Global Precipitation Monitoring - Satellite-based Applications on Climate Change - page 85

For a little historical perspective on this topic, several efforts to intercompare
and evaluate various types of precipitation algorithms using remotely sensed
information were carried out during the 1990s. The WetNet (Dodge and Goodman
1994 ) Precipitation Intercomparison Projects (PIP) evaluated multiple global and
near-global precipitation algorithms including merged satellite datasets (Barrett
et al. 1994 ; Kniveton et al. 1994 ; Smith et al. 1998 ; Adler et al. 2001 ). The Global
Precipitation Climatology Project (GPCP) similarly sponsored three Algorithm
Intercomparison Projects (AIP; Ebert et al. 1996 ) that compared precipitation
estimated from satellite observations against high-resolution observations from
rain gauges and radars over limited domains (Arkin and Xie 1994 ; Ebert and
Manton 1998 ). For the most part, these studies showed that PMW estimates were
more accurate than IR estimates on an instantaneous basis, but algorithms which
combine PMW and IR estimates were superior. However, these intercomparisons
did not show significant differences between individual algorithms of a common
type. It remains the case that several merged satellite products exist without a clear
consensus on which is superior, and it is common to see a range of similar datasets
used in the literature.
Perhaps the best known and widely used global precipitation climatology comes
from the Global Precipitation Climatology Project (GPCP; Huffman et al. 1997 ;
Adler et al. 2003 ). The GPCP was developed in the 1990s and during the pre-
TRMM era. It combines satellite precipitation from SSM/I with IR estimates (from
both geosynchronous and low orbit) and then anchors the estimates with a robust
surface rain gauge dataset which takes precedence over land. A similar product, the
Climate Prediction Center Merged Analysis of Precipitation (CMAP; Xie and
Arkin 1997 ), also emerged in the same time frame as GPCP and yields similar
results when looking at global-scale precipitation patterns on seasonal to interan-
nual time scales. The current version of the GPCP 2.5 monthly mean dataset is the
version two dataset (Adler et al. 2003 ) which improved on the first version with a
longer record and the addition of TOVS data for improved estimates at mid- and
higher latitudes. Both CMAP and GPCP have problems with high-latitude precipi-
tation due to the lack of reliable data: there are few gauges in these sparsely
populated regions, and available satellite-derived precipitation estimates are of
limited use over ice- or snow-covered surfaces.
Figure 6.3 shows the GPCP V2 global precipitation product for a 30-year period,
1979-2008. The figure shows the seasonal precipitation for December-January
(DJF) and June-August (JJA). The heaviest precipitation over land occurs during
the summer season, as evident by the shifts between hemispheres during winter to
summer. The tropical zones, namely, the ITCZ, exhibits the wettest precipitation on
the Earth, over both the land and ocean zones. Other seasonal features are evident
such as monsoonal regions (e.g., India and North America), midlatitude cyclone
storm tracks, and the shift of the ITCZ.
The GPCP dataset is also extremely useful for monitoring seasonal to interan-
nual changes in precipitation patterns. A good example of this is provided in
Fig. 6.4 , which shows the tropical rainfall anomalies over the central Pacific
Ocean for the 30-year period (1979-2008). This region is typically where the