Remote sensing datasets are organized in a hierarchy of processing levels that range from raw instrument output to analyzed data products. A short description of each level is provided below:
Level 1: Calibrated and Geolocated Data¶
In level 1 products are distributed after the data (e.g. observed radiances or brightness temperatures measured by the sensor) has been calibrated. Each observation is geolocated meaning that it has been identified with a corresponding location on the Earth surface, often in terms of longiude and latitude. These data products are provided along the satellite’s ground track in a set of overlapping swaths or segments.
Level 2: Geophysical Variables¶
In the second product level, the calibrated data from Level 1 is converted to a geophysical variable, such as sea surface temperature. These products are also provided along-track and in the same resolution as the previous level.
Level 3: Gridded Observations¶
The third level product uses a collection of Level 2 data to create a gridded composite in a regular grid (e.g. on a consistent longitude-latitude grid). These gridded observation often average observations across time such that each scene may reflect daily, monthly, or even annual data.
Level 4: Modeled and Analyzed Data¶
In the final data level, the observations from previous products are again used to create a gridded product. However, Level 4 data is merged with some sort of model that has been used to augment the data. For example, this could include an optimal interpolation model to fill gaps obscured by clouds or times between retrievals.
For many oceanographic applications, Level 4 SST products are convenient because they provide “complete” maps on regular grids. However, they should be interpreted differently than direct satellite retrievals because much of this data is modeled data rather than direct observations.