Interpolation/Summary: Add dataset to DataFrame

earthobservations · Jan 24, 2025 · 69c7104 · 69c7104
1 parent a069124
commit 69c7104
Show file tree

Hide file tree

Showing 9 changed files with 97 additions and 30 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -6,6 +6,7 @@
 - Stripes: Replace matplotlib by plotly
 - Explorer: Add download button for plot
 - Split up plotting extras into `plotting` and `matplotlib`
+- Interpolation/Summary: Add dataset to DataFrame
 
 ### Refactor
 - Update docker image extras

diff --git a/tests/core/timeseries/test_interpolation.py b/tests/core/timeseries/test_interpolation.py
@@ -25,6 +25,7 @@ def df_interpolated_empty():
     return pl.DataFrame(
         schema={
             Columns.STATION_ID.value: pl.String,
+            Columns.DATASET.value: pl.String,
             Columns.PARAMETER.value: pl.String,
             Columns.DATE.value: pl.Datetime(time_zone="UTC"),
             Columns.VALUE.value: pl.Float64,
@@ -50,6 +51,7 @@ def test_interpolation_temperature_air_mean_2m_hourly_by_coords(default_settings
         [
             {
                 "station_id": "f674568e",
+                "dataset": "temperature_air",
                 "parameter": "temperature_air_mean_2m",
                 "date": dt.datetime(2022, 1, 2, tzinfo=ZoneInfo("UTC")),
                 "value": 4.56,
@@ -74,6 +76,7 @@ def test_interpolation_temperature_air_mean_2m_daily_by_station_id(default_setti
         [
             {
                 "station_id": "6754d04d",
+                "dataset": "climate_summary",
                 "parameter": "temperature_air_mean_2m",
                 "date": dt.datetime(1986, 10, 31, tzinfo=ZoneInfo("UTC")),
                 "value": 6.37,
@@ -82,6 +85,7 @@ def test_interpolation_temperature_air_mean_2m_daily_by_station_id(default_setti
             },
             {
                 "station_id": "6754d04d",
+                "dataset": "climate_summary",
                 "parameter": "temperature_air_mean_2m",
                 "date": dt.datetime(1986, 11, 1, tzinfo=ZoneInfo("UTC")),
                 "value": 8.7,
@@ -117,6 +121,7 @@ def test_interpolation_precipitation_height_minute_10(default_settings):
         [
             {
                 "station_id": "f674568e",
+                "dataset": "precipitation",
                 "parameter": "precipitation_height",
                 "date": dt.datetime(2021, 10, 5, tzinfo=ZoneInfo("UTC")),
                 "value": 0.03,

diff --git a/tests/core/timeseries/test_summary.py b/tests/core/timeseries/test_summary.py
@@ -28,6 +28,7 @@ def test_summary_temperature_air_mean_2m_daily(default_settings):
         [
             {
                 "station_id": "7ac6c582",
+                "dataset": "climate_summary",
                 "parameter": "temperature_air_mean_2m",
                 "date": selected_dates[0],
                 "value": 0.5,
@@ -36,6 +37,7 @@ def test_summary_temperature_air_mean_2m_daily(default_settings):
             },
             {
                 "station_id": "7ac6c582",
+                "dataset": "climate_summary",
                 "parameter": "temperature_air_mean_2m",
                 "date": selected_dates[1],
                 "value": -5.5,
@@ -44,6 +46,7 @@ def test_summary_temperature_air_mean_2m_daily(default_settings):
             },
             {
                 "station_id": "7ac6c582",
+                "dataset": "climate_summary",
                 "parameter": "temperature_air_mean_2m",
                 "date": selected_dates[2],
                 "value": -2.7,
@@ -73,6 +76,7 @@ def test_not_summarizable_parameter(default_settings):
     expected_df = pl.DataFrame(
         schema={
             Columns.STATION_ID.value: pl.String,
+            Columns.DATASET.value: pl.String,
             Columns.PARAMETER.value: pl.String,
             Columns.DATE.value: pl.Datetime(time_zone="UTC"),
             Columns.VALUE.value: pl.Float64,

