Estimates of errors in radio occultation and multiple reanalyses
Presenter:
Jeremiah P. Sjoberg
UCAR COSMIC Program
UCAR COSMIC Program
Talk
In this talk we present results showing vertical profiles of estimated error statistics using the three-cornered hat (3CH) method for radio occultation observations and several model analyses and reanalyses, including ERA-Interim, JRA-55, JRA-55C, MERRA, MERRA-2, 20CR, and ERA-20C. We co-locate all RO observations during selected spans of time with the model-derived variables and compute estimated error variances of temperature, specific humidity, refractivity, and bending angle for the different data sets. The estimated error statistics differ depending on which three data sets are included in the triplets used in the 3CH method, with correlations of errors and differences of representation among some of the data sets likely responsible for many of these error estimate differences (see Anthes et al. in another presentation at this conference). We nonetheless find overall agreement, suggesting that these influences are minimal and thus the error estimates are accurate within the context of these models’ resolutions. We discuss the implications of these results through the lens of data assimilation and of precision estimation of RO.
Presentation in PDF:
