Access
Overview
The are two access methods available in earthaccess. The first method is download, which downloads a file or files to a local computer, a laptop, workstation or the storage space for a high-performace-computer (HPC). The second method is open, which streams data directly into memory. Streaming data is analogous to streaming a TV show or film. This approach can be used whethor you are accessing data from a local machine or accessing data from a cloud-hosted Jupyter Hub or other cloud computing environment but is most useful when you are working in the cloud.
You must login to download or stream data
You need to authenticate using earthaccess.login to download or stream data. See the Access section of the User Guide for more information.
Downloading data
Basic download
earthaccess.download only requires a list of results returned by search_data.
QUEUEING TASKS | : 100%|█████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 1706.04it/s]
PROCESSING TASKS | : 100%|█████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:13<00:00, 6.89s/it]
COLLECTING RESULTS | : 100%|████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 26379.27it/s]
If you want to download a single elements from results, for example results[0], you have make sure you pass a list to download, this is done by putting the single element inside square brackets.
By default files are downloaded to a sub-directory of the current working directory with a name similar to ./data/YYYY-MM-DD_UUID, where YYYY-MM-DD is the year, month and day of the current date, and UUID is a unique identifier. The path to each file is returned and, here, stored in the variable filelist.
[PosixPath('/current/working/directory/data/2025-09-12-c58fe9/ATL10-01_20181014000347_02350101_006_02.h5'),
PosixPath('/current/working/directory/data/2025-09-12-c58fe9/ATL10-01_20181014000347_02350101_007_01.h5')]
You can override this behaviour by setting the local_path keyword to a local directory.
The progress bar is displayed if you are in an interactive session such as a notebook or IPython session. Otherwise the progress bar is not displayed. Whether or not the progress bar is displayed is controlled by setting show_progress=True or show_progress=True.
The list of files in filelist, or one element of that list can be passed to an application such as xarray to load the data for analysis.
or
See download documentation in the User Reference section for more information.
Streaming data
A basic streaming access pattern
Streaming data is an efficient way to access data, without downloading and saving it to disk. This is especially useful if you are working in a cloud-hosted computing environment such as a Jupyter Hub or Google CoLab because storing data incurrs charges. Streaming data when you are working in a cloud computing environment in the same region as the data are stored because connections between the computing and storage environments have high bandwidth. For NASA Earthdata, this is the AWS region us-west-2.
Streaming data with the open method is very similar to download.
QUEUEING TASKS | : 100%|█████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 1706.04it/s]
PROCESSING TASKS | : 100%|█████████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:13<00:00, 6.89s/it]
COLLECTING RESULTS | : 100%|████████████████████████████████████████████████████████████████████████████████████████████| 2/2 [00:00<00:00, 26379.27it/s]
Rather than returning a list of paths, earthaccess.open returns a list of file-like objects. These can be used in the same way as a filepath with third-party applications such as xarray.
or
Optimizing earthaccess.open
When using earthaccess.open to stream data from NASA Earthdata cloud, access times depend on the file format of the data you want to stream and how this data is read by the tools you are using (e.g. xarray, pandas, pyarrow). The default access strategies and parameters for earthaccess.open have been chosen to give good performance for the broad range of datasets, file formats and structures archived by NASA. However, for specific datasets, fine tuning access strategy and parameters may provide improved performance. Fine tuning access is done by setting parameters in fsspec_open_kwargs.
Here, we give a brief overview of the default parameters. For more nformation see this Gist. Some of this gets technical. If it is too technical, just go ahead and use the defaults. Or ask a question in the `earthaccess GitHub Issues.
earthaccess.open uses fsspec to access data stored in NASA Earthdata cloud. fsspec provides a standard interface to local, remote and embedded file systems and byte storage. Essentially, it allows you (and Python tools) to treat remote file storage as virtual file system in the same way you would interact with your local file system.
Two key parameters control the efficiency of file access; cache type and block-size. The default cache type in earthaccess is blockcache. Block-size is set depending on the file size as follows:
| File Size | Block Size | |
|---|---|---|
| Small | < 1000 MB | 4 MB |
| Medium | < 1 GB | 8 MB |
| Large | > 1 GB | 16 MB |