Using probability distribution functions for satellite validation

Abstract

Probability distribution functions (PDFs) can be used to assist in the validation of trace gas retrievals made by satellites. A major advantage of this approach is that large statistical samples are used that do not require correlative measurements to be co-located in space and time. Examples are shown from the launch of UARS through to the present. This approach is also useful to evaluate the consistency among Aura instruments as well as their agreement with other datasets. A key feature of this work is putting the observations of Aura in their long-term historical context via statistical comparisons with previous datasets collected over more than a decade. To validate the Aura data, we use data from a variety of platforms including solar occultation (Canadian ACE) and limb sounder satellite instruments, ozonesondes (WOUDC), lidar (NDSC), and aircraft instruments (AVE, PAVE, and MOZAIC). The width of the trace gas PDFs can be used to accurately estimate the atmospheric spatial variability (or representativeness uncertainty) of trace gases as a function of time and location. This statistical analysis is also being used as preparation for full Kalman filter chemical assimilations. The analysis is presented online at http://www.PDFCentral.info. © 2006 IEEE.