The principles behind GridSat-B1 are:
- Easy data access
- Plenty of information to understand the data
- Efficient storage
Easy data access
The GridSat-B1 dataset was developed to simplify access to geostationary data. It removes the need for users to apply calibration to raw satellite data or navigate the native satellite projection. Instead, calibrated data are provided in an equal angle map projection (that is, an equirectangular: plate carree projection).
Data for experts, too
The data are provided in a means to meet the most basic user's needs: the best estimate of the geostationary data for each point. However, much has been done to allow expert users the ability to modify and work with the data:
- Nadir correction - The infrared window channel (irwin_cdr) has been corrected using a view zentih angle correction. However, the variable irwin_vza_adj allows a user to un-adjust the data to calculate the original brightness temperature observation.
- Multiple layers - The geostationary data often overlap, particularly in recent years. The infrared window variables provide the nadir-most observation (irwin_cdr) along with the second-most nadir observation (irwin_2) and the third-most nadir observation (irwin_3). Similarly, the primary and secondary observations are provided for water vapor (irwvp_2) and visible (vschn_2) channels.
- Backing out satellite digital counts - Multiple variables are provided to back out the original satellite digital counts. For example, the digital counts for the visible channel can be estimated from the visible channels (vschn and vschn_2) using the satellite-specific calibration variables: vis_dc_slope and vis_dc_offset.
Data stored efficiently
Data are stored in netCDF-4, which provides the capability to compress variables internally to save space.
Primary variables (irwin_cdr, irwvp, vschn) are always stored as 2-dimensional variables. However, secondary (e.g., irwin_2) and tertiary (e.g., irwin_3) views can be either stored as 2-D or sparse arrays, whichever is more efficient. The use of these sparse arrays is CF-compliant and is described by CF as compression by gathering
Viewing and solar geometry are not provided explicitly. Solar angles (zenith and azimuth) can be computed directly from the time of observation and the gridcell latitude/longitude. Satellite view angles (zenith and azimuth) can be computed from the satellite position and the gridcell position. Satellite-specific positions are provided (satrad, satlon, and satlat) for these calculations. Calculations of these angles are described here.
Note: Courier font denotes netCDF variable names in the GridSat-B1 files.