Skip to content

FAQ & Troubleshooting

Installation

ModuleNotFoundError: No module named 'pykrige'

Kriging and ML methods require the kriging extra: 

pip install "geointerpo[kriging]"

ModuleNotFoundError: No module named 'rioxarray'

GeoTIFF export and boundary clipping require the raster extra: 

pip install "geointerpo[raster]"

ModuleNotFoundError: No module named 'matplotlib'

Plotting requires the viz extra: 

pip install "geointerpo[viz]"

ModuleNotFoundError: No module named 'plotly'

Interactive maps require the interactive extra: 

pip install "geointerpo[interactive]"
# or just: pip install plotly

ModuleNotFoundError: No module named 'gstools'

Cokriging and Sequential Gaussian Simulation require the geostat extra: 

pip install "geointerpo[geostat]"

ModuleNotFoundError: No module named 'mapie'

Conformal prediction intervals for GBM require the uncertainty extra: 

pip install "geointerpo[uncertainty]"
RF bootstrap intervals are built-in and need no extra install.

ModuleNotFoundError: No module named 'cdsapi'

ERA5 reanalysis requires the era5 extra plus a free CDS API account: 

pip install "geointerpo[era5]"
Then create ~/.cdsapirc following the CDS API setup guide.

Boundaries

ValueError: Could not geocode location 'X'

Nominatim couldn't find the place name. Try:

  • A more specific name: "Calgary, Alberta, Canada" instead of "Calgary"
  • A bbox tuple: boundary=(-114.5, 50.8, -113.8, 51.3)
  • A file: boundary="my_region.geojson"
Boundary clips everything to NaN

The boundary polygon may not overlap your station data. Check:

  1. boundary_bbox(boundary) returns a bbox that covers your data
  2. rioxarray is installed: pip install "geointerpo[raster]"
  3. Try clip_to_boundary=False to isolate the issue

Data sources

No stations returned from Meteostat

Meteostat may have no stations for that bbox/date. Try:

  • A wider bbox (increase padding_deg)
  • A different date (some stations have gaps)
  • data="sample" to confirm the pipeline works offline
ImportError: No module named 'meteostat' 
pip install "geointerpo[data]"

Interpolation

ValueError: Not enough points for spline

Spline requires at least 16 points. Either add more stations or use method="rbf" or method="idw".

kriging returns all NaN

Check that pykrige is installed and try variogram_model="linear" as a safe fallback.

natural_neighbor output looks like IDW

With very sparse data most grid points fall outside the station convex hull, so the IDW fallback activates. This is correct behaviour — the Voronoi weights are only applied to interior points.

Plotting

RuntimeError: Colorbar layout not compatible with tight_layout

Use layout="constrained" in your figure: 

fig, ax = plt.subplots(layout="constrained")

Interactive map doesn't display in Jupyter 
pip install "geointerpo[notebooks]"

Performance

Speed up large grids

  • Increase resolution (0.5° instead of 0.1°) for a quick first pass
  • natural_neighbor is O(n_stations × n_grid) — use idw or rbf for large grids
  • GP scales as O(n³) — prefer rf or gbm for more than ~500 stations