diff --git a/tests/variables.py b/tests/variables.py
new file mode 100644
index 00000000..a945348c
--- /dev/null
+++ b/tests/variables.py
@@ -0,0 +1,144 @@
+"""Module to generate datasets for testing."""
+
+from typing import Any, Callable, Optional, TypeVar
+
+import numpy as np
+import xarray as xr
+from xarray import DataArray, Dataset
+
+F = TypeVar("F", bound=Callable[..., Any])
+
+
+class Variables:
+    """Class to generate idealized variable datasets."""
+
+    def __init__(self, *args, **kwargs):
+        """
+        Initialize class generating a NEMO style variable datasets.
+
+        Parameters
+        ----------
+        *args
+            Arguments passed on to :py:func:`_generate_grid`
+        *kwargs
+            Keyword arguments passed on to
+            :py:func:`_generate_grid`.
+        """
+        self._vars = _generate_variables(*args, **kwargs)
+
+    def const(self, var, value) -> Dataset:
+        """
+        Produce a constant variable field..
+
+        Parameters
+        ----------
+        var: string
+           variable name
+        value: float
+           constant value for the field
+
+        Returns
+        -------
+        Dataset.
+        """
+        ds = self._vars
+        print(var)
+        print(value)
+        return ds
+
+    def lin_grad(self, var, direction) -> Dataset:
+        """
+        Generate linear gradient in variable field.
+
+        Produces ...
+
+        Parameters
+        ----------
+        var: string
+            variable name
+        direction: string
+            i,j,k,t
+
+        Returns
+        -------
+        Dataset.
+        """
+        ds = self._vars
+        print(var)
+        print(direction)
+        return ds
+
+    def real_grad(self, var, direction) -> Dataset:
+        """
+        Generate realistic gradient in variable field.
+
+        Produces ...
+
+        Parameters
+        ----------
+        var: string
+            variable name
+        direction: string
+            i,j,k,t
+
+        Returns
+        -------
+        Dataset.
+        """
+        ds = self._vars
+        print(var)
+        print(direction)
+        return ds
+
+def _generate_variables(
+    ds_domcfg,
+) -> Dataset:
+    """
+    Generate a Dataset of T/S/U/V given a NEMO bathy Dataset.
+
+    Parameters
+    ----------
+    ds_domcfg: Dataset
+
+    Returns
+    -------
+    Dataset
+        Equivalent of NEMO output file.
+
+    Raises
+    ------
+    ValueError
+    """
+    ds = Dataset()
+
+    # Define variables
+    temp = xr.full_like(ds_domcfg.e3t, 20.0, dtype=np.double)
+    salt = xr.full_like(ds_domcfg.e3t, 35.0, dtype=np.double)
+    uvel = xr.full_like(ds_domcfg.e3t, 1.0, dtype=np.double)
+    vvel = xr.full_like(ds_domcfg.e3t, 1.0, dtype=np.double)
+    ssh = xr.full_like(ds_domcfg.e2t, 0.0, dtype=np.double)
+
+    # Add attributes
+    temp.attrs = dict(units="C", long_name="Temperature (degrees centigrade)")
+    salt.attrs = dict(units="", long_name="Salinity (unitless)")
+    uvel.attrs = dict(units="m/s", long_name="Zonal Velocity (m/s)")
+    vvel.attrs = dict(units="m/s", long_name="Meridional Velocity (m/s)")
+    ssh.attrs = dict(units="m", long_name="Sea Surface Height (m)")
+
+    # Add to ds
+    ds["votemper"] = temp
+    ds["vosaline"] = salt
+    ds["vozocrtx"] = uvel
+    ds["vomecrty"] = vvel
+    ds["sossheig"] = ssh
+
+    # Add t dim
+    # times = pd.date_range("2020","2023",freq='Y')+ DateOffset(months=6)
+    # TODO: add input to adjust frequency of data
+    times = xr.cftime_range("2000", periods=3, freq="A-JUN", calendar="standard")
+    da_t = xr.DataArray(times, [("t", times)])
+    ds = ds.expand_dims(dim={"t": da_t})
+    ds = ds.transpose("t", "z", "y", "x")
+    ds["t"] = ds["t"]
+
+    return ds