536 add check for nan values before running the soil model

tests/models/soil/test_soil_model.py

@@ -40,7 +40,6 @@
     (DEBUG, "soil model: required var 'soil_p_pool_secondary' checked"),
     (DEBUG, "soil model: required var 'soil_p_pool_labile' checked"),
     (DEBUG, "soil model: required var 'pH' checked"),
-    (DEBUG, "soil model: required var 'bulk_density' checked"),
     (DEBUG, "soil model: required var 'clay_fraction' checked"),
 )
 POST_SETUP_LOG = (
@@ -511,6 +510,27 @@ def test_integrate_soil_model(
     log_check(caplog, expected_log)
 
 
+def test_integrate_with_nans(caplog, fixture_soil_model):
+    """Test that integration fails if NaN values are in the input data."""
+
+    # Add Nan value to data and then clean up caplog
+    fixture_soil_model.data["pH"] = DataArray([3.3, np.nan, 5.6, 7.9], dims=["cell_id"])
+    caplog.clear()
+
+    with pytest.raises(ValueError):
+        _ = fixture_soil_model.integrate()
+
+    expected_log = (
+        (
+            ERROR,
+            "Soil model integration cannot proceed because the following variables "
+            "have unexpected NaN values: {'pH'}",
+        ),
+    )
+
+    log_check(caplog, expected_log)
+
+
 def test_order_independance(
     dummy_carbon_data,
     fixture_soil_model,
@@ -595,6 +615,87 @@ def test_order_independance(
         assert np.allclose(output[pool_name], output_reversed[pool_name])
 
 
+@pytest.mark.parametrize(
+    argnames=["unexpected_nans", "variable_name", "input_data"],
+    argvalues=[
+        pytest.param(
+            False,
+            "pH",
+            DataArray([3.3, 4.3, 5.6, 7.9], dims=["cell_id"]),
+            id="no NaNs",
+        ),
+        pytest.param(
+            True,
+            "pH",
+            DataArray([3.3, np.nan, 5.6, 7.9], dims=["cell_id"]),
+            id="NaN",
+        ),
+    ],
+)
+def test_check_for_unexpected_nan_value_flat(
+    fixture_soil_model, unexpected_nans, variable_name, input_data
+):
+    """Test unexpected NaN checking values works for variables without layers."""
+
+    fixture_soil_model.data[variable_name] = input_data
+
+    assert unexpected_nans == fixture_soil_model.check_for_unexpected_nan_values(
+        var=variable_name
+    )
+
+
+@pytest.mark.parametrize(
+    argnames=["unexpected_nans", "variable_name", "layer_name", "input_data"],
+    argvalues=[
+        pytest.param(
+            False,
+            "air_temperature",
+            "index_surface",
+            np.array([3.3, 4.3, 5.6, 7.9]),
+            id="surface, good",
+        ),
+        pytest.param(
+            True,
+            "air_temperature",
+            "index_surface",
+            np.array([3.3, np.nan, 5.6, 7.9]),
+            id="surface, bad",
+        ),
+        pytest.param(
+            False,
+            "soil_temperature",
+            "index_all_soil",
+            np.array([[3.3, 4.3, 5.6, 7.9], [23.4, 26.1, 24.4, 29.8]]),
+            id="soil, good",
+        ),
+        pytest.param(
+            True,
+            "soil_temperature",
+            "index_all_soil",
+            np.array([[3.3, 4.3, 5.6, 7.9], [np.nan, 26.1, 24.4, 29.8]]),
+            id="soil, bad",
+        ),
+    ],
+)
+def test_check_for_unexpected_nan_value_layered(
+    fixture_soil_model,
+    fixture_core_components,
+    unexpected_nans,
+    variable_name,
+    layer_name,
+    input_data,
+):
+    """Test unexpected NaN checking values works for variables without layers."""
+
+    lyr_str = fixture_core_components.layer_structure
+    fixture_soil_model.data[variable_name] = lyr_str.from_template()
+    fixture_soil_model.data[variable_name][getattr(lyr_str, layer_name)] = input_data
+
+    assert unexpected_nans == fixture_soil_model.check_for_unexpected_nan_values(
+        var=variable_name
+    )
+
+
 def test_convert_fruiting_body_production_to_rate(fixture_soil_model):
     """Test that conversion of fruiting body production to a rate works."""
 

