Sorry, you need to enable JavaScript to visit this website.

Algorithm Overview

L2 Joint Atmosphere Algorithm

The MODIS Atmosphere Level 2 Joint Product (ATML2) algorithm gathers numerous geophysical parameters from the Level 2 Aerosol, Water Vapor, Cloud, Atmospheric Profile, and Cloud Mask products and assembles them into a single, user-friendly file. The idea behind the ATML2 product is to provide users a sampling of the key MODIS Atmosphere Level 2 data products while minimizing file size and preserving all relevant pixel-level information.

To reduce file size, several limitations are implemented in the ATML2 product. First, the parameters that are reported at 1km resolution in their original (source) Level 2 product files are sampled to 5km in the ATML2 product, i.e., every fifth 1km pixel is reported in the ATML2 file. These parameters include the cloud mask, cloud optical thickness, cloud effective radius, cloud quality assurance (QA), cloud phase, cirrus reflectance, and precipitable water (near IR). This sampling scheme is identical to that used for calculating global aggregated statistics in the MOD08 Level 3 product. Second, geolocation arrays that are stored as 4-byte floating-point real numbers in the original (source) product files are compressed to 2-byte scaled integers; this sacrifices 0.001 degrees of geolocation accuracy. Finally, only QA arrays deemed as necessary are included, primarily for those retrieval parameters for which QA is critical for data interpretation. Nevertheless, these limitations do not impact the utility of the ATML2 product for large-scale spatial and/or temporal analyses and other uses.

A notable complexity in this product relates to the geolocation arrays. The Aerosol (04_L2) parameters, which are reported at 10km resolution in their original (source) Level 2 product files, are not sampled in the ATML2 product and are thus reported at their original 10km resolution. This introduces a disconnect between the geolocation for the Aerosol parameters and that for the remainder of the parameters that are sampled to 5km resolution. While the 5km resolution geolocation is simply sampled from the original (source) 1km geolocation using the same sampling scheme discussed above, the 10km resolution geolocation is computed by averaging the four central 1km pixels within each 10x10 1km pixel region (e.g, pixels [5,5], [5,6], [6,5], [6,6]). Thus, two sets of geolocation parameters are included in the ATML2 product, one at 5km resolution (SDS names Latitude and Longitude), and one at 10km resolution (SDS names Latitude_10km and Longitude_10km) for the Aerosol parameters.

A positive feature of the ATML2 product is enhanced program executable (PGE) tracking capability. Each Atmosphere Level 2 product algorithm has an associated PGE number; this number is incremented each time a production algorithm is updated (see PGE Versions Primer for more information). Tracking PGE versions is important for scientific research efforts. Since the ATML2 product collects geophysical parameters from all Atmosphere Level 2 products, each created with its own unique PGE version, tracking PGE versions is a more difficult task. To help solve this problem, a new local attribute "source_info" is added to each SDS in the ATML2 file. This local attribute contains: 1) the file name of the source Atmosphere product, 2) the PGE ID number and PGE version used to create the source file, and 3) the source SDS name. PGE version information is provided for each Atmosphere Level 2 product under the Collections & PGE Versions page (for example, Joint Atmosphere PGE83 Version).