Landsat Collection 2 download and formats: COG imagery, metadata, access, and GIS workflows

Landsat Collection 2 is the U.S. Geological Survey’s reprocessed Landsat archive, providing Level-1 imagery for Landsat 1–9, global Level-2 surface reflectance and surface temperature products for eligible Landsat 4–9 observations, and tiled U.S. Analysis Ready Data. Products are distributed primarily as Cloud Optimized GeoTIFFs with machine-readable metadata, quality-assessment bands, browse images, and cloud-access options. This page focuses on selecting the appropriate product, downloading only the files needed, understanding formats and coverage, and building reproducible GIS workflows without treating satellite observations as a complete or continuously current inventory of physical features.

Get Started Browse datasets

Landsat Collection 2 is a family of Earth-observation products rather than one single file or layer. It reorganizes and reprocesses the Landsat archive with improved geometric accuracy, updated elevation inputs, calibration improvements, and expanded Level-2 science products. For GIS users, the practical distinction is between Level-1 scene products, Level-2 surface reflectance and surface temperature products, and U.S. Analysis Ready Data tiles. The right download depends on whether the workflow needs calibrated sensor measurements, atmospherically corrected imagery, thermal information, or a tiled analysis-ready structure.

How EDK fits

For: GIS analysts, remote-sensing practitioners, Earth-observation researchers, environmental planners, developers building geospatial pipelines, and data engineers working with large raster archives.

Landsat Collection 2 is a strong external observation layer for understanding how parts of the physical world change over time. It contributes time-stamped raster evidence about land surface reflectance, thermal conditions, vegetation, water, disturbance, and other observable characteristics. In relation to a world model such as E2, Landsat is most useful as a temporal and environmental measurement source that can be connected to physical places and features—such as buildings, roads, waterways, parcels, agricultural areas, or administrative regions—rather than as a replacement for those feature layers. The key connection is computational: Landsat scenes and tiles can be searched by geography and date, filtered by quality, aligned with other spatial objects, summarized over regions, and compared across time. That makes the dataset valuable when the question is not only “where is this place?” but also “what has changed here, how warm is this surface, or what environmental signal is visible at this location?”

•Overlay Landsat observations with physical-world features such as buildings, roads, parks, waterways, or infrastructure to compute environmental indicators for specific places or regions. For example, a workflow could summarize surface temperature or vegetation-related spectral measures within buffers around roads or across building-dense areas, while preserving the Landsat acquisition date and quality masks.
•Use spatial and temporal search to compare the same geographic features across multiple observations. A world-model workflow could identify locations near construction activity, urban expansion, vegetation loss, water change, or heat anomalies, then rank or aggregate those observations by neighborhood, city, watershed, or other geographic region.
•Combine Landsat rasters with other external layers such as surface-water products, elevation models, administrative boundaries, or open building footprints to support change detection and regional computation. Landsat supplies the time-varying observation; the surrounding geographic layers supply the physical context needed for interpretation.

Publisher

U.S. Geological Survey, Landsat Missions and EROS Center

Geography

Global Level-1 archive; global eligible Level-2 scene products; U.S. Analysis Ready Data tiles

Temporal coverage

Level-1 archive since 1972; eligible Level-2 products generally from 1982 to present

Primary raster format

Cloud Optimized GeoTIFF

Access

EarthExplorer, Bulk Download, AWS cloud access, STAC, and selected related services

Update cadence

New observations and products are added operationally; inventory metadata files are updated daily. There is no single fixed cadence for usable imagery at every location.

Dataset profile

license

USGS states that there are no restrictions on the use of Landsat products and maintains a no-cost open data policy. Users should provide appropriate dataset citation and check the terms of any external cloud, software, or derived-data service used in a workflow.

geography

Global for Landsat Collection 2 Level-1 scene products; global for eligible Level-2 surface reflectance and surface temperature scene products; U.S. Analysis Ready Data tiles cover the conterminous United States, Alaska, and Hawaii.

resolution

Resolution depends on sensor, band, and product. Common Landsat multispectral bands are approximately 30 meters, while thermal measurements and derived products have sensor-specific native or resampled characteristics. Consult the applicable USGS product guide before comparing bands or products.

update cadence

New acquisitions and derived products are added operationally, but usable imagery varies by location, acquisition schedule, cloud conditions, sensor, and processing status. USGS bulk metadata files are updated daily; there is no single fixed observation cadence for every place.

temporal coverage

The Level-1 archive contains Landsat observations beginning in 1972. Eligible global Level-2 surface reflectance and surface temperature products generally cover Landsat 4–9 observations from 1982 to the present, with sensor- and product-specific availability constraints.

