ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Intensification of the tropical hydrological cycle? Richard P. Allan Environmental Systems Science Centre, University of Reading, UK Thanks to Brian Soden

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Climate Impacts How the hydrological cycle responds to a radiative imbalance is crucial to society (e.g. water supply, agriculture, severe weather) Motivation

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk IPCC 2007 WGI

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Introduction Hydrological cycle linked to radiative energy balance How will global water cycle respond to warming? Method: compare observations and reanalysis products with model simulations of the present day climate over the period

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Earths energy balance Kiehl and Trenberth, 1997; Also IPCC 2007 tech. summary, p.94

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Earths energy balance Kiehl and Trenberth, 1997; Also IPCC 2007 tech. summary, p.94 Precip: +78 Wm -2 SW heating +67 Wm -2 LW cooling -169 Wm -2

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Changing character of precipitation Convective rainfall draws in moisture from surroundings

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Changing character of precipitation Moisture is observed & predicted to increase with warming ~7%K -1 (e.g. Soden et al. 2005, Science) Thus convective rainfall also expected to increase at this rate (e.g. Trenberth et al BAMS)

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Precipitation also linked to clear-sky longwave radiative cooling of the atmosphere

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Global precipitation (P) changes constrained by atmospheric net radiative cooling (Q) Changes in Q expected to be ~3 Wm -2 K -1 (e.g. Allen and Ingram, 2002) If so, changes in P with warming 3%K -1 …substantially lower than changes in moisture (~7%K -1 )

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Models also display a muted precipitation response (~1-3%K -1 ) Held and Soden (2006) J. Clim P (%) 7 % K -1 T (K)

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Held and Soden (2006) J. Clim P (%) heavy rain: ~7 % K -1 T (K) Mean: ~2 % K -1 light rain: -? % K -1

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk But recent results suggest that the muted precipitation response is not found in the observations Is the global water cycle/radiation budget not captured by models? Wentz et al. (2007) Science

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Are model simulated changes in clear-sky radiative cooling robust? σT4 σT4 ε A σT 4 OLR Ts T, H 2 O, GHG ? Cloud ? Aerosol

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk How does atmospheric net radiative cooling respond to warming? Sensitivity test (Edwards/Slingo): Assume no change in clouds, aerosol, ozone (1)Two warming scenarios (C1, C2) (2)Greenhouse gas increase ( ) (3)Shortwave absorption case

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Sensitivity test (Edwards/Slingo): Assume no change in clouds, aerosol, ozone (1)Two warming scenarios (C1, C2) (2)Greenhouse gas increase ( ) (3)Shortwave absorption case How does atmospheric net radiative cooling respond to warming?

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Increased moisture enhances atmospheric radiative cooling to surface ERA40 NCEP Allan (2006) JGR 111, D22105 dSNLc/dCWV ~ W kg -1 SNLc = clear-sky surface net down longwave radiation CWV = column integrated water vapour

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk What about in observations, reanalyses and models? Method: Compare monthly mean changes in temperature, water vapour and clear-sky longwave radiation, Estimate water vapour driven changes in surface radiation using satellite and climatological ocean data, and an empirical model (Prata 1996)

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Tropical ocean variability SST Water vapour Clear net LW down at surface

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Increase in clear-sky longwave radiative cooling to the surface CMIP3 CMIP3 volcanic NCEP ERA40 SSM/I-derived ~ +1 Wm -2 per decade SNLc (Wm -2 )

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Tropical Oceans dCWV/dTs ~2 4 mm K -1 dSNLc/dTs ~3 5 Wm -2 K -1 AMIP3 CMIP3 non- volcanic CMIP3 volcanic Reanalyses/ Observations

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Tropical ocean variability LWc: TOA LWc: ATM Precip

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk AMIP3 CMIP3 non- volcanic CMIP3 volcanic Reanalyses/ Observations Increase in atmospheric cooling over tropical ocean descent ~4 Wm -2 K -1

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Increased moisture (~7%/K) – increased convective precipitation Increased radiative cooling – smaller mean rise in precipitation (~3%/K) Implies reduced precipitation away from convective regimes (less light rainfall?) Locally, mixed signal from the above RECAP…

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Method: Analyse separately precipitation over the ascending and descending branches of the tropical circulation –Use reanalyses to sub-sample observed data –Employ widely used precipitation datasets –Compare with atmosphere-only and fully coupled climate model simulations

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk GPCP CMAP AMIP3 Model precipitation response smaller than the satellite observations –see also Wentz et al. (2007) Science Tropical Precipitation Response Allan and Soden, 2007, GRL

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Tropical Subsidence regions dP/dt ~ -0.1 mm day -1 decade -1 OCEANLAND AMIP SSM/IGPCPCMAP

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Projected changes in Tropical Precipitation

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Calculated trends Models understimate mean precipitation response by factor of ~2-3 Models severely underestimate precip response in ascending and descending branches of tropical circulation

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Questions (1) Is the analysis flawed? (2) Are the observed changes physically plausible? Are the observations wrong? (3) Can decadal changes in cloud and aerosol explain the discrepancy? (4) Are the models missing fundamental physics?

