Troubleshooting

Output Is Invalid

• Check 1: Are the coordinates over land or outside the model domain?

Most ocean tide models only have data over the ocean, which may not extend to all coastal areas. Any point interpolated over land, inland water, or beyond the model’s geographic boundary would return as NaN. For points at or near the coast: can try using one of the extrapolation options.

Warning

Extrapolated values have degraded accuracy, and may lead to misleading results. Both bathymetry and the extrapolation distance need to be carefully considered when using these options.

Additionally, some models are purely regional and cover only part of the globe. For points outside the domain of a regional model: probably need to use a different model. Post on the pyTMD discussions board if you want help choosing a model.

• Check 2: Are the coordinates in a different projection or coordinate reference system?

Some models are defined in a projected coordinate system rather than geographic coordinates (longitude, latitude). Other models are defined in geographic coordinates but use a different longitudinal convention (e.g. [-180, 180] vs. [0, 360]). In either case: transform the coordinates to the model coordinate reference system.

m = pyTMD.io.model().from_database("GOT4.10_nc")
ds = m.open_dataset()
x, y = ds.tmd.transform_as(lon, lat, crs=4326)

Output Looks Inaccurate

• Check 1: Is the time standard and zone correct?

The most common cause of a systematic phase shift is passing times in the wrong time zone or time standard. pyTMD uses times in UTC for its astronomical calculations, and so either condition would cause the predicted tides to be out of phase with the observed tides.

• Check 2: Are the coordinates in a different projection or coordinate reference system?

Similar to the case that returned NaN, the returned values may be finite but mapped to an incorrect location. Verify your coordinates are in the model’s coordinate reference system.

• Check 3: Is the extrapolation distance too large?

If extrapolating over an extended distance (e.g. np.inf), the returned values may be interpolated from a location far from the intended point [see Interpolation and Extrapolation].

Model Not Found

If the model files are not found, try the following checks:

• Check 1: Is the model in the database?

• Check 2: Are the model files downloaded locally?

• Check 3: Is the directory set correctly (if using a custom model directory)?

• Check 4: Does the directory path exist and contain the expected files? [see Elevation Directory Table and Current Directory Table]

Predictions Are Slow

If the predictions are taking a long time (or there are memory issues), try the following fixes:

• Fix 1: Crop the model to the region of interest

• Fix 2: Read the model in chunks

• Fix 3: Use nearest-neighbour interpolation

• Fix 4: Reduce the constituent list (will likely reduce accuracy)

Getting Help

If this page does not resolve your problem:

1. Search existing Discussions and Issues

2. Open a new Discussion for usage questions or Issue for potential problems [see Guidelines]