Issue: geemap cartoee.get_map() ignores region parameter and results in low resolution when using ax.set_extent()

[chenbadan]

When using cartoee.get_map() with a specified region parameter, the function still renders the entire global extent instead of the specified region. Additionally, when manually setting the extent with ax.set_extent(), the final output image has very low resolution, as the initial rendering is global and then cropped.

Code to Reproduce

import ee
import ee.mapclient
from matplotlib import pyplot as plt
from geemap import cartoee
import cartopy.crs as ccrs

# Initialize Earth Engine
ee.Initialize()

projection = ccrs.LambertConformal(
    central_longitude=105,
    central_latitude=40,
    standard_parallels=(25.0, 47.0)
)

ocean = (
    ee.ImageCollection("NASA/OCEANDATA/MODIS-Terra/L3SMI")
    .filter(ee.Filter.date("2018-01-01", "2018-03-01"))
    .median()
    .select(["sst"], ["SST"])
)

visualization = {"bands": "SST", "min": -2, "max": 30}
bbox = [70, 15, 140, 55]  # Expected region: East Asia

plt.rcParams['font.family'] = 'Times New Roman'
fig = plt.figure(figsize=(3.5, 3), dpi=300)

# Problem 1: `region=bbox` is ignored, global map is rendered
ax = cartoee.get_map(
    ocean,
    vis_params=visualization,
    cmap='plasma',
    proj=projection,
    region=bbox
)

# Problem 2: Using `ax.set_extent()` fixes the region but results in low resolution
# ax.set_extent([70, 140, 5, 55], crs=ccrs.PlateCarree())

plt.show()

Environment Information

Actual Behavior

1. The region parameter is ignored, and the entire globe is rendered (as shown in the attached screenshot, where only a small portion of the global Lambert projection contains data).
2. When ax.set_extent() is used to manually crop the plot, the final image resolution is drastically reduced because the initial rendering was global and then scaled down.

Possible Cause

It appears that the region parameter is not being properly passed to the underlying Earth Engine getThumbURL or image export logic, causing the function to default to the global extent of the projection. When ax.set_extent() is used, the matplotlib figure is cropped after the low-resolution global image has been rendered, leading to pixelation.

Suggested Fix

Ensure that the region parameter is correctly converted into the appropriate coordinate system for the given projection and used to clip the Earth Engine image before rendering, so that only the required area is fetched and rendered at full resolution.

[cpicazo8304]

Hi! I'm a student in the CodePath AI301 program and I'd like to work on this issue. I'll start by reproducing the problem locally this week. Any pointers on where to start in the codebase would be appreciated!

[chenbadan]

Hi! Thanks for volunteering to tackle this issue.
Starting points for debugging
Open geemap/cartoee.py and locate the get_map() function.
Inspect how the region parameter is handled when requesting image thumbnails from GEE — currently the region is unused in the render request, so the globe is rendered by default.
The low resolution after ax.set_extent() is a side effect: the full low-res global raster is downloaded first, then cropped locally, rather than fetching a high-res raster limited to your region.
Reproduction hint
Test with a small bounding box ROI, log the thumbnail API request payload inside get_map() to confirm the missing region argument.
Feel free to reach out if you need more guidance during debugging!

[alepr]

Hi there! The add_layer function in cartoee seems to be the issue. It defaults ax.imshow() to the ccrs.PlateCarree() coordinate system, which causes rendering errors when a different projection is intended. If you agree, I’m happy to submit a pull request to fix this. When using the original projection, the figure looks like this: