Error Estimates for AMSR-E Rainfall Data

PI: Thomas L. Bell

Institution: NASA / Goddard Space Flight Center
Code 913
NASA/Goddard Space Flight Center
Greenbelt, MD 20771
Phone: (301)-614-6197
FAX: (301)-614-6197
Email: bell@climate.gsfc.nasa.gov
WWW: http://climate.gsfc.nasa.gov/~bell

Co-investigator: Prasun K. Kundu, UMBC, MD

EOS Team: AMSR-E

NASA EOS-PSO funding through FY02: $21,300

ABSTRACT

The research proposed here addresses issues related to validation of gridded climatological averages of AMSR-E-derived rain rates. The objective of the work is to provide a set of rain statistics as a function of the rain climatology and use them to determine error estimates from a simple previously developed formula. These error estimates are essential both for validation of the AMSR-E rainfall products as well as for their utilization in comparison with climate model predictions.
Results obtained from previous studies of error estimates for TRMM and SSM/I indicate that the rms sampling error for monthly averages of gridded rain rates can be simply expressed in terms of the variance of the box-averaged rain rate and the total area of the box sampled by the satellite in the course of a month. These studies include analysis of ground-based data with simulated satellite sampling and also TRMM and SSM/I retrievals over the tropical western Pacific. A larger variance of the satellite-derived averages compared to those obtained by ground radars strongly indicates the presence of retrieval errors in the satellite estimates. This is further confirmed by linear regression of the satellite and radar estimates obtained from collocated rain images which moreover shows the error to be dependent on the rain type (convective/stratiform).

The proposed research is a natural extension of ongoing work on the errors in TRMM gridded rainfall products and will build our current and previous research on TRMM and SSM/I error estimates.