Data Processing Levels

Data Processing Levels or also known as Data Product Levels help researchers and scientists access data at different stages of refinement, depending on their specific needs and research objectives.

NASA (1986) defined a set of processing "levels" to classify standard data products. The idea was that the level would reflect the type of data processing that had been performed, allowing the user to determine the suitable uses for it. The levels are cumulative, meaning that as processing input the data of the level below is used.

This classification system has evolved over time to what is used today. While the levels follow the same common scheme, they may be defined slightly differently today by different agencies and for different platforms and sensors.

With out the details the levels can be summarised as follows:

Overview of Data Processing Levels

Level	Description
L0	Unprocessed raw instrument data
L1	Georeferenced data in sensor units
L2	Derived geophysical variables projected to coordinate reference systems
L3	Combined georeferenced data
L4	Processed to greatest amount, possibly including modelled or interpolated data and data from different sensors.

Level-0 (L0)

Level-0 data contains raw instrument and payload data at full resolution, with communication artifacts removed. These are the initial measurements captured by the sensors. Usually there is no public access to this level of data.

Level-1 (L1)

Level-1 data contains mainly unprocessed instrument data amended with calibration and georeferencing data. Level 1 data can be traced back to its level 0 data. Data may be extended by information, like meteorological or classification data based on 3^rd party data.

Level-1A (L1A)

Reconstructed, unprocessed instrument data at full resolution, time-referenced, and annotated with ancillary information (e.g., calibration coefficients). Georeferencing parameters are computed but not applied.

Level-1B (L1B)

Derived from L1A data, L1B data are processed to sensor units (not all instruments have L1B source data).

Level-1C (L1C)

Includes new variables supporting the other data. Users can identify which channels are directly copied from L1B and which are synthesized from L1B. Sometimes the data is already georeferenced and orthorectified at this level.

Level-2 (L2)

Level-2 data contains derived geophysical variables at the same resolution and location as L1 source data. The data is mapped into cartographic map reference systems. At L2 the data starts to be useful for most scientific applications.

Level-2A (L2A)

Contains information derived from geolocated sensor data, such as ground elevation. Often Level2A contains atmospherically corrected variables and other metrics describing the intercepted surface.

Level-2B (L2B)

L2B contains data improved in spatial accuracy achieved by image rectification process and geo-registering the data with known locations. Processed to sensor units (not all instruments have an L2B equivalent)

Level-3 (L3)

Variables mapped on uniform space-time grid scales, often summarized periodically (weekly, monthly). Level-3 data is often reduced to one variable and also reduced in spatial resolution. But the spatial extended might be increased to cover certain regions or even the globe.

Level-4 (L4)

Level 4 are high level data. For example, animated images showing the development of a region or statistical values. They can combine data from several satellites, or modelled output.

Exceptions

There are various exceptions and extensions to the before mentioned levels. For example, the data from the upcoming ESA mission FLEX shall be provided also at Level-2C and Level-2D. USGS defines the Level-1 data of the Landsat mission as level L1TP, L1GT and L1GS.

Useful Resources:

• Copernicus - Which levels are used for data processing?
• NASA - Data Processing Levels
• NV5 Geospatial - A Quick Data Product Levels Primer