virtual_ecosystem/models/soil/soil_model.py

@@ -89,7 +89,6 @@ class SoilModel(
         "soil_p_pool_secondary",
         "soil_p_pool_labile",
         "pH",
-        "bulk_density",
         "clay_fraction",
     ),
     vars_populated_by_init=(
@@ -128,9 +127,12 @@ class SoilModel(
         "soil_p_pool_primary",
         "soil_p_pool_secondary",
         "soil_p_pool_labile",
+        "pH",
+        "clay_fraction",
         "matric_potential",
         "vertical_flow",
         "soil_temperature",
+        "air_temperature",
         "soil_moisture",
         "litter_C_mineralisation_rate",
         "litter_N_mineralisation_rate",
@@ -396,6 +398,7 @@ def integrate(self) -> dict[str, DataArray]:
 
         Raises:
             IntegrationError: When the integration cannot be successfully completed.
+            ValueError: If any of the variables used by the soil model have NaN values.
         """
 
         # Find number of grid cells integration is being performed over
@@ -431,6 +434,22 @@ def integrate(self) -> dict[str, DataArray]:
             **{name: np.array([]) for name in self.refreshed_variables},
         }
 
+        # Check if any values used by the soil model integration have NaN values (these
+        # can stall the integration)
+        unexpected_nans = set()
+
+        for var in self.vars_required_for_update:
+            if self.check_for_unexpected_nan_values(var=var):
+                unexpected_nans.add(var)
+
+        if unexpected_nans:
+            to_raise_nan = ValueError(
+                "Soil model integration cannot proceed because the following "
+                f"variables have unexpected NaN values: {unexpected_nans}"
+            )
+            LOGGER.error(to_raise_nan)
+            raise to_raise_nan
+
         # Carry out simulation
         output = solve_ivp(
             construct_full_soil_model,
@@ -453,12 +472,12 @@ def integrate(self) -> dict[str, DataArray]:
 
         # Check if integration failed
         if not output.success:
-            LOGGER.error(
-                "Integration of soil module failed with following message: {}".format(  # noqa: UP032
-                    str(output.message)
-                )
+            to_raise = IntegrationError(
+                "Integration of soil module failed with following message: "
+                f"{output.message!s}"
             )
-            raise IntegrationError()
+            LOGGER.error(to_raise)
+            raise to_raise
 
         # Construct index slices
         slices = make_slices(no_cells, round(len(y0) / no_cells))
@@ -471,6 +490,29 @@ def integrate(self) -> dict[str, DataArray]:
 
         return new_c_pools
 
+    def check_for_unexpected_nan_values(self, var: str) -> bool:
+        """Check if there are unexpected NaN values in the data for a specific variable.
+
+        The soil model needs the air_temperature variable to have non-NaN values at the
+        soil surface, and the other layer structured variables to be defined for every
+        soil layer. For these variables, this function takes the appropriate subset.
+
+        Args:
+            var: The name of the variable being checked
+
+        Returns:
+            Whether the data for the variable has any unexpected NaN values.
+        """
+
+        if var == "air_temperature":
+            subset = self.data[var].isel(layers=self.layer_structure.index_surface)
+        elif "layers" in self.data[var].dims:
+            subset = self.data[var].isel(layers=self.layer_structure.index_all_soil)
+        else:
+            subset = self.data[var]
+
+        return bool(subset.isnull().any())
+
     def convert_fruiting_body_production_to_rate(
         self, total_production: DataArray
     ) -> dict[str, DataArray]: