Merge pull request #180 from GispoCoding/178-fix-kriging-interpolatio…

…n-grid-creation

178 fix kriging interpolation grid creation

eis_toolkit/vector_processing/kriging_interpolation.py

@@ -2,11 +2,11 @@
 
 import geopandas as gpd
 import numpy as np
-import numpy.ma as ma
 from beartype import beartype
 from beartype.typing import Literal, Optional, Tuple
 from pykrige.ok import OrdinaryKriging
 from pykrige.uk import UniversalKriging
+from rasterio import transform
 
 from eis_toolkit.exceptions import EmptyDataFrameException, InvalidParameterValueException
 
@@ -17,13 +17,13 @@ def _kriging(
     resolution: Tuple[Number, Number],
     extent: Optional[Tuple[Number, Number, Number, Number]],
     variogram_model: Literal,
+    coordinates_type: Literal,
     method: Literal,
     drift_terms: list,
 ) -> Tuple[np.ndarray, dict]:
 
-    points = np.array(list(data.geometry.apply(lambda geom: [geom.x, geom.y])))
-    x = points[:, 0]
-    y = points[:, 1]
+    x = data.geometry.x
+    y = data.geometry.y
     z = data[target_column].values
 
     if extent is None:
@@ -35,20 +35,23 @@ def _kriging(
     else:
         grid_x_min, grid_x_max, grid_y_min, grid_y_max = extent
 
-    grid_x = np.linspace(grid_x_min, grid_x_max, resolution[0])
-    grid_y = np.linspace(grid_y_min, grid_y_max, resolution[1])
+    grid_x = np.arange(grid_x_min, grid_x_max + resolution[0], resolution[0])
+    grid_y = np.arange(grid_y_min, grid_y_max + resolution[1], resolution[1])
 
     if method == "universal":
         universal_kriging = UniversalKriging(x, y, z, variogram_model=variogram_model, drift_terms=drift_terms)
         z_interpolated, _ = universal_kriging.execute("grid", grid_x, grid_y)
 
     if method == "ordinary":
-        ordinary_kriging = OrdinaryKriging(x, y, z, variogram_model=variogram_model)
+        ordinary_kriging = OrdinaryKriging(x, y, z, variogram_model=variogram_model, coordinates_type=coordinates_type)
         z_interpolated, _ = ordinary_kriging.execute("grid", grid_x, grid_y)
 
-    z_interpolated = ma.getdata(z_interpolated)
-
-    out_meta = {"crs": data.crs, "width": len(grid_x), "height": len(grid_y)}
+    out_meta = {
+        "crs": data.crs,
+        "width": len(grid_x),
+        "height": len(grid_y),
+        "transform": transform.from_bounds(grid_x_min, grid_y_min, grid_x_max, grid_y_max, len(grid_x), len(grid_y)),
+    }
 
     return z_interpolated, out_meta
 
@@ -60,6 +63,7 @@ def kriging(
     resolution: Tuple[Number, Number],
     extent: Optional[Tuple[Number, Number, Number, Number]] = None,
     variogram_model: Literal["linear", "power", "gaussian", "spherical", "exponential", "hole-effect"] = "linear",
+    coordinates_type: Literal["euclidean", "geographic"] = "geographic",
     method: Literal["ordinary", "universal"] = "ordinary",
     drift_terms: list = ["regional_linear"],
 ) -> Tuple[np.ndarray, dict]:
@@ -69,10 +73,12 @@ def kriging(
     Args:
         data: GeoDataFrame containing the input data.
         target_column: The column name with values for each geometry.
-        resolution: Size of the output grid.
+        resolution: The resolution i.e. cell size of the output raster.
         extent: The extent of the output grid.
             If None, calculate extent from the input vector data.
         variogram_model: Variogram model to be used. Defaults to 'linear'.
+        coordinates_type: Determines are coordinates on a plane ('euclidean') or a sphere ('geographic').
+            Defaults to 'geographic'.
         method: Kriging method. Defaults to 'ordinary'.
         drift_terms: Drift terms used in universal kriging.
 
@@ -103,7 +109,7 @@ def kriging(
         )
 
     data_interpolated, out_meta = _kriging(
-        data, target_column, resolution, extent, variogram_model, method, drift_terms
+        data, target_column, resolution, extent, variogram_model, coordinates_type, method, drift_terms
     )
 
     return data_interpolated, out_meta