pyissm 0.0.1.dev1__py3-none-any.whl

pyissm/__init__.py

@@ -0,0 +1 @@
+from . import analysis, data, learn, plot, model, tools

pyissm/analysis/__init__.py

@@ -0,0 +1 @@
+from . import ismip

pyissm/analysis/ismip.py

@@ -0,0 +1,103 @@
+"""
+ISMIP-related functions for ISSM
+
+This module contains functions for handling ISMIP-related data and operations.
+"""
+
+from pyissm import tools
+
+def calc_perc_ice_cover(total_ice_area, ice_area):
+    """
+    Calculate the percentage ice cover, relative to the of total ice area.
+
+    This function computes the fraction of the ice area that is grounded (i.e.,
+    not floating). The result is returned as a percentage.
+
+    Parameters
+    ----------
+    total_ice_area : ndarray
+        Array representing the total ice area.
+
+    ice_area : ndarray
+        Array representing the ice area for percentage coverage to be computed.
+
+    Returns
+    -------
+    ice_area_perc : ndarray
+        Array with the same shape as `ice_area`, where each
+        element represents the percentage of the total_ice_area that is covered by the ice_area.
+
+    """
+
+    ## Calculate ice area fraction
+    ice_area_perc = (ice_area / total_ice_area) * 100
+
+    return ice_area_perc
+
+
+def get_ismip_variable(md, ismip_variable_name):
+    """
+    Retrieve a specific ISMIP variable from a model object.
+
+    This function extracts and calculates a requested ISMIP (Ice Sheet Model Intercomparison Project)
+    variable from a given model object, if the necessary data is available. It supports area-based
+    variables calculated from grounded and floating ice components stored in the model's
+    transient solution.
+
+    Parameters
+    ----------
+    md : object
+        The model object containing transient solution results, typically with
+        attributes like `md.results.TransientSolution.GroundedArea` and `FloatingArea`.
+
+    ismip_variable_name : str
+        Name of the ISMIP variable to compute. Must be one of:
+        - 'land_ice_area_fraction' (assumes all ice in ISSM is land ice and returns 100 %)
+        - 'floating_ice_shelf_area_fraction'
+        - 'grounded_ice_sheet_area_fraction'
+
+    Returns
+    -------
+    variable : array-like or None
+        The computed variable as a percentage, or `None` if the required data is
+        not available in the model object.
+
+    Raises
+    ------
+    None explicitly, but prints a warning if the requested variable cannot be computed.
+
+    Notes
+    -----
+    - 'land_ice_area_fraction' assumes all ice is land ice (returns 100% where data is available).
+    - 'floating_ice_shelf_area_fraction' computes the fraction of total ice that is floating.
+    - 'grounded_ice_sheet_area_fraction' computes the fraction of total ice that is grounded.
+    - Missing required attributes in `md` will result in the variable being skipped.
+    """
+
+    variable = None
+
+    if ismip_variable_name == 'land_ice_area_fraction':
+        if tools.general.has_nested_attr(md, 'results', 'TransientSolution', 'GroundedArea') and tools.general.has_nested_attr(md, 'results', 'TransientSolution', 'FloatingArea'):
+            total_area = md.results.TransientSolution.GroundedArea + md.results.TransientSolution.FloatingArea
+            variable = calc_perc_ice_cover(total_area, total_area)
+        else:
+            print("\033[1mland_ice_area_fraction\033[0m cannot be computed and will be skipped.")
+
+    elif ismip_variable_name == 'floating_ice_shelf_area_fraction':
+        if tools.general.has_nested_attr(md, 'results', 'TransientSolution', 'FloatingArea'):
+            total_area = md.results.TransientSolution.GroundedArea + md.results.TransientSolution.FloatingArea
+            variable = calc_perc_ice_cover(total_area, md.results.TransientSolution.FloatingArea)
+        else:
+            print("\033[1mfloating_ice_shelf_area_fraction\033[0m cannot be computed and will be skipped.")
+
+    elif ismip_variable_name == 'grounded_ice_sheet_area_fraction':
+        if tools.general.has_nested_attr(md, 'results', 'TransientSolution', 'GroundedArea'):
+            total_area = md.results.TransientSolution.GroundedArea + md.results.TransientSolution.FloatingArea
+            variable = calc_perc_ice_cover(total_area, md.results.TransientSolution.GroundedArea)
+        else:
+            print("\033[1mgrounded_ice_sheet_area_fraction\033[0m cannot be computed and will be skipped.")
+
+    else:
+        print(f"\033[1m{ismip_variable_name}\033[0m is not recognized and will be skipped.")
+
+    return variable

