Céline Bonfils*, Kate Marvel, Benjamin Santer, Thomas Phillips Lawrence Livermore National Laboratory External influences on precipitation mean

Ruby Leung state and variability Pacific Northwest National Laboratory

Contact: [email protected]

The relative contribution of Historical changes in zonal-mean Identification of human fingerprint Observed trend is compatible with precursors may change in precipitation in zonal-mean precipitation externally forced

response to global warming Smooth and Zonally average DJF Precipitation Fingerprint = Spaal paern expected in response to external forcing Projection analysis

(left) Wind circulation patterns in each hemisphere help transport moist equatorial air toward the poles. We project observed D(t) and T(t) anomalies onto fingerprint: Introduction These loops are driven by such properties as Earth’s How do we get it? size, rotation rate, atmospheric depth, and heating. = measure the similarity between 1980-2012 data and fingerprint 60°N (below) A smoothed map of satellite data reveals 1. Average over historical simulaons (world with us) à beat down noise Changes in global (ocean and land) precipitation are among the distinct zones of wet and dry land produced by the (uncorrelated noise across simulaons) àThe posive trend in projecon me-series means that the fingerprint is circulation patterns. Livermore researchers used this Cool, dry air information to detect changes in the location and present and growing in the observaons. most important and least well-understood consequences of climate Desert 2. Fingerprint = first EOF of mulmodel average àextract mode of variability descends intensity of global precipitation over the past three change. decades. explaining most of the variance 30°N Equatorial tropical peak In an idealized world… Air cools, Midlatitude storm tracks Subtropical dry zones clouds 90°N § In a simple case (Collins et al. 2010)… form, Inflection points Aempt to remove ENSO first has rain falls no impact (mulvariate fingerprint 60°N Fingerprint expected in response to external is an efficient noise filter) Tropicsropics 30°N Thermal forcing 0° equator: warm, moist air 0° rises Latitude 30°S Desert 60°S Storm Thermod Wet track 30°S 90°S Storm tropics 0 1 2 3 4 5 6 track

Smoothed precipitation, millimeters per day .: wet-get-wetter • Will#ENSO#change#in#the#future?# • Is#this#already#occurring#(detec3on)?## 60°S Observed trend cannot be explained by • Can#we#a8ribute#it#to#specific#external# • Will#ENSO#teleconnec3ons#change#in# internal variability forcings#(a8ribu3on)?# the#future,#invalida3ng#this#scenario?# dry zone dry zone We project D(t) and T(t) anomalies from hundreds 33-year periods of unforced climate Rendering by Kwei-Yu Chu (world without us) to obtain noise me-series ENSO: primary source of drought variability There is no reason for the fingerprint to project, except by chance ENSO: primary source of drought variability in many Dynamic: poleward expansion à Typical Signal-to-noise problem

inregions many of regions the globe via via “teleconnections”teleconnections Thermodynamic indicator T(t) We also project D(t) and T(t) anomalies Walker circulation Natural external forcing only from runs forced with natural forcings CMIP3 data with ozone depletion 0.5 CMIP3 data without ozone depletion alone, or various combinaons of CMIP5 data with ozone depletion Preindustrial control El!Niño! Primary noise mode (ENSO) natural and human forcings. Best fit 0.4 OBS ENSO mode cannot project on our fingerprint ! This multivariate fingerprint acts as an automatic noise filter 0.3 •!Warmer/we)er!tropical! eastern!Pacific!!

density Probability, •!Drier!than!usual!in! 0.2 absence of trends in either Do t or To t , if thermodynamic This method identifies physical effects that are robust across Thermod Indonesia!and!part!of! Wet ð Þ ð Þ Thermod changes and dynamic changes increasingly occur in tandem. multiple models, even in thetropics presence of model errors. We have Australia! Storm Moisture transport P!!! fi fi fi track It is, of course, possible to calculate single-variable ngerprints to identi ed a ngerprintStorm pattern that characterizes the simulta- 0.1 track