diff --git a/tests/ui/cli/test_cli_interpolate.py b/tests/ui/cli/test_cli_interpolate.py
@@ -29,6 +29,7 @@ def test_cli_interpolate():
     assert response["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 6.37,
@@ -37,6 +38,7 @@ def test_cli_interpolate():
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -203,6 +205,7 @@ def test_cli_interpolate_geojson():
                 "values": [
                     {
                         "station_id": "6754d04d",
+                        "dataset": "climate_summary",
                         "parameter": "temperature_air_mean_2m",
                         "date": "1986-10-31T00:00:00+00:00",
                         "value": 6.37,
@@ -212,6 +215,7 @@ def test_cli_interpolate_geojson():
                     {
                         "station_id": "6754d04d",
                         "date": "1986-11-01T00:00:00+00:00",
+                        "dataset": "climate_summary",
                         "parameter": "temperature_air_mean_2m",
                         "value": 8.7,
                         "distance_mean": 0.0,
@@ -246,6 +250,7 @@ def test_cli_interpolate_interpolation_station_distance():
     assert response["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": None,
@@ -254,6 +259,7 @@ def test_cli_interpolate_interpolation_station_distance():
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -286,6 +292,7 @@ def test_cli_interpolate_dont_use_nearby_station():
     assert response["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 6.37,
@@ -294,6 +301,7 @@ def test_cli_interpolate_dont_use_nearby_station():
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -326,6 +334,7 @@ def test_cli_interpolate_custom_units():
     assert response["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 43.47,
@@ -334,6 +343,7 @@ def test_cli_interpolate_custom_units():
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 47.66,

diff --git a/tests/ui/cli/test_cli_summarize.py b/tests/ui/cli/test_cli_summarize.py
@@ -28,6 +28,7 @@ def test_cli_summarize():
     assert response["values"] == [
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 6.6,
@@ -36,6 +37,7 @@ def test_cli_summarize():
         },
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -96,6 +98,7 @@ def test_cli_summarize_geojson():
                 "values": [
                     {
                         "station_id": "a87291a8",
+                        "dataset": "climate_summary",
                         "parameter": "temperature_air_mean_2m",
                         "date": "1986-10-31T00:00:00+00:00",
                         "value": 6.6,
@@ -104,6 +107,7 @@ def test_cli_summarize_geojson():
                     },
                     {
                         "station_id": "a87291a8",
+                        "dataset": "climate_summary",
                         "parameter": "temperature_air_mean_2m",
                         "date": "1986-11-01T00:00:00+00:00",
                         "value": 8.7,
@@ -139,6 +143,7 @@ def test_cli_summarize_custom_units():
     assert response["values"] == [
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 43.88,
@@ -147,6 +152,7 @@ def test_cli_summarize_custom_units():
         },
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 47.66,

diff --git a/tests/ui/test_restapi.py b/tests/ui/test_restapi.py
@@ -402,6 +402,7 @@ def test_interpolate_dwd(client):
     assert response.json()["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 6.37,
@@ -410,6 +411,7 @@ def test_interpolate_dwd(client):
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -436,6 +438,7 @@ def test_interpolate_dwd_lower_interpolation_distance(client):
     assert response.json()["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": None,
@@ -444,6 +447,7 @@ def test_interpolate_dwd_lower_interpolation_distance(client):
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -470,6 +474,7 @@ def test_interpolate_dwd_dont_use_nearby_station(client):
     assert response.json()["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 6.37,
@@ -478,6 +483,7 @@ def test_interpolate_dwd_dont_use_nearby_station(client):
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -507,6 +513,7 @@ def test_interpolate_dwd_custom_unit(client):
     assert response.json()["values"] == [
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 43.47,
@@ -515,6 +522,7 @@ def test_interpolate_dwd_custom_unit(client):
         },
         {
             "station_id": "6754d04d",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 47.66,
@@ -540,6 +548,7 @@ def test_summarize_dwd(client):
     assert response.json()["values"] == [
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 6.6,
@@ -548,6 +557,7 @@ def test_summarize_dwd(client):
         },
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 8.7,
@@ -577,6 +587,7 @@ def test_summarize_dwd_custom_unit(client):
     assert response.json()["values"] == [
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-10-31T00:00:00+00:00",
             "value": 43.88,
@@ -585,6 +596,7 @@ def test_summarize_dwd_custom_unit(client):
         },
         {
             "station_id": "a87291a8",
+            "dataset": "climate_summary",
             "parameter": "temperature_air_mean_2m",
             "date": "1986-11-01T00:00:00+00:00",
             "value": 47.66,