Published July 29, 2025 | Version v1
Dataset Open

Supplemental Training Coordinates

Description

Overview

This dataset was developed to provide a representative sample of Earth’s terrestrial land surface and near-shore ecosystems while optimizing for coverage across space, time, and availability of data sources. Our sampling strategy prioritizes coverage of locations where we have geocoded text information by first taking a gridded sample that covers point locations for geotagged features from Wikipedia and GBIF species observations (1). To cover the remaining land surface, we use the 2017 RESOLVE Ecoregions dataset ("RESOLVE/ECOREGIONS/2017" in the Earth Engine Data Catalog) (2) to draw an additional random stratified sample by ecoregion ID. We supplement our initial RESOLVE sample, which largely targets terrestrial ecosystems, with additional stratified samples from the Allen Coral Atlas (3) and Global Intertidal Zones datasets (4) to improve representation of near-shore ecosystems. We sample 4,141 locations from the Allen Coral Atlas ("ACA/reef_habitat/v2_0") and 2,968 from the Murray Global Intertidal dataset ("UQ/murray/Intertidal/v1_1/global_intertidal"). We ensure a minimum distance of 1.28 km between sampled locations, and we sample two year-long periods for each location. After culling locations with insufficient image availability, the final dataset hosted here includes 8,412,511 unique (x, y, t_start, t_end) rows that can be used to query imagery from publicly available image collections.

License

Copyright 2025 Google LLC

All software is licensed under the Apache License, Version 2.0 (Apache 2.0); you may not use this file except in compliance with the Apache 2.0 license. You may obtain a copy of the Apache 2.0 license at: https://www.apache.org/licenses/LICENSE-2.0 All other materials are licensed under the Creative Commons Attribution 4.0 International License (CC-BY). You may obtain a copy of the CC-BY license at: https://creativecommons.org/licenses/by/4.0/legalcode

Unless required by applicable law or agreed to in writing, all software and materials distributed here under the Apache 2.0 or CC-BY licenses are distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the licenses for the specific language governing permissions and limitations under those licenses.

This is not an official Google product.

Files

training_sites.zip

Files (266.6 MB)

Name Size Download all
md5:4ba3010f2e556072acce7f402768ad0a
266.6 MB Preview Download