• .: wet-get-wetter ! Equatorward!retreat!of!! examine changes in Do t (SI Appendix, S6.1)orTo t (SI Appendix, neous response of precipitation location and.: wet-get-wetter intensity to ex- subtropical!dry!zones! ð Þ fi ð Þ fi This graph shows the probability distribution of the signal S6.2)separately.Thenoise ltering aspect of the multivariate ternal forcing and acts as a noise lter. Observed changes in this (that is, similarity of the fingerprint relative to the noise

fi from naturally occurring climate variability). The observed 0 ngerprint is lost in the single-variable cases: natural variability, multivariate response are incompatible with our best estimates signal matches predictions generated by climate models –6 –4 –2 0 2 4 6 1980-2012 intensity anomaly at peaks/troughs fi dry zone that include both natural and human-induced external Signal-to-noise ratio ENSO in particular, will project onto these ngerprints and of natural variability and consistent with model predictions of forcings. decreaseHow wet theis the weest latude? S/N ratio (SI AppendixHow dry is the driest? ,Fig.S8). However, con- externally forced change. The synchronicity of these changes is Rendering by Kwei-Yu Chu Alternative drought drought precursors precursors expected expected in a sidering each variable separatelyallowsforcomparisonswith key, however: considering either changedry zone in isolation does not lead warmingin a warming world world previous studies that have detected changes in the hydrological to detection and attribution (SI Appendix, Fig. S3). By focusing on Equatorward contraction cycle or atmospheric circulation. both the underlying mechanisms that drive changes in global Thermodynamic- Dynamic- Coupled- fi We nd, considering the dynamical indicator alone, that the precipitation, and by restricting our analysis to the large scales Alexander 2011 fi observations do show a poleward shift in the main features of where we have some con dence in models’ ability to reproduce the Dynamic indicator D(t) This work, in a more global context global precipitation. As previous authors (14) have found, the current climate, we have shown that the changes observed in the

observed trend is much larger than the trends found in forced satellite era are externally forced, and likely to be anthropogenic Future changes in P will be increasingly driven by human- model runs, although including anthropogenic stratospheric ozone in nature. Acknowledgements induced mechanisms superimposed to intensified historical depletion reduces the discrepancy. Santer et al 2007 ACKNOWLEDGMENTS. We are very grateful to Ben Santer and Karl Taylor Seidel et al 2007 Although other studies (7, 26) have found evidence for thermo- ENSO-driven P variability. for helpful comments and infinite patience. This work was supported by the dynamic changes in the hydrological cycle, we do not detect a trend Intensifica7on-of-current-zonal- La7tudinal-redistribu7on-of- Global Climate Modeling Program of the US Department of Energy (DOE)

wet;dry-pa

in SI Appendix,TableS2of this paper) for producing and making available ’ Can we see theses changes emerge in the observations? Conclusions their model output. For CMIP the US DOE s Program for Di- agnosis and Intercomparison provides coordinating support and led devel- Results In this paper, we have presented a simple method to track opment of software infrastructure in partnership with the Global Organization

