Health & Environmental Research Online (HERO)


Print Feedback Export to File
1608036 
Journal Article 
Global daily precipitation fields from bias-corrected rain gauge and satellite observations. Part I: Design and Development 
Kottek, M; Rubel, F 
2007 
1 
Meteorologische Zeitschrift
ISSN: 0941-2948
EISSN: 1610-1227 
16 
5 
525-539 
WOS:000250777200007 
Global daily precipitation analyses are mainly based on
satellite estimates, often calibrated with monthly ground analyses or merged with model
predictions. We argue here that an essential improvement of their accuracy is only possible by
incorporation of daily ground measurements. In this work we apply geostatistical methods to
compile a global precipitation product based on daily rain gauge measurements. The raw ground
measurements, disseminated via Global Telecommunication System (GTS), are corrected for their
systematic measurement errors and interpolated onto a global 1 degree grid. For interpolation
ordinary block kriging is applied, with precalculated spatial auto-correlation functions (ACFs).
This technique allows to incorporate additional climate information. First, monthly ACFs are
calculated from the daily data; second, they are regionalised according to the five main climatic
zones of the Koppen-Geiger climate classification. The interpolation error, a by-product of
kriging, is used to flag grid points as missing if the error is above a predefined threshold. But
for many applications missing values constitute a problem. Due to a combination of the ground
analyses with the daily multi-satellite product of the Global Precipitation Climatology Project
(GPCP-1DD) not only these missing values are replaced but also the spatial structure of the
satellite estimates is considered. As merging method bivariate ordinary co-kriging is applied.
The ACFs necessary for the gauge and the satellite fields as well as the corresponding spatial
cross-correlation functions (CCFs) are again precalculated for each of the five main climatic
zones and for each individual month. As a result two new global daily data sets for the period
1996 up to today will be available on the Internet (www.gmes-geoland.info): A precipitation
product over land, analysed from ground measurements; and a global precipitation product merged
from this and the GPCP-1DD multi-satellite product.