Spatial and Temporal Variability of GPCP Precipitation Estimates

Spatial and Temporal Variability of GPCP Precipitation Estimates By C. F. Ropelewski Summarized from the generous input Provided by G. Huffman, R. Adler, S. Curtis, P. Arkin, X. Yin J. Janowiak, P. Xie R. Ferraro P. Bauer C. Beck, J. Grieser, B. Rudolf M. Bell, B. Blumenthal, B. Lyon F. R. Robertson T. Smith Presented at the 30 th Climate Diagnostics and Prediction WS, 24 -28 Oct 2005 State College, PA Go LIONS

Characteristics of the GPCP Data Set • Global Complete Monthly Precipitation Analysis • January 1979 to Present • 2. 5 o latitude by 2. 5 o longitude • Input data • Satellite Infrared (geostationary 0 • Microwave (from mid-1987) • Gauge data (Global Precipitation Climatology Center (GPCC) operated by the DWD • Output Data • Satellite only • Merged gauge and satellite • Monthly, pentad, daily

“Real-time” monthly gauge data from the GTS

Major GPCP contributions Estimates of the global Precipitation patterns Estimated mean global rainfall rates P estimated a 2. 61 mm/day Yearly Standard Deviation 0. 03 mm/day

Figure 1 a. Zonal mean precipitation, land versus ocean, 1979 -2003, after Adler et al, (2003) Courtesy of Scott Curtis

Figure 1 b. Zonal mean precipitation, land versus ocean, 1979 -86 versus 1988 -2003 After Adler et al, (2003) Courtesy of Scott Curtis

Fig 2. Mean GPCP precipitation 1979 -2002 (mm/day) Courtesy John Janowiak

Mean Annual Cycle

Standard deviations of Annual mean GPCP Precipitation Courtesy Michael Bell

Fig 5 b. Mean annual cycle (mm/day) Global (white), Northern Hemisphere (green), Southern Hemisphere (yellow).

Fig 5 c. Mean annual cycle of precipitation. Oceans (green), Global (white), Land (yellow). Courtesy P. Arkin

Fig. 6 a Annual cycle of zonal mean GPCP Precipitation (mm/day) and b) the difference between total period zonal mean 1979 – 2002 and 1979 -86/87. Courtesy J. Janowiak

Fig 3 (top) Mean GPCP Precipitation P 1 -1979 -86 (left) P 2 -1988 -95 (middle) P 3 - 1996 -2003 (right) Courtesy P. Arkin Fig 4 Mean Differences P 1 minus P 2 left-middle P 2 minus P 3 rt-middle P 1 minus P 3 bottom Courtesy P. Arkin

Fig. 3 a, Mean precipitation for P 1 (1979 -86) – Includes a Major ENSO Warm episode but no microwave estimates. (Courtesy P. Arkin)

Fig. 3 b, Mean precipitation for P 2 (1988 -95) – no major ENSO warm Episodes, includes microwave estimates. (Courtesy P. Arkin)

Fig 4 a P 1 minus P 2 8 year mean GPCP precipitation. (Courtesy P. Arkin)

Fig. 3 c, Mean precipitation for P 3 (1996 -2003) – Major ENSO warm Episode, includes microwave estimates. (Courtesy P. Arkin)

Fig. 4 b P 2 minus P 3 eight year periods. Highlights ENSO-No ENSO

Fig 4 c P 1 minus P 3. Shows the differences in strengths between the 1982/83 and 1997/98 ENSOs and due to the addition of microwave estimates. (Courtesy P. Arkin)

Fig. 7 Seasonal mean precipitation (mm/day) for a) DJF, b) MAM, c) JJA and d) SON. Courtesy P. Arkin

Fig. 8 Standard deviations

Time series of GPCP annual mean global precipitation 1979 -2003 P estimated at 2. 61 mm/day Yearly Standard Deviation 0. 03 mm/day See Allen, M. R. and W. J. Ingram, 2002: Constraints on future changes in climate and the Hydrologic cycle. Nature, 419, 224 -232

Figure 9. (a) Global averages of monthly precipitation (mm day-1) for ocean, total, and land. From Adler et al. 2003.

Fig. 9 b) Tropical (30 o. N-30 o. S) averages of monthly precipitation anomalies (mm day-1) for (top) total, (middle) ocean, and (bottom) land. Vertical dashed lines indicate the months of significant volcanic eruptions. The thin black curves indicate the Niño-3. 4 SST index (o. C). After Adler et al 2003.

Fig. 11. Time series of zonal mean precipitation anomaly. Vertical arrow shows the approximate introduction of microwave data Courtesy P. Arkin

Fig 10 Alternate. Zonal means for tropics only.

a e b f c g d h Fig 10. Time series of zonally averaged GPCP estimates over land (left hand panels) and over the ocean (right hand panels). Zonal averages are for 50 N to 90 N (a, e); 25 N to 50 N (b, f); 25 S to 25 N (c, g) and 50 S to 25 S (d, h). The introduction of microwave data in 1987 is evident for the data over land.

4. LINEAR TREND FIT ZONAL AVERAGE Zonal averages of 25 year linear trend show quite different trends in different lat. bands – again, note regional coherence. 25°S-25°N ~. 08 / 3. 1 = ~ 2. 5% 25°-50°N ~ -. 12 / 2. 4 = ~ -5. 4% Land Ocean are not uniformly related. Fig 10 a Linear trends in the zonally averaged GPCP estimated precipitation. Huffman et al.

Figure 12. Map of linear changes in GPCP precipitation anomalies from January 1979 to December 2003. The thin black contour outlines the local 1% significance level. Courtesy Huffman, Curtis and Adler

Fig 12 Spatial correlations between global trend maps computed for the 1979 to 2004 period and trends computed from successively fewer years. E. g. , the spatial correlation at “lag” 1 is for the 1980 to 2004 period and so on. (from Smith et al. , 2005).

Fourth Rotated EOF of the GPCP Annual data for the period 1979 - 2004. This REOF accounts for 6% of the total variance. (From Smith et al. 2005)

Summary: • The GPCP Data provide (relatively) consistent and complete global precipitation estimates from (1979) 1988 to the present. • These data identify (illuminate) features of the large-scale precipitation fields not (well) known before. E. g, oceanic precipitation patterns, storm tracks, individual ENSO patterns. • Global precipitation shows no significant trends over the period of record…however regional “trends” are evident in the tropics. These aren’t easily untangled from instrumental differences and differences in ENSO in the 17 (25) yr record. • These are research data. For real-time monitoring go to CAMS -OPI or other “operational” estimates (eventually go to CMORPH).