Cartopy - Python for climatology, oceanograpy and atmospheric science

Wind roses are a powerful tool for visualizing wind patterns, but what if you could overlay them on real-world maps for better geographic context? In this week's MetPy Monday, we take wind data analysis to the next level by plotting wind roses on OpenStreetMap layers using CartoPy. By integrating Python's windrose package with CartoPy image tiles, we create a geospatial visualization that combines meteorology and mapping. In this tutorial, you'll learn how to extract wind speed and direction data, generate wind rose plots, and overlay them onto OpenStreetMap using customized map layers. Whether you're working in meteorology, environmental analysis, or GIS, this approach provides deeper insight into regional wind patterns. Watch the full video to see how Python can help you bridge the gap between weather data and spatial analysis!

Examples

GOES IR images using Cartopy

https://nbviewer.jupyter.org/gist/ajdawson/8637019

python4oceanographers

Using SRTM data to correct GPS altimetry

https://ocefpaf.github.io/python4oceanographers/blog/2014/09/08/srtm/

Plotting shapefiles with cartopy and folium

http://ocefpaf.github.io/python4oceanographers/blog/2015/02/02/cartopy_folium_shapefile/

Qiitaのcartopyタグ記事

https://qiita.com/tags/cartopy

Ocean Science Hack

surface drifter (in Japanese)

http://oceansciencehack.blogspot.com/2014/09/surface-drifter.html

Surface velocity data of MOVE/MRI.com at NEAR-GOOS RRTDB (in Japanese)

http://oceansciencehack.blogspot.com/2014/11/surface-velocity-data-of-movemricom-at.html

Tips

python - Increasing resolution of Cartopy stock background? - Geographic Information Systems Stack Exchange

https://gis.stackexchange.com/questions/313490/increasing-resolution-of-cartopy-stock-background

python - Download data from Natural Earth and OpenStreetMap for Cartopy - Stack Overflow

https://stackoverflow.com/questions/35707514/download-data-from-natural-earth-and-openstreetmap-for-cartopy?rq=1

Cartopy Performance Optimization: transform_first

What is transform_first?

A keyword argument in Cartopy plotting functions (e.g., contourf, pcolormesh) that changes the timing of coordinate transformation.

Default (False): Transforms every single data point/polygon from source to target projection. High precision, but high CPU load for large grids.

True: Renders data as an image in the source projection first, then warps the resulting image to the target projection. Significant speedup for high-resolution 2D grids.

Case-by-Case Analysis

table: Cases

Case Transformation Speedup Verdict

1 Different (e.g., Robinson) High Recommended for massive grids

2 Same (PlateCarree -> PC) None Not necessary

3 Shifted (PC -> PC 180) High Useful to bypass wrapping math

Sample Code

code:python

import cartopy.crs as ccrs

import matplotlib.pyplot as plt

ax = plt.axes(projection=ccrs.Robinson())

plt.contourf(lons, lats, data, transform=ccrs.PlateCarree(), transform_first=True)

Disadvantages and Cautions: The "Seam" Issue

When changing central_longitude, using transform_first=True may cause a thin, blank vertical line to appear at the new map boundaries (the "seam" where the data should be continuous).

This is caused by sampling errors that occur when the rendered image is "torn" and rearranged to fit the new projection. While applying cartopy.util.add_cyclic_point to your data is the standard way to ensure periodicity, its effect may be diminished or bypassed when transform_first=True is enabled, as the transformation happens at the image level rather than the coordinate level.

# N