Dataset description fields

If you have a dataset uploaded into Earth Engine and you'd like to mirror it into the Earth Engine catalog, we ask you to fill in a form specifying various fields for the dataset description. This document guides you through this form.

Earth Engine data structure and dataset ID:

We recommend creating as few assets as possible. Generally, all data from a single observation or computation should be in the same asset. Limiting the number of datasets makes navigating the catalog easier.

We recommend that you include a version number like v1 into the ID, even if you don't plan to produce more versions. If a version number is not applicable, we suggest using the publication date of the dataset.

Sometimes, there are multiple but related families of datasets that differ in their permutations of parameters, e.g., due to different simulation algorithms, different geographical coverage, different input variables, etc. These variations might affect how you should name your Earth Engine asset.

The decision of whether to use multiple levels of folders (e.g. ORG/A/B/C/D/V001/20180310) vs. collapsing levels into one folder (e.g. ORG/A_B_C_D/V001/20180310) also depends on how likely it is that there will be more datasets of the same type. If it is likely that there will be a new dataset with different values, use multiple levels (e.g. FOO/E/F/G/V001/20180310). If it is unlikely that there will be more datasets in the future, then it is okay to collapse levels.

If your data differ in their permutations of parameters and you feel it is necessary to split it into multiple assets, be sure to group all relevant bands together. For instance, a dataset with temperature, humidity, and vegetation type may be broken into two assets, with temperature and humidity in one asset and the vegetation type in another.

Example combined and divided assets:

Feature Collections and Image Collections

Feature Collections and Image Collections can be used to group related features or images.

Examples of Feature Collections and Image Collections:

  • BLM AIM TerrADat TerrestrialAIM Point v1 Combines related table data into a FeatureCollection. Since there is no easy way to make continuous updates to tables, each update should be a new FeatureCollection.
  • LANDFIRE covers the entire CONUS, Alaska,and Hawaii. Instead of ingesting them together into one asset, we had to put the assets into an ImageCollection, as the source files have different projections.

Dataset title

A short, descriptive, plain-text string. It will appear at the top of the dataset page.

Example: Hansen Global Forest Change v1.7 (2000-2019)

To help distinguish between similarly named datasets, we recommend making the first 50 characters unique. This will help users find the appropriate dataset more easily in search results.

Provider name

Specify the dataset provider name. Use multiple provider names if the dataset is produced by a collaboration of different organizations.

Example: European Union/ESA/Copernicus

Dataset description URL

Add a link to the main web page of the dataset.

File download URL

Add a link to the location where the data can be downloaded directly, if applicable.

Dataset start and end dates

Earth Engine requires a start and end date to be set. It’s recommended that the dates represent the observation interval. If this is an ongoing dataset, you can leave the end date blank.

There are a few cases when it's not obvious what the dates are; e.g. a 30-year average temperature dataset, in which case we set the start and end dates to span the 30 years.

Dataset coverage area

Please choose either GLOBAL or CUSTOM_EXTENT. For a dataset to be considered GLOBAL, it should either overlap the antimeridian or cover more than half of the globe.

If you choose CUSTOM_EXTENT, you will be guided to a separate page where you can specify a bounding box. This box can be larger than the actual coverage area, but please make sure that it does not overlap the antimeridian. The geojson.io site can be useful to quickly look up coordinates of points on a map.

Dataset description

This should contain enough information for a new user to get started.

Please keep the description limited to no more than five paragraphs. The first paragraph should stand on its own to briefly describe the dataset. Links to more detailed documentation are welcome.

Description can use markdown for links and bullet points.

The description should not duplicate information from fields found elsewhere in this form.

Tags

We use tags to categorize the data in the Earth Engine Data Catalog. This will allow users to find your data by browsing related datasets.

Tags can be any words - e.g., "carbon", "pollution", "ocean", "landcover", "nasa", etc.

Please see the list of existing tags and try to use them.

Dataset license (terms of use)

If you have already chosen the terms of use, please specify them in this section. We recommended applying the CC-BY-4 license. More restrictive terms will need to be discussed with the Earth Engine Data team.

Please specify a link to the page on the main site of your dataset where the license is listed.

Citations

List the requirements of how and when the data should be cited. Use citation APA format if applicable.

Examples from existing datasets:

Use the format https://doi.org/10.7265/N5V98602

Raster bands or vector properties

This section allows you to list raster band or vector (table) properties.

Examples:

If you have more than 10 entries, create a list in a separate file and upload it to a Google Sheet.

The entry type (STRING / INT / DOUBLE) needs to be specified only for vector properties.

Please create a sample thumbnail image to display in the catalog page. Follow these guidelines when creating a thumbnail:

  • Sample images must be 256x256 pixels.
  • Don't include anything else (no background map, borders, city names etc). If there is an area with no data (e.g., oceans around continents), make the background white to better match the web page background. For example, to add a white layer in Code Editor:
var white = ee.Image(1);
Map.addLayer(white, {min: 0, max: 1});
  • Center on something recognizable, e.g, a continent.
  • Try to show a clean visualization. Use a mean or median if individual data is too messy/noisy.

Share a link to the sample visualization in the code editor. This code snippet will be shared in the catalog page for the dataset. Below is an example used for the Copernicus CORINE Land Cover dataset.

var dataset = ee.Image('COPERNICUS/CORINE/V20/100m/2012');
var landCover = dataset.select('landcover');
Map.setCenter(16.436, 39.825, 6);
Map.addLayer(landCover, {}, 'Land Cover');

If you'd like to share a more complex example, please check it into the scripts section of the Earth Engine Community repo.