pyissm/data/__init__.py

@@ -0,0 +1 @@
+from . import interp

pyissm/data/interp.py

@@ -0,0 +1,256 @@
+"""
+Tools to interpolate data to/from ISSM model mesh.
+
+This module contains various interpolation functions that can be used in conjunction with ISSM models.
+"""
+
+import xarray as xr
+import numpy as np
+from scipy.interpolate import RegularGridInterpolator, griddata
+from pyissm import tools
+
+def xr_to_mesh(data,
+               var_name,
+               mesh_x,
+               mesh_y,
+               x_var = 'x',
+               y_var = 'y',
+               default_value = np.nan,
+               interpolation_type = 'bilinear',
+               issm_wrapper = True):
+    
+    """
+    Interpolate a variable from an xarray dataset onto mesh nodes.
+    
+    Assumes rectilinear (structured) grid in the xarray dataset.
+    
+    Parameters
+    ----------
+    data : str or xr.Dataset
+        Path to a netCDF file or an xarray Dataset containing the gridded data.
+    var_name : str
+        Name of the variable to interpolate.
+    mesh_x : ndarray
+        X-coordinates of mesh nodes.
+    mesh_y : ndarray
+        Y-coordinates of mesh nodes.
+    x_var : str, optional
+        Name of the x-coordinate variable in the dataset. Default is 'x'.
+    y_var : str, optional
+        Name of the y-coordinate variable in the dataset. Default is 'y'.
+    default_value : float, optional
+        Value to assign to points outside the grid domain. Default is np.nan.
+    interpolation_type : str, optional
+        Type of interpolation method. For ISSM wrapper: 'bilinear', 'nearest', etc.
+        For scipy: 'linear', 'nearest', 'slinear', 'cubic', 'quintic', 'pchip'.
+        Default is 'bilinear'.
+    issm_wrapper : bool, optional
+        If True, use ISSM wrapper functions for interpolation. If False, use scipy.
+        Default is True.
+    
+    Returns
+    -------
+    ndarray
+        Interpolated variable values at mesh nodes.
+    
+    Raises
+    ------
+    TypeError
+        If data is neither a file path nor an xarray Dataset.
+    ValueError
+        If variable is not 2D, coordinates are inconsistent, or grid is not rectilinear.
+    ImportError
+        If issm_wrapper is True but ISSM wrappers are not installed.
+    """
+
+    # Load xarray dataset if a filepath was given
+    if isinstance(data, str):
+        data = xr.open_dataset(data)
+        close = True
+    elif isinstance(data, xr.Dataset):
+        close = False
+    else:
+        raise TypeError("pyissm.data.interp.xr_to_mesh: data must be a file path or an xarray Dataset")
+
+    # Extract and squeeze arrays
+    x = np.asarray(data[x_var].values).squeeze()
+    y = np.asarray(data[y_var].values).squeeze()
+    var_data = np.asarray(data[var_name].values).squeeze()
+
+    if close:
+        data.close()
+
+    # Convert everything to float64 (but keep shapes)
+    x = x.astype(np.float64, copy = False)
+    y = y.astype(np.float64, copy = False)
+    var_data = var_data.astype(np.float64, copy = False)
+    mesh_x = mesh_x.astype(np.float64, copy = False)
+    mesh_y = mesh_y.astype(np.float64, copy = False)
+
+    # Check for rectilinear grid
+    if var_data.ndim != 2:
+        raise ValueError(f"pyissm.data.interp.xr_to_mesh: variable '{var_name}' must be 2D on a rectilinear grid")
+
+    # If coordinates are 1D arrays, check shapes
+    if x.ndim == 1 and y.ndim == 1:
+        if var_data.shape != (y.size, x.size):
+            raise ValueError(f"pyissm.data.interp.xr_to_mesh: variable '{var_name}' has shape {var_data.shape}, "
+                             f"expected ({y.size}, {x.size})")
+
+    # If coordinates are 2D arrays, check shapes and rectilinearity
+    # If 2D, they should be repeated 1D arrays
+    elif x.ndim == 2 and y.ndim == 2:
+        if x.shape != var_data.shape or y.shape != var_data.shape:
+            raise ValueError("pyissm.data.interp.xr_to_mesh: x, y, and variable must have identical shapes "
+                             "for 2D coordinate grids")
+
+        # Check rectilinearity
+        if not np.allclose(x, x[0, :][None, :]):
+            raise ValueError("pyissm.data.interp.xr_to_mesh: 2D x-coordinate is not rectilinear")
+
+        if not np.allclose(y, y[:, 0][:, None]):
+            raise ValueError("pyissm.data.interp.xr_to_mesh: 2D y-coordinate is not rectilinear")
+
+        # Reduce to 1D
+        x = x[0, :]
+        y = y[:, 0]
+
+    else:
+        raise ValueError(
+            "pyissm.data.interp.xr_to_mesh: x and y must both be 1D or both be 2D"
+        )
+
+
+    # Ensure monotonic increasing coordinates
+    if np.any(np.diff(x) < 0):
+        x = x[::-1]
+        var_data = var_data[:, ::-1]
+
+    if np.any(np.diff(y) < 0):
+        y = y[::-1]
+        var_data = var_data[::-1, :]
+
+    # If use_wrapper is True, use ISSM wrappers for interpolation