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk (1) Is the analysis flawed? Changes in the reanalyses cannot explain the bulk of the trends in precipitation

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk (2a) Are the observations wrong? Many of the global precipitation datasets use satellite data –Changes in the observing system –Potential algorithm errors sensitive to temperature and/or water vapour amount Evidence from land and ocean observations suggest models underestimate precipitation and evaporation response to warming –e.g. Wentz et al Science; Zhang et al Nature, Chou et al GRL, …

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Zhang et al Nature

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Surface Evaporation and wind speed Surface evaporation depends primarily upon –Wind stress –Surface humidity gradient (expected to change due to Clausius Clapeyron equation ~7%/K) Globally, evaporation must equal precipitation –Therefore a muted precipitation response requires a muted evaporation response (e.g. muted wind stress)

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Changes in Evaporation over ocean Observed increases in surface evaporation larger than climate model simulations; caused by –increased surface humidity gradient (Clausius Clapeyron) –Little trend in wind stress changes over ocean (Yu and Weller, 2007; Wentz et al., 2007) but some evidence over land (Roderick et al GRL) Yu and Weller (2007) BAMS

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk (2b) Are the changes plausible? Increases in tropical precipitation and evaporation at ~7%/K plausible But increases in ascent region precip ~ 30%K -1, much larger than Clausius Clapeyron Implies intensification of hydrological cycle At odds with observations and model simulations of a reduction in Walker circulation (Vecchi and Soden, 2007)

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Vecchi and Soden (2006) Nature Evidence for weakening of Walker circulation in models and observations

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Vecchi and Soden (2006) Nature Evidence for weakening of Walker circulation in models and observations

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk (3) What else could explain this apparent discrepancy? Changes in land/ocean tropics/extra-tropical transport of heat and moisture? Decadal variability in radiation balance and hydrological cycle? –Changes in clouds –Changes in aerosol

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Wong et al J. Climate, Wielicki et al. 2002, Science Climate models appear to underestimate variability in radiation budget Does this relate to clouds and/or aerosol?

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Global dimming to Global Brightening Stanhill and Cohen EOS (below), Wild et al., Pinker et al Science

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Could decadal changes in aerosol have short-circuited the global water cycle through direct and indirect effect on cloud radiation? See Mishchenko et al. (2007) Science

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Aerosol Hypothesis? Could changes in aerosol directly and indirectly (through cloud) alter the radiation balance and precipitation? Aerosol

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Aerosol Hypothesis? Could changes in aerosol directly and indirectly (through cloud) alter the radiation balance and precipitation? Aerosol

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Aerosol Hypothesis? Could changes in aerosol directly and indirectly (through cloud) alter the radiation balance and precipitation? Aerosol Rad cooling Solar heating

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Aerosol Hypothesis? Could changes in aerosol directly and indirectly (through cloud) alter the radiation balance and precipitation? Aerosol Rad cooling Precip Solar, Evap Circulation?

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Summary Global water and energy cycles coupled Theoretical changes in clear-sky radiative cooling of atmosphere implies muted precipitation response Models simulate muted response, observations show larger response Possible artifacts of data? Possible mechanisms (aerosol, cloud) Implications for climate change prediction

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk References Allen and Ingram (2002) Nature Trenberth et al. (2003) BAMS Wentz et al. (2007) Science My meagre contributions: –Allan (2006) JGR –Allan and Soden (2007) GRL

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Conclusions Heavy rainfall and areas affected by drought expected to increase with warming [IPCC 2007] Heavy precipitation increases with moisture ~7%K -1 Mean Precipitation constrained by radiative cooling –Models simulate increases in moisture (~7%K -1 ) and clear-sky LW radiative cooling (3-5 Wm -2 K -1 ) But large discrepancy between observed and simulated precipitation responses… –Model inadequacies or satellite calibration/algorithm problems? –Changes in evaporation and wind-speed over ocean at odds with models? (Yu and Weller, 2007 BAMS; Wentz et al. 2007, Science; Roderick et al GRL) Observing systems: capturing decadal variability problematic

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Implications for tropical precipitation (GPCP)? ERA40 Q LWc GPCP P OBS Q LWc Pinatubo?

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Comparison of AMIP3 models, reanalyses and observations over the tropical coeans

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Also considering coupled model experiments including greenhouse gas and natural forcings

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Clear-sky vs resolution

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Sensitivity study Based on GERB- SEVIRI OLR and cloud products over ocean: dOLRc/dRes ~0.2 Wm -2 km -0.5 Suggest CERES should be biased low by ~0.5 Wm -2 relative to ERBS

ESSC Seminar, October 18, 2007© University of Reading 2007www.reading.ac.uk Links to precipitation