Satellite composites: techniques in combining geostationary and polar orbiting observations

The combining and compositing of satellite observations over the polar-regions has been accomplished for many years. The methods for doing these mosaics have evolved. Beyond crude overlay composites and merging, conditional minimum compositing methods have lead to eliminating limb darkening effects on infrared composite imagery, and result is reasonable satellite depiction of the polar hemispheres. New advanced image compositing techniques, currently under development, are designed to blend the data from a variety of satellites with differences in calibration, viewing geometry, and temporal offsets. The resulting images are composites of the geostationary and polar orbiting satellites. Spatially, two hurdles to overcome that are related to the image combination procedure include: 1) Test compositing techniques for optimal spatial and temporal resolution of geostationary and polar orbiting satellite data set; 2) Take the time of the observation into account during the compositing process. Temporally, two major factors are the 1) timeliness of the data and 2) time interval between composite images. These parameters affect the mix of geostationary and polar satellite data within a composite. Timeliness of the data is typically not an issue for the geostationary satellite data. Making use of the polar data requires delaying the composite building by a minimum of a couple hours. Based on a typical data cutoff for derived product uses of composites such as atmospheric motion vectors (AMV) for numerical weather prediction (NWP) applications, a three-hour delay was chosen for building the composite. For example, the 1200 UTC combined image is created at 1500 UTC. Determining an optimal time interval between images depends on the desired mix of geostationary and polar satellite data within the composite and a commensurate pixel resolution for an application of the composite, such as cloud tracking. Since most of the global geostationary satellite data is available regularly at ½ hour intervals, this was chosen to visually assess how much polar satellite data would also be included. By choosing one-half hourly images, a pixel resolution of 4 km is considered ideal for a cloud tracking application; this also is the mid-resolution of the input satellite data (ranging from 1 km MODIS and AVHRR HRPT to about 10 km for the geostationary data at high latitudes, with the AVHRR GAC averaging at 4 km).