22nd International Conference on Interactive Information Processing Systems for Meteorology, Oceanography, and Hydrology


Temporal comparison of the Comprehensive Pacific Rainfall Database (PACRAIN) with satellite rainfall estimates

Michael D. Klatt, Univ. of Oklahoma, Norman, OK; and M. L. Morrissey and J. S. Greene

Temporal comparison of the Comprehensive Pacific Rainfall Database (PACRAIN) with satellite rainfall estimates

The Comprehensive Pacific Rainfall Database (PACRAIN) is a database of tropical Pacific rainfall observations. PACRAIN merges rainfall records from various weather agencies and other sources into a unified format and makes them available via the World Wide Web. The database currently contains nearly 1.8 million records from almost 800 sites. The PACRAIN project is also part of the Pacific Island Global Climate Observing System (PI-GCOS) initiative to expand and enhance observation networks in the tropical Pacific.

Data accuracy is a top priority for PACRAIN. A comparison of PACRAIN data and satellite rainfall estimates from the Tropical Rainfall Measuring Mission (TRMM) was performed to assess the temporal accuracy of daily records in the database. The TRMM 3B-42 product, which has a 3-hour resolution, was used for the comparison. First, a range of daily time series was created from the 3B-42 data by calculating 24-hour totals with a range of start times (00Z, 03Z, etc.). Next, the correlations were calculated between a PACRAIN series and a satellite series for a range of time offsets, where an offset of 0 hours meant the PACRAIN series was coincident with the satellite series. Finally, the time offset that resulted in the maximum correlation was determined for each site. The time offsets were used to determine if there were any systematic PACRAIN errors and to identify problem sites individually.

The comparison of PACRAIN data to the 3B-42 product produced several findings. First, the maximum correlation between the PACRAIN and 3B-42 rainfall amounts was 0.5 or greater for more than half of the sites. This positive correlation suggested that the results of the temporal comparison would be meaningful. The temporal comparison itself showed that the time offsets for the majority of sites were within 6 hours, with +3 hours being the most common value. While the overall temporal agreement was good, two systematic database errors were discovered. One error was the result of a misinterpretation of certain data from the National Climatic Data Center (NCDC), and all affected records have been corrected. The other error concerned data from French Polynesia, which had inaccurate documentation. The temporal comparison confirmed existing suspicions about the times for these records and they have also been corrected. Work continues to investigate other sites identified as having significant time offsets.

extended abstract  Extended Abstract (92K)

wrf recording  Recorded presentation

Supplementary URL: http://pacrain.evac.ou.edu

Session 4, International Applications Part II
Monday, 30 January 2006, 1:30 PM-5:00 PM, A412

Previous paper  Next paper

Browse or search entire meeting

AMS Home Page