OverflowError: Python int too large to convert to C long

[scottyhq]

Getting this error trying to sample "esa-worldcover-10meter"

from sliderule import raster
import geopandas as gpd

region = [  {"lon": -108.34, "lat": 38.89},
            {"lon": -107.76, "lat": 38.90},
            {"lon": -107.78, "lat": 39.26},
            {"lon": -108.36, "lat": 39.25},
            {"lon": -108.34, "lat": 38.89}  ]

gfp = gpd.GeoDataFrame(geometry=gpd.points_from_xy([-108.1, -108.3], [39.1, 39.2]), crs='EPSG:4326')
points = [[x,y] for x,y in zip(gfp.geometry.x , gfp.geometry.y)]
gdf = raster.sample("esa-worldcover-10meter", points)

---------------------------------------------------------------------------
OverflowError                             Traceback (most recent call last)
Cell In[1], line 12
     10 gfp = gpd.GeoDataFrame(geometry=gpd.points_from_xy([-108.1, -108.3], [39.1, 39.2]), crs='EPSG:4326')
     11 points = [[x,y] for x,y in zip(gfp.geometry.x , gfp.geometry.y)]
---> 12 gdf = raster.sample("esa-worldcover-10meter", points)
     13 gdf

File ~/GitHub/uw-cryo/coincident/.pixi/envs/dev/lib/python3.12/site-packages/sliderule/raster.py:145, in sample(asset, coordinates, parms)
    143 per_coord_index += 1
    144 for raster_sample in input_coord_response:
--> 145     columns['time'][i] = numpy.int64((raster_sample['time'] + 315964800.0) * 1000000000)
    146     columns['longitude'][i] = numpy.double(longitude)
    147     columns['latitude'][i] = numpy.double(latitude)

OverflowError: Python int too large to convert to C long

[elidwa]

Fixed in PR #447

[scottyhq]

Thanks @elidwa ! Looks like this can be closed after a 4.8.15 release?

[scottyhq]

On the topic, I'm also unsure what is the worldcover.time unit for 1309046418000.0? I'm not sure how to convert that to a human readable timestamp...

[elidwa]

@scottyhq the worldcover.time value 1309046418000.0 is GPS time in seconds. Unfortunately, Open Data on AWS only hosts TIFF files and does not provide any STAC endpoint or VRT file. Because of this, we generate the VRT file from their data and host it in our SlideRule S3 buckets. The process for this is detailed in the README file, with the top sections specifically describing the ESA worldcover portion:
https://github.com/SlideRuleEarth/sliderule/blob/main/datasets/opendata/utils/build_vrt_readme.txt

Since we create the VRT, the date of creation would not make much sense as a time reference. Instead, for time, we chose the midpoint of 2021.

1309046418000.0 = TimeLib::datetime2gps(2021, 06, 30, 0, 0, 0) / 1000

[scottyhq]

Closed for server version>=4.8.15