The data found in these directories are written as text files, in a format created by the group at NASA's Goddard Space Flight Center that processes the raw satellite data. (Note that this is not the same group that maintains the NASA Ozone Watch website.) These files contain total column ozone data--that is, the total amount of ozone in a vertical column of air aeen by the satellite instrument as it looks down towards the earth's surface. The units are "Dobson units", or "DU". These data are "Level 3" (or "L3") data, which means that the data are the physical quantities of interest (total column ozone in this case) that have been averaged into a two-dimensioan (longitude by latitude) grid. (In contrast, "Level 2" data are the physical quantities of interest at the original locations sampled by the satellite instrument along an orbital swath. "Level 1" data consist of the raw instrument radiances, from which the Level 2 data can be calculated through a process known as "retrieval".) To make the Level 3 gridded data, all of the Level 2 orbital swath data for a given day (UTC) are collected and averaged into gridded bins. One consequence of this practice is that the total ozone field shows discontinuities across the 180-degree longitude line, as data from a few seconds after midnight on one side of the international date line are adjacent to data measured nearly 24 hours later on the other side of the date line. Fortunately the ozone field generally evolves slowly enough that such discontinuities are usually fairly small; but they are noticeable. The Level 3 data here are provided by Ozone Watch as a convenience for our users. To obtain the official and authoritative Level 3 and Level 2 data sets from NASA's permanent archives, visit the Goddard Earth Science Data and Information Services Center (GES DISC) at https://www.earthdata.nasa.gov/eosdis/daacs/gesdisc Note also that the instruments work by measuring certain frequencies of the sunlight that is scattered from ozone molecules in the atmosphere. Consequently, measurements are made only in the daylit portions of the satellite orbits. This also means that no measurements can be taken in the polar night of the winter hemisphere. When making the various global images of total ozone for the Ozone Watch website, we fill in gaps with products from the Goddard Modeling and Assimilation Office (GMAO) GEOS FP and MERRA2 assimilation products. Those data assimilation products are more than model output but less than measurements. So while they are appropriate for making media-friendly pictures, we take care *not* to include them in the statistics and data files on Ozone Watch. Now for the file format: Each data file begins with three header lines of text that identify the date, the satellite instrument, and certain information related to production. The header lines are followed by a sequence of multi-line groups. Each group of lines is terminated by a latitude identifier (e.g., " lat = -89.5"). Each group contains the data around a single latitude circle. The number of longitude points around the latitude circle is given in the header; they are equally-spaced within the longitude range described in the header. To read the numbers in the group, concatenate its text lines together and divide the resulting string of characters into three-character subgroups. Each subgroup is an integer representing the number of Dobson Units. For example, a string that stsrts with "250248249249250245" would be parsed into an array of numbers that starts with [ 250, 248, 249, 249, 250, 245 ] Dobson Units. If there are 288 longitude bins specified in the header, there will be 288 numbers to be extracted from the group. A value of 0 indicates that no measurement was taken at the corresponding gridpoint; this usually indicates polar night or a gap between successive orbital swaths. The number of groups in the file will of course be the same as the number of latitude bins specified in the header.