixes
 
+- Stop `dbt docs generate` from emitting a `RuntimeWarning` per schema for catalog metadata columns that are absent from the results (thanks @emrekabalcii!) ([#1749](https://github.com/databricks/dbt-databricks/pull/1749) resolves [#1748](https://github.com/databricks/dbt-databricks/issues/1748))
 - Unset tags on columns that a `table` model or incremental full refresh removes, so replacing a relation no longer fails with `CANNOT_DROP_TAGGED_COLUMN` when a removed column carries a governed tag ([#1707](https://github.com/databricks/dbt-databricks/pull/1707) resolves [#1684](https://github.com/databricks/dbt-databricks/issues/1684))
 - Load wide seeds faster by resolving the placeholder and `parquet` cast types once per seed instead of once per cell ([#1722](https://github.com/databricks/dbt-databricks/pull/1722))
 - Stop materialized views with a Compatibility Mode location from being dropped and lost when `use_safer_relation_operations` or a `partition_by` change would recreate them at the same location. Unchanged partitions now use `CREATE OR REPLACE`, and location or partition changes fail before anything is dropped ([#1727](https://github.com/databricks/dbt-databricks/pull/1727))
diff --git a/dbt/adapters/databricks/impl.py b/dbt/adapters/databricks/impl.py
index f6e3c6aa1..6d69e8f6d 100644
--- a/dbt/adapters/databricks/impl.py
+++ b/dbt/adapters/databricks/impl.py
@@ -902,11 +902,11 @@ def _get_schema_for_catalog(self, catalog: str, schema: str, identifier: str) ->
             )
             for relation, information in self._list_relations_with_information(schema_relation):
                 columns.extend(self._get_columns_for_catalog(relation, information))
-        # An empty input produces a 0-column table; force-typing absent columns
-        # would emit one RuntimeWarning per name. Skip the override in that case.
-        column_types = (
-            build_type_tester(self.CATALOG_TEXT_ONLY_COLUMNS) if columns else DEFAULT_TYPE_TESTER
-        )
+        # Rows omit None-valued keys, so a forced column can be absent from the table, and
+        # agate emits one RuntimeWarning per absent name. Only force the columns present.
+        present = {key for row in columns for key in row}
+        text_only = [col for col in self.CATALOG_TEXT_ONLY_COLUMNS if col in present]
+        column_types = build_type_tester(text_only) if text_only else DEFAULT_TYPE_TESTER
         return Table.from_object(columns, column_types=column_types)
 
     def _get_columns_for_catalog(  # type: ignore[override]
diff --git a/tests/unit/test_adapter.py b/tests/unit/test_adapter.py
index fc81a9d26..18fe98628 100644
--- a/tests/unit/test_adapter.py
+++ b/tests/unit/test_adapter.py
@@ -570,6 +570,28 @@ def test_get_schema_for_catalog__no_columns_emits_no_warnings(self, _):
                     table = adapter._get_schema_for_catalog("database", "schema", "name")
         assert len(table.rows) == 0
 
+    @patch("dbt.adapters.databricks.api_client.DatabricksApiClient")
+    def test_get_schema_for_catalog__omitted_none_columns_emit_no_warnings(self, _):
+        # `to_column_dict` omits None-valued keys, so a schema with no owner or comments
+        # yields rows without `table_owner`, `table_comment` and `comment`.
+        row = {
+            k: v
+            for k, v in _catalog_row("12345").items()
+            if k not in ("table_owner", "table_comment", "comment")
+        }
+        with patch.object(DatabricksAdapter, "_list_relations_with_information") as list_info:
+            list_info.return_value = [(Mock(), "info")]
+            with patch.object(DatabricksAdapter, "_get_columns_for_catalog") as get_columns:
+                get_columns.return_value = [row]
+                adapter = DatabricksAdapter(Mock(flags={}), get_context("spawn"))
+                with warnings.catch_warnings():
+                    warnings.simplefilter("error", RuntimeWarning)
+                    table = adapter._get_schema_for_catalog("cat", "schema", "name")
+
+        assert "comment" not in table.column_names
+        types = dict(zip(table.column_names, table.column_types))
+        assert isinstance(types["column_name"], agate.data_types.text.Text)
+
     def test_simple_catalog_relation(self):
         self.maxDiff = None
         rel_type = DatabricksRelation.get_relation_type.Table