Formats

Cloud Optimized GeoTIFF for Level-1, Level-2, and U.S. ARD raster products

GeoTIFF browse imagery and JPEG full-resolution browse images

MTL metadata files

XML metadata files

STAC metadata and STAC API records for cloud-hosted assets

Apache Parquet bulk metadata files

Compressed product bundles and individual downloadable band files

Access methods

USGS EarthExplorer interactive search and download

EarthExplorer Bulk Download Application

AWS cloud access for Collection 2 operational products

USGS Landsat STAC API and cloud-hosted STAC metadata

USGS Bulk Metadata Service

NASA AppEEARS for selected Landsat U.S. Analysis Ready Data workflows

GloVis and other USGS visualization or distribution services for selected products

Quality and limits

Clouds, cloud shadows, haze, snow, saturation, terrain occlusion, and fill pixels can make observations unsuitable for a particular analysis.

Level-2 products are not generated for every Level-1 observation; solar-angle, auxiliary-data, sensor, and processing constraints apply.

Digital numbers must be interpreted with the product-specific scale factors, fill values, and metadata definitions.

Surface reflectance and surface temperature products depend on algorithms, atmospheric inputs, emissivity information, and other modeled or auxiliary data.

Landsat’s pixel size and revisit pattern may be inadequate for narrow infrastructure, individual buildings, parcel boundaries, addresses, or rapidly changing events.

Historical scenes may have different sensor characteristics, missing bands, scan-line issues, or processing histories that complicate long time-series comparisons.

A daily-updated metadata inventory does not mean every geographic location receives a new cloud-free or analysis-ready observation every day.

Analysis workflows

Search by area of interest, date range, sensor, product level, tier, and cloud-related criteria before downloading data.

Select Level-1 for custom radiometric processing, Level-2 surface reflectance for many land-surface spectral analyses, surface temperature for thermal studies, or U.S. ARD for tiled U.S. analysis.

Read product metadata, apply scale factors, decode QA bit fields, and mask invalid or unsuitable pixels before calculating indices or statistics.

Use COG-aware GIS or cloud libraries to read only required bands and windows rather than downloading complete archives unnecessarily.

Reproject or resample consistently when comparing scenes, while documenting the target coordinate system, pixel size, and resampling method.

Build temporal composites or change-detection stacks using acquisition dates, valid-pixel counts, and consistent masking rules.

Join raster summaries to geographic features such as buildings, roads, watersheds, administrative units, parks, or infrastructure corridors.

Validate important findings against higher-resolution imagery, field observations, or authoritative local datasets.

Preserve product identifiers, acquisition and download dates, metadata, QA rules, processing versions, and derived outputs for reproducibility.

Value inside EDK

What Landsat Collection 2 contains

Collection 2 is the second major USGS reprocessing of the Landsat Level-1 archive. It contains Level-1 data for all Landsat sensors since 1972, while global scene-based Level-2 surface reflectance and surface temperature products are available for eligible Landsat 4–9 observations from 1982 onward. U.S. Analysis Ready Data provides tiled products for the conterminous United States, Alaska, and Hawaii. (usgs.gov)

Level-1 products preserve calibrated sensor observations as digital numbers and include spectral bands, angle-coefficient files, quality-assessment bands, radiometric-saturation and terrain-occlusion information, metadata, and full-resolution browse images. Users can convert Level-1 values to top-of-atmosphere reflectance or radiance using scaling information in the scene metadata. (usgs.gov)

•Level-1: calibrated scene-based imagery and sensor measurements.
•Level-2: surface reflectance and surface temperature science products.
•U.S. ARD: tiled analysis-ready products for eligible U.S. regions.
•Supporting assets: QA bands, metadata, browse images, angle files, and cloud-oriented catalog metadata.

Geography, temporal coverage, and update cadence

Landsat Collection 2 has global Level-1 coverage for the Landsat archive, with sensor-specific periods extending from Landsat 1 acquisitions beginning in 1972 through current Landsat operations. Level-2 surface reflectance and surface temperature coverage is global for eligible Landsat 4–9 observations, but not every historical observation produces every Level-2 science product. The USGS identifies constraints involving solar angle, required auxiliary data, sensor history, and processing maturity. (usgs.gov)

The temporal record is therefore long but irregular at any individual location. Revisit frequency depends on the satellite constellation, latitude, cloud conditions, acquisition strategy, scene processing, and whether a usable observation exists. Landsat should not be interpreted as a daily global image feed. For current observations, the USGS reports that Landsat 9 Level-1 products can become available within roughly 4–6 hours after acquisition, with corresponding Level-2 products typically available within about three days; historical reprocessing and selected scenes can follow different timelines. (usgs.gov)

•Global Level-1 archive: Landsat 1–9, beginning in 1972.
•Global Level-2 scene products: eligible Landsat 4–9 observations from 1982 onward.
•U.S. ARD: tiled products for the conterminous United States, Alaska, and Hawaii.
•Inventory metadata: updated daily; observation availability varies by place, date, clouds, and processing conditions.