+    if issm_wrapper:
+        
+        # Check that wrappers are installed
+        if not tools.wrappers.check_wrappers_installed():
+            raise ImportError("pyissm.data.interp.xr_to_mesh: ISSM wrappers are not installed. Please install them or set issm_wrapper = False to use scipy interpolation.")
+        
+        # Interpolate using ISSM wrapper
+        var_on_mesh = tools.wrappers.InterpFromGridToMesh(
+            x,
+            y,
+            var_data,
+            mesh_x,
+            mesh_y,
+            default_value,
+            interpolation_type
+        )
+    
+    # Otherwise, use scipy for interpolation        
+    else:
+
+        scipy_method_list = ['linear', 'nearest', 'slinear', 'cubic', 'quintic', 'pchip']
+        if interpolation_type not in scipy_method_list:
+            raise ValueError(f"pyissm.data.interp.xr_to_mesh: interpolation_type '{interpolation_type}' is not supported by scipy. Choose from {scipy_method_list}.")
+
+        interp = RegularGridInterpolator(
+                (y, x),
+                var_data,
+                method = interpolation_type,
+                bounds_error = False,
+                fill_value = default_value,
+            )
+
+        # scipy expects points as (y, x)
+        mesh_points = np.column_stack((mesh_y, mesh_x))
+
+        # Extract variable on mesh
+        var_on_mesh = interp(mesh_points)
+
+    return var_on_mesh
+
+def points_to_mesh(data_x,
+                   data_y,
+                   data_values,
+                   mesh_x,
+                   mesh_y,
+                   default_value = np.nan,
+                   interpolation_type = 'linear'):
+    """
+    Interpolate scattered points onto mesh node coordinates using scipy.
+
+    Parameters
+    ----------
+    data_x : array_like
+        X-coordinates of the scattered data points. Can be 1D or 2D; if 2D it will be flattened.
+    data_y : array_like
+        Y-coordinates of the scattered data points. Must have the same shape as `data_x`.
+    data_values : array_like
+        Values at the scattered data points. Must have the same shape as `data_x`.
+    mesh_x : array_like
+        X-coordinates of mesh nodes. 1D array.
+    mesh_y : array_like
+        Y-coordinates of mesh nodes. 1D array.
+    default_value : float, optional
+        Value used to fill points outside the convex hull of the input data. Default is `np.nan`.
+    interpolation_type : str, optional
+        Interpolation method passed to `scipy.interpolate.griddata`. Supported options: 'linear', 'nearest', 'cubic'.
+        Default is 'linear'.
+
+    Returns
+    -------
+    ndarray
+        1D array of interpolated values at the mesh nodes (shape equals `mesh_x`/`mesh_y`).
+        Points outside the convex hull of the input data are assigned `default_value`.
+
+    Raises
+    ------
+    ValueError
+        If `interpolation_type` is not supported by scipy, if input shapes are inconsistent, or if no valid
+        data points remain after removing NaNs/Infs.
+    """
+
+    # Check interpolation type    
+    scipy_method_list = ['linear', 'nearest', 'cubic']
+    if interpolation_type not in scipy_method_list:
+        raise ValueError(f"pyissm.data.interp.points_to_mesh: interpolation_type '{interpolation_type}' is not supported by scipy. Choose from {scipy_method_list}.")
+    
+    # Validate input shapes
+    if data_x.shape != data_y.shape or data_x.shape != data_values.shape:
+        raise ValueError("pyissm.data.interp.points_to_mesh: data_x, data_y, and data_values must have the same shape.")
+    
+    # Flatten if needed:
+    if data_values.ndim == 2:
+        data_x = data_x.flatten()
+        data_y = data_y.flatten()
+        data_values = data_values.flatten()
+    elif data_values.ndim == 1:
+        pass
+    else:
+        raise ValueError("pyissm.data.interp.points_to_mesh: data_values must be 1D or 2D array.")
+    
+    # Remove NaNs/Inf from input data
+    mask = np.isfinite(data_x) & np.isfinite(data_y) & np.isfinite(data_values)
+
+    if not np.any(mask):
+        raise ValueError("pyissm.data.interp.points_to_mesh: no valid data points available for interpolation")
+
+    data_x = data_x[mask]
+    data_y = data_y[mask]
+    data_values = data_values[mask]
+
+    # Prepare points for interpolation
+    data_points = np.column_stack((data_y, data_x))
+    interp_points = np.column_stack((mesh_y, mesh_x))
+    
+    # Perform interpolation    
+    data_on_mesh = griddata(data_points,
+                            data_values,
+                            interp_points,
+                            method = interpolation_type,
+                            fill_value = default_value)
+
+    return data_on_mesh
+    

pyissm/learn/__init__.py

@@ -0,0 +1 @@
+from . import tutorials

pyissm/learn/tutorials.py

@@ -0,0 +1,3 @@
+"""
+ISSM tutorials and help functions
+"""