Fix bounding box clipping bug #50 (#55)

* Fix bbox clipping bug #50

* Update lockfile

* Updates for compat with pytmd==2.2.2

* Fix time

* New version 0.6.1

* Fix wording

* Add test of global clipping options

* Autouse synethic hamtide fixture

* Fix output path

docs/changelog.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## v0.6.1
+
+### Bug fixes
+
+- Fixed bug causing tide model clipping with `clip_tides` to fail for bounding boxes completely west of the prime meridian ([#50](https://github.com/GeoscienceAustralia/eo-tides/issues/50))
+
 ## v0.6.0
 
 ### New features

eo_tides/model.py

@@ -19,6 +19,7 @@
 import pandas as pd
 import pyproj
 import pyTMD
+import timescale.time
 from tqdm import tqdm
 
 from .utils import DatetimeLike, _set_directory, _standardise_models, _standardise_time, idw
@@ -95,7 +96,7 @@ def _model_tides(
     lon, lat = transformer.transform(x.flatten(), y.flatten())
 
     # Convert datetime
-    timescale = pyTMD.time.timescale().from_datetime(time.flatten())
+    ts = timescale.time.Timescale().from_datetime(time.flatten())
 
     try:
         # Read tidal constants and interpolate to grid points
@@ -136,10 +137,10 @@ def _model_tides(
     # Compute delta times based on model
     if pytmd_model.corrections in ("OTIS", "ATLAS", "TMD3", "netcdf"):
         # Use delta time at 2000.0 to match TMD outputs
-        deltat = np.zeros_like(timescale.tt_ut1)
+        deltat = np.zeros_like(ts.tt_ut1)
     else:
         # Use interpolated delta times
-        deltat = timescale.tt_ut1
+        deltat = ts.tt_ut1
 
     # In "one-to-many" mode, extracted tidal constituents and timesteps
     # are repeated/multiplied out to match the number of input points and
@@ -149,7 +150,7 @@ def _model_tides(
     points_repeat = len(x) if mode == "one-to-many" else 1
     time_repeat = len(time) if mode == "one-to-many" else 1
     t, hc, deltat = (
-        np.tile(timescale.tide, points_repeat),
+        np.tile(ts.tide, points_repeat),
         hc.repeat(time_repeat, axis=0),
         np.tile(deltat, points_repeat),
     )

eo_tides/utils.py

@@ -181,9 +181,9 @@ def _clip_model_file(
     """
     Clips tide model netCDF datasets to a bounding box.
 
-    If the bounding box crosses 0 degrees longitude (e.g. Greenwich),
-    the function will clip the dataset into two parts and concatenate
-    them along the x-dimension to create a continuous result.
+    If the bounding box crosses 0 degrees longitude (e.g. Greenwich prime
+    meridian), the dataset will be clipped into two parts and concatenated
+    along the x-dimension to create a continuous result.
 
     Parameters
     ----------
@@ -226,20 +226,22 @@ def _clip_model_file(
     xcoords = nc[xcoord].compute()
     ycoords = nc[ycoord].compute()
 
-    # If data falls within 0-360 degree bounds, then clip directly
-    if (bbox.left >= 0) & (bbox.right <= 360):
+    # Convert longitudes to 0-360 convention
+    left = bbox.left % 360
+    right = bbox.right % 360
+
+    # If left coordinate is smaller than right, bbox does not cross
+    # zero longitude and can be clipped directly
+    if left <= right:  # bbox does not cross 0
         nc_clipped = nc.sel({
             ydim: (ycoords >= bbox.bottom) & (ycoords <= bbox.top),
-            xdim: (xcoords >= bbox.left) & (xcoords <= bbox.right),
+            xdim: (xcoords >= left) & (xcoords <= right),
         })
 
-    # If bbox crosses zero longitude, extract left and right
-    # separately and then combine into one concatenated dataset
-    elif (bbox.left < 0) & (bbox.right > 0):
-        # Convert longitudes to 0-360 range
-        left = bbox.left % 360
-        right = bbox.right % 360
-
+    # If left coordinate is larger than right, bbox crosses zero longitude.
+    # If so, extract left and right separately and then combine into one
+    # concatenated dataset
+    elif left > right:  # bbox crosses 0
         # Extract data from left of 0 longitude, and convert lon
         # coords to -180 to 0 range to enable continuous interpolation
         # across 0 boundary
@@ -357,7 +359,7 @@ def clip_models(
         model_files = model_database[m].get("model_file", [])
         grid_file = model_database[m].get("grid_file", [])
 
-        # Convert to list if strings and combine
+        # Convert to list of strings and combine
         model_files = model_files if isinstance(model_files, list) else [model_files]
         grid_file = grid_file if isinstance(grid_file, list) else [grid_file]
         all_files = model_files + grid_file

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "eo-tides"
-version = "0.6.0"
+version = "0.6.1"
 description = "Tide modelling tools for large-scale satellite earth observation analysis"
 authors = [
     { name = "Robbi Bishop-Taylor" },
@@ -44,10 +44,11 @@ dependencies = [
     "psutil>=5.8.0",
     "pyogrio>=0.10.0",
     "pyproj>=3.7.0",
-    "pyTMD==2.2.1",
+    "pyTMD>=2.2.2",
     "scikit-learn>=1.4.0",
     "scipy>=1.14.1",
     "shapely>=2.0.6",
+    "timescale>=0.0.3",
     "tqdm>=4.55.0",
     "xarray>=2022.3.0",
 ]

tests/conftest.py

@@ -3,11 +3,14 @@
 """
 
 from copy import deepcopy
+from pathlib import Path
 
+import numpy as np
 import odc.stac
 import pandas as pd
 import pystac_client
 import pytest
+import xarray as xr
 
