allow timezone aware time selection

Author: vincentsarago

CHANGES.md

@@ -2,6 +2,10 @@
 
 ## Unreleased
 
+### titiler.xarray
+
+* add support for timezone aware time selection (e.g `sel="time=2023-01-01T00:00:00+03:00"`)
+
 ## 0.22.2 (2025-06-02)
 
 ### titiler.application

src/titiler/xarray/tests/test_io_tools.py

@@ -169,6 +169,46 @@ def test_get_variable():
         da = get_variable(ds, "dataset")
 
 
+def test_get_variable_datetime_tz():
+    """test io.get_variable with datetime and timezones."""
+    arr = numpy.arange(0, 33 * 35 * 2).reshape(2, 33, 35)
+    data = xarray.DataArray(
+        arr,
+        dims=("time", "y", "x"),
+        coords={
+            "x": numpy.arange(-170, 180, 10),
+            "y": numpy.arange(-80, 85, 5),
+            "time": [
+                datetime(2022, 1, 1),
+                datetime(2023, 1, 1),
+            ],
+        },
+    )
+    data.attrs.update({"valid_min": arr.min(), "valid_max": arr.max()})
+    assert not data.rio.crs
+    assert data.dims == ("time", "y", "x")
+    ds = data.to_dataset(name="dataset")
+
+    da = get_variable(ds, "dataset", sel=["time=2023-01-01T00:00:00"], method="nearest")
+    assert da.rio.crs
+    assert da.dims == ("y", "x")
+    assert da["time"] == numpy.datetime64("2023-01-01")
+
+    da = get_variable(
+        ds, "dataset", sel=["time=2023-01-01T00:00:00Z"], method="nearest"
+    )
+    assert da.rio.crs
+    assert da.dims == ("y", "x")
+    assert da["time"] == numpy.datetime64("2023-01-01")
+
+    da = get_variable(
+        ds, "dataset", sel=["time=2023-01-01T00:00:00+03:00"], method="nearest"
+    )
+    assert da.rio.crs
+    assert da.dims == ("y", "x")
+    assert da["time"] == numpy.datetime64("2023-01-01")
+
+
 @pytest.mark.parametrize(
     "protocol,filename",
     [

src/titiler/xarray/titiler/xarray/io.py

@@ -8,6 +8,7 @@
 import pandas
 import xarray
 from morecantile import TileMatrixSet
+from pandas._libs.tslibs.parsing import DateParseError
 from rio_tiler.constants import WEB_MERCATOR_TMS
 from rio_tiler.io.xarray import XarrayReader
 from xarray.namedarray.utils import module_available
@@ -143,6 +144,12 @@ def _cast_to_type(value, dtype: Any) -> Any:
     elif numpy.issubdtype(dtype, numpy.floating):
         value = float(value)
 
+    elif numpy.issubdtype(dtype, numpy.datetime64) or dtype == "O":
+        try:
+            value = pandas.Timestamp(value).to_datetime64()
+        except DateParseError:
+            pass
+
     return value