set an infinity translation speed to NaN (will be filled)

stormevents/nhc/track.py

@@ -1135,6 +1135,8 @@ def __compute_velocity(data: DataFrame) -> DataFrame:
                 .astype(float)
             )
             speeds = pandas.Series(distances / abs(intervals), index=indices)
+            # if Inf (interval is zero but distance != 0) set to NaN
+            speeds[speeds == numpy.inf] = numpy.nan
             bearings = pandas.Series(inverse_azimuths % 360, index=indices)
             # use forward azimuths for negative intervals
             bearings[intervals < 0] = pandas.Series(

tests/test_nhc.py

@@ -312,6 +312,25 @@ def test_vortex_track_forecast_time_fromfile_arg():
     assert pandas.to_datetime(dates) == pandas.to_datetime("09-10-2018")
 
 
+def test_vortex_track_forecast_time_fromfile_duplicated_forecast_hours():
+    # Test from_file to accept forecast_time argument
+    # this file has analysis times with several 0 forecast hours
+    # followed by hourly timestep forecasts from the forecast_time.
+    # One of the >0 forecast_hours is duplicated which creates and Inf
+    # trans speed which is then set to NaN by the code to avoid an error
+    input_directory = INPUT_DIRECTORY / "test_vortex_track_from_file"
+
+    track = VortexTrack.from_file(
+        input_directory / "helene2024_fort.22",
+        file_deck="a",
+        advisories="OFCL",
+        forecast_time="2024-09-24 12:00",
+    )
+
+    dates = track.data.track_start_time.unique()
+    assert len(dates) == 1
+
+
 # def test_vortex_track_forecast_time_outofbound_date():
 #     # Test it raises if forecast time is not between start and end
 #     msg = ""