 GAUGE_X = 122.2183
 GAUGE_Y = -18.0008
@@ -97,3 +100,45 @@ def satellite_ds(satellite_ds_load):
     each test to ensure each test is independent
     """
     return deepcopy(satellite_ds_load)
+
+
+# Run once per session to generate symethic HAMTIDE11 files; autouse=True
+# allows this to run without being specifically called in tests
+@pytest.fixture(scope="session", autouse=True)
+def create_synthetic_model(base_dir="tests/data/tide_models_synthetic"):
+    """
+    Generates and exports synthetic HAMTIDE11 model data
+    to test clipping functionality.
+    """
+    base_dir = Path(base_dir)  # Ensure base_dir is a Path object
+
+    # Create coordinate arrays
+    lon = np.arange(0, 360.125, 0.125)  # 2881 points
+    lat = np.arange(-90, 90.125, 0.125)  # 1441 points
+
+    # List of hamtide tidal constituents
+    constituents = ["2n", "k1", "k2", "m2", "n2", "o1", "p1", "q1", "s2"]
+
+    # Create hamtide output directory
+    hamtide_dir = base_dir / "hamtide"
+    hamtide_dir.mkdir(parents=True, exist_ok=True)
+
+    # Create and save a NetCDF for each constituent
+    for constituent in constituents:
+        # Create synthetic hamtide dataset with random data
+        shape = (len(lat), len(lon))  # 1441, 2881
+        data = np.random.random(shape).astype(np.float32)
+        ds = xr.Dataset(
+            {
+                "RE": (("LAT", "LON"), data),
+                "IM": (("LAT", "LON"), data),
+                "AMPL": (("LAT", "LON"), data),
+                "PHAS": (("LAT", "LON"), data),
+            },
+            coords={"LON": lon, "LAT": lat},
+            attrs={"title": f"HAMTIDE11a: {constituent} ocean tide"},
+        )
+
+        # Export
+        filename = hamtide_dir / f"{constituent}.hamtide11a.nc"
+        ds.to_netcdf(filename)

tests/test_utils.py

@@ -3,9 +3,11 @@
 from datetime import datetime
 
 import numpy as np
+import odc.geo.geom
 import pandas as pd
 import pytest
 
+from eo_tides.model import model_tides
 from eo_tides.utils import _standardise_models, _standardise_time, clip_models, idw, list_models
 
 
@@ -53,17 +55,96 @@ def test_standardise_models(model, ensemble_models, exp_process, exp_request, ex
 
 
 def test_clip_models():
+    # Set input and output paths
+    in_dir = "tests/data/tide_models"
     out_dir = pathlib.Path("tests/data/tide_models_clipped")
 
+    # Clip models to bbox
     clip_models(
-        input_directory="tests/data/tide_models",
+        input_directory=in_dir,
         output_directory=out_dir,
         bbox=(122.27, -18.07, 122.29, -18.05),
     )
 
+    # Assert that files were exported for all available models
     output_files = set([i.stem for i in out_dir.iterdir()])
     assert output_files == set(["GOT5", "EOT20", "hamtide"])
 
+    # Set modelling location
+    x, y = 122.28, -18.06
+    time = pd.date_range(start="2000-01", end="2001-03", freq="5h")
+
+    # Model using unclipped vs clipped files
+    df_unclipped = model_tides(
+        x=x,
+        y=y,
+        time=time,
+        model="HAMTIDE11",
+        directory=in_dir,
+        crop=False,
+    )
+    df_clipped = model_tides(
+        x=x,
+        y=y,
+        time=time,
+        model="HAMTIDE11",
+        directory=out_dir,
+        crop=False,
+    )
+
+    # Verify both produce the same results
+    assert np.allclose(df_unclipped.tide_height, df_clipped.tide_height)
+
+
+# Test clipping across multiple global locations using synthetic HAMTIDE11 data
+@pytest.mark.parametrize(
+    "bbox, name",
+    [
+        ((-166, 14, -151, 29), "hawaii"),  # entirely W of prime meridian
+        ((-13, 49, 6, 60), "uk"),  # crossing prime meridian
+        ((105.292969, -47.872144, 160.312500, -5.266008), "aus"),  # entirely E of prime meridian
+        ((-256.640625, 7.013668, -119.794922, 63.391522), "pacific"),  # crossing antimeridian
+    ],
+)
+def test_clip_models_bbox(bbox, name):
+    # Set input and output paths
+    in_dir = "tests/data/tide_models_synthetic/"
+    out_dir = f"tests/data/tide_models_synthetic_{name}/"
+
+    # Clip models to input bbox
+    clip_models(
+        input_directory=in_dir,
+        output_directory=out_dir,
+        bbox=bbox,
+        model="HAMTIDE11",
+        overwrite=True,
+    )
+
+    # Set modelling location based on bbox centroid
+    x, y = odc.geo.geom.BoundingBox(*bbox, crs="EPSG:4326").polygon.centroid.xy
+    time = pd.date_range(start="2000-01", end="2001-03", freq="5h")
+
+    # Model using unclipped vs clipped files
+    df_unclipped = model_tides(
+        x=x,
+        y=y,
+        time=time,
+        model="HAMTIDE11",
+        directory=in_dir,
+        crop=False,
+    )
+    df_clipped = model_tides(
+        x=x,
+        y=y,
+        time=time,
+        model="HAMTIDE11",
+        directory=out_dir,
+        crop=False,
+    )
+
+    # Verify both produce the same results
+    assert np.allclose(df_unclipped.tide_height, df_clipped.tide_height)
+
 
 @pytest.mark.parametrize(
     "input_value, expected_output",