Download options and file formats

The primary interactive access method is USGS EarthExplorer. Users can search by map-drawn area, place, address, date range, cloud-cover criteria, and dataset category, then download individual bands, browse images, or product bundles. The USGS also provides an EarthExplorer Bulk Download Application for larger selections. (usgs.gov)

Collection 2 Level-1 and Level-2 imagery is delivered in Cloud Optimized GeoTIFF format. COG preserves GeoTIFF interoperability while enabling range-based access to internal tiles and overviews, which is useful when a cloud-aware client needs only a spatial window or selected band rather than a complete archive download. Product documentation and metadata files describe band names, scaling factors, fill values, coordinate information, quality flags, and processing details. (usgs.gov)

•Raster imagery: Cloud Optimized GeoTIFF.
•Metadata: MTL, XML, STAC-linked metadata, and bulk metadata files.
•Browse assets: COG TIFF and JPEG.
•Bulk and cloud workflows: EarthExplorer, Bulk Download, AWS, STAC, and selected NASA access tools.

Suitable GIS workflows and interpretation cautions

A defensible workflow begins by choosing the measurement rather than downloading the largest available bundle. Use Level-1 when the analysis requires calibrated sensor data or custom processing. Use Level-2 surface reflectance for many land-surface spectral analyses, and use surface temperature when thermal conditions are the target. Use U.S. ARD when a tiled, consistently projected structure is more important than global scene coverage. Always inspect the product guide and metadata for the sensor-specific band definitions and scaling rules. (usgs.gov)

In desktop GIS, common steps include adding COG bands, applying scale factors, decoding QA bit fields, masking clouds and shadows, harmonizing projections, clipping to a study area, calculating indices, and compositing observations over a defined time period. For time-series work, retain acquisition dates and avoid comparing unmasked pixels as if they represented equivalent surface conditions. For regional statistics, decide whether the summary should use mean, median, percentile, valid-pixel count, or another measure, and report the share of the region that was actually observed.

•Select Level-1, Level-2, or ARD according to the measurement and geography required.
•Apply documented scale factors before interpreting digital numbers as physical quantities.
•Use QA bands to mask clouds, shadows, saturation, snow, terrain occlusion, and fill values as appropriate.
•Keep acquisition dates and valid-pixel counts in every time-series or regional summary.
•Treat Landsat as evidence about observed surface conditions, not as a complete inventory of physical-world objects.

Questions teams ask

What is the best Landsat Collection 2 format for GIS?

For most raster GIS and cloud workflows, Cloud Optimized GeoTIFF is the main working format because it remains compatible with GeoTIFF tools while supporting efficient access to internal tiles and overviews. The best product level still depends on the task: Level-1 for calibrated sensor data, Level-2 for surface reflectance or temperature, and U.S. ARD for tiled analysis-ready processing. Metadata and QA files should be retained with the imagery. (usgs.gov)

Can Landsat Collection 2 be downloaded in bulk?

Yes. EarthExplorer supports interactive searching and downloading, and the USGS provides an EarthExplorer Bulk Download Application. Landsat Collection 2 is also available through AWS cloud access, while STAC and bulk metadata services support programmatic discovery. Bulk workflows are usually more efficient when they query metadata first, select only the needed scenes or bands, and process COGs without downloading unnecessary files. (usgs.gov)

Does Landsat Collection 2 provide current imagery everywhere?

No. Landsat is an operational Earth-observation archive, but usable observations depend on acquisition timing, revisit geometry, cloud and atmospheric conditions, sensor history, product processing, and quality constraints. USGS reports rapid availability for some newly acquired Landsat 9 products, but the archive does not guarantee a cloud-free, analysis-ready image for every location on every day. (usgs.gov)

What is the difference between Landsat Level-1 and Level-2?

Level-1 products contain calibrated and geometrically processed sensor observations, including spectral bands and quality information. Level-2 products are derived science products, including surface reflectance and surface temperature, generated only when the input and auxiliary conditions support the relevant processing. Level-2 products are more directly suited to many environmental analyses, but they still require quality masking, scaling, and interpretation of product documentation. (usgs.gov)

Keep exploring