thermodynamic and dynamic changes in global precipitation. for Earth System Science Portals. Methods 1980-2012 latude anomaly at peaks/troughs This study provides evidence that human activities are affecting precipitation over Where are the weest and driest latudes? Alexander, L. (2011) Extreme heat rooted in dry , Nature Geoscience, 4, 12–13. land and oceans. Anthropogenic increases in greenhouse gases and stratospheric Observaons: Global Precipitaon Climatology Project (GPCP); 33 1. Stott PA, et al. (2010) Detection and attribution of climate change: A regional per- Collins,15. Seager M. et al. (2010) R, Naik The impact N, Vecchi of global GAwarming (2010) on the Thermodynamictropical Pacific ocean and and El Nino. dynamic Nature Geoscience mechanisms 3, 391-397 for Held, I. & Soden, B. (2006) Robust responses of the hydrological cycle to global warming. Journal of Climate 19, 5686-5699 ozone depletion are expected to lead to a latitudinal intensfication and redistribution years of data (1980-2012) spective. Wiley Interdisciplinary Reviews: Climate Change 1(2):192–211. large-scale changes in the hydrological cycle in response to global warming. J Clim 23: Marvel, K & Bonfils C. (2013) Identifying external influences on global precipitation, PNAS, 110(48):19301-19306 of global precipitation. However, detecting these mechanisms in the observational 2. Allen MR, Ingram WJ (2002) Constraints on future changes in climate and the hy- Boreal winter (Dec-Jan-Feb) means Santer 4651BD, et –al.4668. (2007) Identification of human-induced changes in atmospheric moisture content. PNAS 104(39):15248–15253 record is complicated by strong climate noise and model errors. We establish that the drologic cycle. Nature 419(6903):224–232. Seidel16. DJ, Santer Fu Q, Randel BD, et WJ, al. Reichler (1995) TJ (2007) Ocean Widening variability of the tropical and belt its in ina changingfluence climate. on theNat Geosci detectability. 1:21–24 of 70+ simulaons of the “historical” + “RCP8.5” climate that include changes in land and ocean precipitation predicted by theory are indeed present in the 3. Held IM, Soden BJ (2006) Robust responses of the hydrological cycle to global Vecchi,greenhouse G. & Wittenberg, warmingA. (2010) El Niño signals. and ourJGeophysRes future climate: where100(C6):10693 do we stand? Wiley–10725. Interdisciplinary Reviews-Climate climate noise + natural forcings + human forcings (“world with us”) – Change 1, 260-270 observational record, that these changes are unlikely to arise purely due to natural warming. J Clim 19:5686 5699. 17. Adler RF, et al. (2003) The version-2 global precipitation climatology project (GPCP) 4. Santer BD, et al. (2007) Identification of human-induced changes in atmospheric climate variability, and that external influences, probably anthropogenic in origin, are Control simulaons to obtain ~20k years of climate noise (“world monthly precipitation analysis (1979-present). JHydrometeorol4:1147–1167. moisture content. Proc Natl Acad Sci USA 104(39):15248–15253. 18. This Marvel study, K, Ivanovapublished D, Taylor in PNAS K (2013) (Marvel Scale spaceand Bonfils, methods for2013) climate ,will model also analysis.be responsible. without us”) fl 5. Fyfe J, Gillett N, Marshall G (2012) Human in uence on extratropical southern featuredJournal in of the Geophysical June 2014 Research: issue Atmospheres of LLNL’s Science118(11):5082 &Technology–5097. Review. hemisphere summer precipitation. Geophys Res Lett 39:L23711. 19. Santer BD, et al. (2013) Identifying human influences on atmospheric temperature. 6. Min S-K, Zhang X, Zwiers F (2008) Human-induced Arctic moistening. Science Proc Natl Acad Sci USA 110(1):26–33. 320(5875):518–520. 20. Lu J, Chen G, Frierson DM (2008) Response of the zonal mean atmospheric circulation TEMPLATE DESIGN © 2008 This work is supported by the Regional and Global Climate and Earth System Modeling programs for the Office of Science of the 7.United Zhang States X, Department et al. (2007) of Energy. Detection This work of human was performed influence under on the twentieth-century auspices of the U. S. pre- Department of Energy by Lawrence Livermore National Laboratory under contract DE-AC52-07NA27344. LLNL-POST- May 12th-14th Integrated Climate Modeling PI Meeting www.PosterPresentations.com to El Niño versus global warming. J Clim 21:5835–5851. cipitation trends. Nature 448(7152):461–465. 21. Seager R, Naik N (2012) A mechanisms-based approach to detecting recent anthro- 8. Noake K, Polson D, Hegerl G, Zhang X (2012) Changes in seasonal land precipitation pogenic hydroclimate change. J Clim 25:236–261. during the latter twentieth-century. Geophys Res Lett 39:L03706. 22. Levy AA, et al. (2013) Can correcting feature location in simulated mean climate 9. Polson D, Hegerl GC, Zhang X, Osborn TJ (2013) Causes of robust seasonal land pre- – cipitation changes. J Climate 26:6679–6697. improve agreement on projected changes? Geophys Res Lett 40:354 358. fi 10. Durack PJ, Wijffels SE, Matear RJ (2012) Ocean salinities reveal strong global water 23. Scheff J, Frierson D (2012) Twenty- rst-century multimodel subtropical precipitation cycle intensification during 1950 to 2000. Science 336(6080):455–458. declines are mostly midlatitude shifts. J Clim 25:4330–4347. 24. Allen MR, Tett SF (1999) Checking for model consistency in optimal fingerprinting. 11. Seidel DJ, Fu Q, Randel WJ, Reichler TJ (2007) Widening of the tropical belt in a changing climate. Nat Geosci 1:21–24. Clim Dyn 15:419–434. 12. Shindell DT, Schmidt GA (2004) Southern hemisphere climate response to ozone 25. Santer BD, et al. (2011) Separating signal and noise in atmospheric temperature changes and increases. Geophys Res Lett 31:L18209. changes: The importance of timescale. JGeophysRes116:D22105. 13. Kang SM, Polvani LM, Fyfe JC, Sigmond M (2011) Impact of polar ozone depletion on 26. Wentz FJ, Ricciardulli L, Hilburn K, Mears C (2007) How much more rain will global – warming bring? Science 317(5835):233–235. subtropical precipitation. Science 332(6032):951 954. 14. Min S-K, Son S-W (2013) Multi-model attribution of the southern hemisphere Hadley cell 27. Allan RP, Soden BJ, John VO, Ingram W, Good P (2010) Current changes in tropical

widening: Major role of ozone depletion. JGeophysResD:Atmospheres118(7):3007–3015. precipitation. Environ Res Lett 5(2):025205.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1314382110 Marvel and Bonfils