1. Observed temperature dependence of precipitation extremes: comparison to results of climate models and reanalyses of NCEP and ECMWF Shaw Chen Liu Research Center for Environmental Change Academia Sinica, Taipei WCRP Regional Climate Workshop Lille, France June 14-16, 2010

2. Indirect Effects of Aerosols The aerosol indirect effect is the change in cloud properties caused by a change in the Cloud Condensation Nuclei (CCN) (Twomey, 1977). There are two kinds of indirect effects: Albedo, surface area increase (First indirect effect) Cloud lifetime increase (Second indirect effect) Possible impacts on precipitation: Delay and/or suppression of precipitation by aerosols ( Twomey et al. 1984; Albrecht, 1989; Rosenfeld and Lensky, 1998 ). activation less polluted more polluted

3. ↑ Since 1985 Taiwan has held the infamous #1 title in per unit area emissions of air pollutants in the world.

4. Direct sunshine hours have decreased by ~15% since 1970 (updated from Liu et al. 2002) Average (2043 hrs)

5. T (D) T (N) Linear trends of daytime & nighttime temperature in Taiwan (1961-2005) Blue is decrease ， Red increase.

6. Updated from Liu et al. (2002) Hsu & Chen (2002) also noticed the decrease in light rain.

8. Increases in very heavy precipitation, and sometimes with decreases in light precipitation have been reported in recent years over most land areas (e.g. Karl & Knight, 1998; Goswami et al., 2006) as well as the tropical oceans (Lau and Wu, 2007).

9. Changes of Precipitation Spectra US (48 states)

10. Increases in heavy precipitation can lead to more and worse floods and mudslides. Light and moderate precipitation is a critical source of soil moisture. Because light and moderate precipitation events frequently occur in different seasons and/or regions from those of heavy precipitation, their decrease poses a serious threat to the drought problem.

11. Theoretical Basis for Changes in Precipitation Intensity Trenberth et al. (2003) hypothesized that the precipitation intensity should increase at about the same rate as atmospheric moisture, i.e. about 7%/K according to the Clausius-Clapeyron equation, because precipitation rates from storms were determined by low-level moisture convergence. Furthermore, they argued that the increase of heavy rainfalls could even exceed the moisture increase because additional latent heat released from the increased water vapor could feed back and invigorate the storms. However, quantitatively the hypothesis was not corroborated by results from an ensemble of 17 current climate models (Sun et al., 2007) which estimates a global mean increase of precipitation intensity to be only ~2%/K.

12. A new analysis strategy By focusing on the interannual temperature difference rather than the time-series, we were able to determine a statistically significant relationship between precipitation extremes of the globe and global temperature (Liu et al. 2009). Acknowledgment A major part of this talk is based on: Liu, S.C., Congbin Fu, Chein-Jung Shiu, Jen-Ping Chen, and Futing Wu, GRL, 2009.

13. Changes in Taiwan’s rain intensity (mm/hr) for each degree warming in western Pacific SST (~ global warming) From Liu et al. (GRL, 2009)

14. ΔP/ΔTc for all 10 bins of precipitation intensity. Open bars denote values of ΔP/ΔTc. The vertical line on top of each open bar denotes the 2 standard deviation. GPCP data Average precipitation intensity increases by 23%

15. From Liu et al. (2009) Average precipitation intensity increases by 23% for GPCP data. Only 2% for the ensemble of 17 current climate models.

16. The current climate models underestimate the temperature dependence of precipitation extremes by about one order of magnitude, raising a serious concern that the risk of floods, mud slides and droughts due to global warming, is severely underestimated by IPCC2007.

17. Concerns about the GPCP data The short length of time: 1979-2007. The coarse temporal and spatial resolutions: 2.5 o x 2.5 o, 5-day average. Different satellite data sets, bias, calibrations, and drifts in the accuracy of instrument. -------------------------- So we check GPCP values against results of reanalyses from operational weather forecast models of NCEP and ECMWF. This check is an objective evaluation of the GPCP results as the precipitation from a reanalysis is a diagnostic quantity. In fact, observed precipitation is not used in the data assimilation of the reanalysis.

18. Note:  T is from 2m air temperature of NCEP R1 Averaged intensity change 20.27% 15.22% 12.70% NCEP Reanalysis (daily data)

19. 1961~2006 (2002 missing) NCEP Climate Forecast System (CFS) long-term simulation Data provided by Dr. Hualu Pan Average precipitation intensity increases by about 12,15, and 19% K -1 for 6-hourly, daily and pentad resolutions, respectively.  T is from CFS simulation

20. Note:  T is from 2m air temperature of EC Averaged intensity change 22.97% 7.29% 23.93% EC Reanalysis (daily data)

21. The temperature dependence of global precipitation intensity derived from GPCP is in reasonably good agreement with results of reanalyses by NCEP and ECMWF. Since the reanalysis doesn’t include effects of aerosols, the increase in precipitation intensity of NCEP and ECMWF is not caused by aerosols. It is highly likely that most of the increase in precipitation intensity derived from GPCP is driven by the global temperature increase. The increase in the top 10% heavy precipitation with temperature is significantly underestimated by the operational weather forecast models, especially the EC model. Does this mean that their deep convection scheme needs to be reexamined? If so, how about climate models? How would this affect the water vapor and cloud feedbacks in the climate models? Also vertical heat transport?

22. Thank you for your attention!

24. Aerosol Optical Depth in Winter 2002 (Li et al., 2004)

25. 1961~2006 (2002 missing) NCEP Climate Forecast System (CFS) long-term simulation Data provided by Dr. Hualu Pan Average precipitation intensity increases by about 12,15, and 19% K -1 for 6-hourly, daily and pentad resolutions, respectively.  T is from CFS simulation

26. GPCP data

27. ERA 40 (1958~2001) within 60 o N~60 o S

28. NCEP R1 (1948~2008) within 60 o N~60 o S